Philosophia Scientiæ, 17-3 | 2013, “Tacit and Explicit Knowledge: Harry Collins’S Framework” [Online], Online Since 01 October 2013, Connection on 15 January 2021

Philosophia Scientiæ Travaux d'histoire et de philosophie des sciences

17-3 | 2013 Tacit and Explicit Knowledge: Harry Collins’s Framework

Léna Soler, Sjoerd D. Zwart and Régis Catinaud (dir.)

Electronic version URL: http://journals.openedition.org/philosophiascientiae/876 DOI: 10.4000/philosophiascientiae.876 ISSN: 1775-4283

Publisher Éditions Kimé

Printed version Date of publication: 1 October 2013 ISBN: 978-2-84174-641-5 ISSN: 1281-2463

Electronic reference Léna Soler, Sjoerd D. Zwart and Régis Catinaud (dir.), Philosophia Scientiæ, 17-3 | 2013, “Tacit and Explicit Knowledge: Harry Collins’s Framework” [Online], Online since 01 October 2013, connection on 15 January 2021. URL: http://journals.openedition.org/philosophiascientiae/876; DOI: https://doi.org/ 10.4000/philosophiascientiae.876

This text was automatically generated on 15 January 2021.

TABLE OF CONTENTS

Editorial Introduction: Collins and Tacit Knowledge Léna Soler and Sjoerd D. Zwart

Building an Antenna for Tacit Knowledge Harry Collins

Tacit Knowledge and Realism and Constructivism in the Writings of Harry Collins Trevor Pinch

At the Margins of Tacit Knowledge Michael Lynch

Taking the Collective Out of Tacit Knowledge Stephen Turner

Tacit Knowledge and Its Antonyms Tim Thornton

Collins’s Taxonomy of Tacit Knowledge: Critical Analyses and Possible Extensions Léna Soler and Sjoerd D. Zwart

Ships that Pass in the Night: Tacit Knowledge in Psychology and Sociology Harry Collins and Arthur Reber

Refining the Tacit Harry Collins

Varia

Unix selon l’ordre des raisons : la philosophie de la pratique informatique Baptiste Mélès

Introduction au rapport inédit de Helmholtz sur Mosso Alexandre Métraux

Ein unveröffentlichter Bericht über verschiedene Arbeiten Angelo Mossos Hermann von Helmholtz

Un rapport inédit sur divers travaux d’Angelo Mosso Hermann von Helmholtz

Philosophia Scientiæ, 17-3 | 2013 2

Editorial Introduction: Collins and Tacit Knowledge

Léna Soler and Sjoerd D. Zwart

Introduction

1 Harry Collins is internationally recognized as a distinguished sociologist of science who writes creatively on a substantial number of varied subjects. He is acknowledged as one of the prominent specialists on the topic of tacit knowledge and has played an important role in the introduction of this topic into science studies. He has investigated the topic extensively, most famously through several case studies of physics [Collins 1974, 1984, 1985, 1990, 2001a,b, 2004], [Collins & Kusch 1995, 1998], [Collins & Pinch 1993], [Pinch, Collins & Carbone 1996]. The publication of his latest book, Tacit and Explicit Knowledge ([Collins 2010], hereafter TEK), provides the occasion and primary motivation for the present thematic issue of Philosophia Scientiæ. The aim of this issue is to discuss Collins’s new ideas on tacit knowledge developed in TEK, and more generally to address issues related to the tacit dimension, especially (but not exclusively) in science.

2 The point of departure for this thematic issue was a conference in December 2011, entitled “Tacit Knowledge in Science: Discussions with Harry Collins”, which was organized by the PratiScienS team, in Nancy, France.1 Harry attended the conference in person, and creatively took part in the lively exchanges about his position. Initially, Harry was ‘only’ asked to introduce the conference and to respond to the different talks, but as a result of these exchanges, he was kind enough to give a spontaneous and very stimulating extra-talk. We were, and remain, very grateful to Harry for the role he has played in making the conference a success and in participating in bringing about the present thematic issue. Although the thematic issue was inspired by the original conference, its contributions are the result of a separate call for papers.

3 The thematic issue begins and ends with two new contributions from Collins. In the first, Collins introduces the core elements of his characterization of tacit knowledge in TEK and reconsiders these elements in retrospect. In the second, he concludes by

Philosophia Scientiæ, 17-3 | 2013 3

reviewing the articles in the issue and addresses the main questions raised by each of them.

4 We would also like to draw particular attention to a contribution of a special type, co- authored by Collins and Arthur Reber. Reber is Broeklundian Professor emeritus at Brooklyn College, and has conducted empirical psychological research into tacit knowledge acquisition for more than four decades. In 1993, he published a book on tacit knowledge from the experimental point of view, Implicit Learning and Tacit Knowledge: An essay on the cognitive unconscious [Reber 1993]. This book was the reason that the editors invited Reber to review one of the manuscripts. A by-product of the reviewing process was that Collins and Reber started a correspondence about some aspects of tacit knowledge, and came to realize they were entangled in a typical example of a longstanding clash between different frames—they came to realize that Collins’s sociological and Reber’s psychological perspectives on tacit knowledge could be thought of as “incommensurable” in a Kuhnian sense. They then conceived the idea to reconstruct their exchange, with its misunderstandings and agreements about tacit knowledge, and put it on paper. The resulting article goes beyond the scope of tacit knowledge: it has become a remarkable and entertaining ‘inside description’ of a clash of paradigms, easily recognizable to anyone who has had similar experiences with a frame-shifting discussion. It also offers some evidence that although the task is hard and success requires willful effort and perseverance, incommensurability does not preclude the possibility of mutual understanding. As editors, we wish to congratulate Collins & Reber on their success in surmounting the difficulties of the process and in providing a living exemplar of how cross-paradigm conversations go; we are delighted and grateful that Collins & Reber contributed this unexpected bonus to our thematic issue.

5 In the remainder of this editorial introduction, we first situate the subject of tacit knowledge in the landscape of the science studies; next we consider Collins’s contribution to this subject; and finally, we sketch a brief overview of the contents of the present thematic issue.

1 Tacit knowledge in the landscape of the science studies

6 Skills and tacit aspects involved in scientific practices are a relatively recent theme in philosophical, sociological and historical studies of science, and Collins has played an important role in establishing the interest in this theme.

7 The origin of reflections about the nature of tacit knowledge and its modes of intervention in human activities aiming at knowledge acquisition is commonly attributed to Michael Polanyi (see especially [Polanyi 1958, Polanyi 1967]). Often, Wittgenstein (in relation to rule-following) and Thomas Kuhn also have been referred to, and rightly so, as providing important insights into tacit knowledge and related epistemic issues ([Wittgenstein 1953, 89, 201]; [Kuhn 1970]; for more details see [Soler 2009, chap. IX]).

8 In the 1970s, oftentimes inspired by these early writings, scholars interested in science became increasingly convinced that tacit knowledge played an important role in science. How did this happen? At the time, researchers, especially in sociology of

Philosophia Scientiæ, 17-3 | 2013 4

science, questioned the widespread belief that science could be adequately described only by its explicit products such as published theories and explicit justifications, and stressed the need to study “science in the making” and scientific practices rather than scientific products. They carried out detailed studies of ongoing experimental practices and claimed that such an approach led to a more profound and realistic conception of science.2 As a spin-off of these practice-oriented tendencies, some researchers, among whom perhaps most importantly Harry Collins (see in particular his pioneering contributions [Collins 1974, 1975]), showed that the actual process of science included many important tacit, unarticulated aspects that resisted any attempt of explication. By doing so they boosted the interest in tacit knowledge in science.

9 Subsequently, more and more sociologists of science, but also philosophers and historians of science, sought to explore and characterize tacit “know-how” and tacit “know-that” (a distinction first introduced by Gilbert Ryle in [Ryle 1945-1946], which later became classical), the acquisition of tacit resources, and their transmission and dissemination within larger social groups. Additionally, but more rarely, they analyzed the ways in which the attribution of tacit resources affects the constitution of scientific results and conclusions.3

10 Traditionally, tacit resources are almost always referred to in the literature by the term “tacit knowledge”. In order to remain consistent with this common usage, the present thematic issue is entitled “Tacit Knowledge in Science”. However, we believe that the term “tacit knowledge” should be questioned, because not every tacit resource relevant to science or knowledge production is straightforwardly well-characterized as knowledge.

11 Admittedly, conceptions of knowledge vary—as is illustrated by some of the contributions in this thematic issue. In TEK, Collins favors the “knowledge-as-stuff” metaphor, according to which knowledge can be detached from the human beings who might possess it, and can be considered “in itself”. Within this framework, we may consider, for example, the knowledge corresponding to “bike-riding” independently of the specific way in which human beings know how to ride bikes. However, other authors favor different perspectives on knowledge (see in particular Thornton’s and Reber’s articles in this issue). According to these authors, tacit knowledge can only be analyzed in reference to the nature of the process through which a task is humanly accomplished, such as, for instance, the nature of the process through which human beings successfully ride bikes. Perhaps in order to emphasize the latter perspective, Polanyi preferred the expression “tacit knowing” to that of tacit knowledge.

12 Even if we put aside differences between conceptions of knowledge, however, some tacit elements commonly referred to as “tacit knowledge” still do not straightforwardly qualify as knowledge—at least not without further discussion. Take for example the tacit presuppositions an analyst attributes to a given scientist from the past in order to explain the behavior and position of this scientist, when these presuppositions are false according to the present-day science of the analyst: are these presuppositions tacit knowledge? Or take some ordinarily-tacit values and norms that were once constitutive of past scientific practices, such as the requirement “reject any force-at-a-distance and throw it out of science”: are these norms tacit knowledge? Even if the answer is affirmative for some conceptions of knowledge, the knowledge’s status surely cannot be taken for granted from the outset. Consequently, the currently standard expression “tacit knowledge” is potentially misleading and confusing. It would be preferable to

Philosophia Scientiæ, 17-3 | 2013 5

start from an agnostic position, and adopt a more neutral vocabulary that does not already commit us to a position on the epistemic status of the elements that we treat as tacit.

13 From such a perspective, we could replace “tacit knowledge” with “tacit resources”, “tacit aspects” or “tacit dimension”. As far as Collins’s three central categories in TEK are concerned—Relational, Somatic and Collective Tacit Knowledge—we could substitute the “K”, which stands for Knowledge in RTK, STK and CTK, with an “X”, and remain agnostic about the relation between X and knowledge.

2 Collins’s contributions to the topic of tacit knowledge

14 Collins creatively and critically exploited the contributions of Polanyi, Wittgenstein, Kuhn and others, and put them to work in his research on science. He is one of those scholars who have helped us to improve our understanding of both the nature of tacit knowledge, and the epistemological consequences of the involvement of tacit resources in scientific and technological practices.

15 From his first publication devoted to the tacit dimension (his now classical paper on the TEA-laser [Collins 1974]) onwards, via his famous book Changing Order [Collins 1985], and many other publications, culminating in TEK [Collins 2010], Collins has developed diverse and helpful analytical tools, as well as insightful, although often controversial, theses related to tacit knowledge. Let us briefly review a selection of these tools and theses.

2.1 Analytical tools for a finer grasp of tacit knowledge

2.1.1 Types of tacit knowledge

16 Collins introduced multiple categories that help to distinguish various types of tacit knowledge and to understand the relations among them.

17 In [Collins 2001a], nearly a decade before the publication of TEK, Collins proposed his first useful classification of tacit knowledge, which was composed of five categories: (i) “concealed knowledge” (such as knowledge that is kept secret); (ii) “ostensive knowledge” (such as knowledge that is transmitted by pointing); (iii) “mismatched saliences” (such as knowledge that is not expressed because the knower does not realize that it needs to be revealed to a recipient); (iv) “unrecognized knowledge” (knowledge that is actually used but not expressed and even not known to be possessed by the user); (v) “uncognized/uncognizable knowledge” (knowledge that is not articulated and not explicable in a given stage of human history, and that is perhaps humanly impossible to explicate).

18 In TEK, Collins introduces three new categories of higher-order, RTK, STK, and CTK, and explains their relations to the five older 2001 categories. In particular, the first four types of tacit knowledge he introduced in 2001 become different sub-types of Relational Tacit Knowledge in TEK.

Philosophia Scientiæ, 17-3 | 2013 6

19 The new three-fold mapping of tacit knowledge constitutes the core of the analytical tools provided by Collins in TEK—even if, as we will see below, at the end of section 3.2., Collins does not conceive the mapping only as a set of useful tools, but also as an ontological division of the realm of tacit knowledge into three altogether exhaustive kinds. In TEK, each of the three kinds of tacit knowledge is classified according to the reason why the knowledge K involved resists explication, and the three kinds are ordered according to the force of the resistance to explication offered by the corresponding kind of K. Relational or weak tacit knowledge (RTK) is knowledge that: (i) is explicable today (i.e., could be explicated without fundamental problem, offers no essential resistance to explication: hence the “weak”); and (ii) remains however tacit in some particular situations, for contingent reasons that have to do with the relations between people or groups of people (hence the “Relational”). Collective or strong tacit knowledge (CTK) corresponds to knowledge that possesses the three following features: (i) it is required to act relevantly and creatively in a certain collective (hence the “Collective”); (ii) it is presently inexplicable because of its collective character; (iii) we cannot see today how it could be explicated in any foreseeable future (hence the “strong”). Somatic or medium tacit knowledge (STK) is positioned in between. It corresponds to somatic skills defined by the three following features: (i) they are difficult to explicate because they involve the body; (ii) as a consequence of this particular kind of difficulty, they remain, as a matter of fact in the present stage of scientific development, inexplicable; (iii) they could nevertheless, assuming further developments in scientific knowledge and technology, be explicated (hence the “medium”), but only in certain ways—ways corresponding to what Collins calls senses (3) and (4) of explication.

20 Mentioning Collins’s senses (3) and (4) of explication leads us to another distinction developed in TEK, which plays an important role in several papers of this thematic issue: the distinction between four different possible meanings of “explicable” [TEK, 81]. This distinction proves to be a source of difficulties for some of Collins’s readers. Or so is the conclusion of the editors, based on their interpretation of several articles submitted to the call for papers for this thematic issue.

2.1.2 Four different meanings of “explicable”

21 According to Collins, tacit knowledge becomes explicit by the mediation of “strings”. Strings are physical patterns without intrinsic meaning, like the material side of a written word or of an uttered sentence. More broadly put, strings may be the physical substrate of any meaningful human interpretation—that is, almost any material thing.

22 There are four ways in which the manipulation of strings can lead to the explication of some knowledge K, which is not explicated. This gives four senses of “explicable”: (1) “explicable by elaboration”: “a longer string affords meaning when a short one does not”; (2) “explicable by transformation”: “a physical transformation of strings enhances their causal effect and affordance”; (3) “explicable by mechanization”: “a string is transformed into mechanical causes and effects that mimic human action”; and (4) “explicable by explanation”: “mechanical causes and effects are transformed into strings called scientific explanations” [TEK, 81]. Senses (1) and (2) are related to the most widespread uses of “explicable” and “explicit”. Roughly speaking, to explicate in these senses is to capture, in ordinary language, what is tacit for an interlocutor, by saying more or saying it differently. By contrast, senses (3) and (4) depart from

Philosophia Scientiæ, 17-3 | 2013 7

common usages. To explicate in sense (4) is to develop a scientific explanation of some performance or knowledge K. It might be, for example, a scientific explanation of “riding a bike” in terms of the curve that should be taken to compensate for a given angle of imbalance [TEK, 101]. To explicate in sense (3) is to succeed in building a machine that is able to realize some K, such as, for example, a robot able to ride a bike as well as humans do—that is, able to go from a point A to a point B without falling— even though for the same performance, the underlying robotic processes differ from those of humans. We think it is important to examine the counterintuitive implications related to Collins’s senses (3) and (4) of explication. Some of these implications are considered in the contributions to the present thematic issue (see notably those by Thornton and Soler-Zwart).

23 Other interesting conceptual distinctions are introduced in TEK in relation to RTK, STK and CTK. To close our selection, we would like to mention the analysis proposed by Collins of eight different “types of cannot” [TEK, 90]. Differentiations of these types were urgently needed, and are considerably clarifying, when compared to crude classical definitions of tacit knowledge as knowledge that cannot be explicated.

24 Let us turn to the epistemological consequences of the inevitable involvement of tacit resources in scientific and technological practices.

2.2 Epistemological implications related to tacit knowledge

2.2.1 Delays, failures and losses in knowledge transmission

25 A first, no longer very controversial but still important implication of Collins’s earlier case studies, is that tacit knowledge can be responsible for impediments, delays or even failures, in the transmission of the experimental or technical protocols. When a

scientist S2 attempts to follow protocols set out by another scientist S1 for re-producing

experimental facts or technical objects that S1 claims to have successfully produced, S2

might fail because S2 does not succeed in acquiring the tacit knowledge possessed by S1. A successful replication is not guaranteed by the conservation of written protocols and public reports, because the tacit knowledge (especially the tacit know-how) is never completely captured by such explicit means. This raises the risk of irreversible losses of knowledge. Limiting such risk requires timely and costly maintenance policies and activities—such as uninterrupted transmission of the specific human know-how from experts to apprentices, or the development, where possible, of mechanical or ‘turnkey’ methods for bypassing human tacit knowledge.

2.2.2 Tacit knowledge and the experimenter’s regress

26 A second, crucial and highly controversial thesis is Collins’s famous thesis of the experimenter’s regress. In Collins’s words: To know whether you have built a good gravitational wave detector, you should try it and see if it works properly. But to know what “works properly” means you have to know what it should see. But to know what it should see, you have to know what gravitational waves look like. But to know what gravitational waves look like, you have to build a good gravitational wave detector and look at them. But to know whether you have built a good gravitational wave detector, you should try it and see if it works properly. And so on! [Collins 2004, 126]

Philosophia Scientiæ, 17-3 | 2013 8

27 The experimenter’s regress locates tacit knowledge at the level of the proper working of the instrumental device (in the above quotation, the gravitational-wave detector). When an experimental controversy arises, any experimental conclusion advocated by a given team can be contested by claiming that the competing experimenters lacked the crucial tacit knowledge needed to carry out the experiment successfully.

2.2.3 Under-determinations in tacit knowledge attribution and variable ideas about success

28 The problems with replication and the experimenter’s regress might be considered to be a pragmatic and elaborated version of the Duhem-Quine thesis about the underdetermination of theories by empirical evidence. Accordingly, tacit aspects introduce an additional source of indeterminacy in science [Soler 2011b]. In a given historical scientific controversy, in principle, at least two interpretations are possible, and alternative interpretations actually are sometimes endorsed by rival teams: either the experimenters do indeed possess some tacit knowledge and their experimental conclusions are well-grounded; or they do not possess the required tacit knowledge and their experimental conclusions must be dismissed as methodologically faulty. But how do we know if some subjects indeed possess tacit knowledge, and who decides in case of disagreements between the actors?

29 The experimenters’ regress thesis draws our attention to the empirical conditions through which we usually attribute tacit knowledge to actors, whether or not they are scientists. Namely: we notice that some people or groups of people, ourselves possibly included, are able to perform valued tasks repeatedly whereas others are not able to do so; we notice that they are able to perform these tasks without telling how they do so, and that many aspects of their successful actions are hard or even impossible to express (when we ask the actors to explicate them or when we try to articulate them ourselves); we conclude that these people have resources at their disposal that others lack and, if we are ready to identify the result of their actions as a true success, we are ready to consider the possessed resources as tacit knowledge.

30 Such an attribution of tacit knowledge is not too problematic when we are dealing with already institutionalized and socially stabilized kinds of action, such as bike riding, normal science and the like. In these cases, what “success” means is relatively clear; individuals or groups by and large agree about what counts as a success and what counts as a failure. But the more we deal with pioneering, creative, open, non-stabilized and non-consensual practices—such as when scientists attempt to investigate new phenomena by means of poorly-mastered experimental devices, or during periods of scientific revolution, or when avant-gardist artists challenge established conventions, and so forth—the more the meaning of success becomes an open-ended and controversial matter. In such pioneering situations, the people involved entertain different (often tacit) intuitions about success and how to recognize it; the parties compete to impose their own (often tacit) values and to have them recognized as the appropriate ones. Two competing groups of people operating with heterogeneous intuitions about success are unlikely to attribute the status of knowledge to the same tacit resources. A tacit resource valued as knowledge by some, is likely to be dismissed by others as a tacit prejudice.

Philosophia Scientiæ, 17-3 | 2013 9

31 In short, tacit knowledge identification and attribution is strongly related to ideas about success, and ideas about success might vary between societies, among individuals, and from one context to another. Among other sources of underdetermination, ‘subjective’ and ‘social’ sources can be suspected to play a role in attributions of tacit knowledge. Examples of such attributions are the tendency of parents to overvalue the capacities of their children, and hence to attribute tacit knowledge to them all too easily; or the hegemonic tendencies of a given community to claim that they are the sole possessors of the inexpressible knowledge required to achieve certain socially valued aims. Trevor Pinch and Michael Lynch give examples in their contributions.

2.2.4 The ontological status of tacit knowledge

32 Reflections such as those we have just discussed lead us to ask questions about the ontological status of tacit knowledge. Is tacit knowledge real? Or, is tacit knowledge relative and inextricably attached to some (collective or individual) subject?

33 Collins was one of the first scholars to draw our attention to the problematic character of the epistemic status of tacit knowledge, when he identified the experimenters’ regress and the under-determinations that go with it. In TEK, however, Collins does not take experimenter’s regress into account and leaves aside its implications. Let us recap Collins’s ontological position in TEK: in that book Collins offers a systematic conceptual analysis of tacit and explicit knowledge, from which follows a new three-fold classification of tacit knowledge, which Collins endows with an empirical and ontological status. RTK, STK and CTK are not just three useful tools to order a ‘hypothetical something’ called “tacit knowledge”. No, for Collins tacit knowledge is real knowledge, and RTK, STK and CTK name three different existing kinds of tacit knowledge. In brief, RTK, STK and CTK cut the realm of tacit knowledge at its joints. Thus in TEK, Collins treats tacit knowledge as if it is unproblematically real; he reasons as though the success, failure, or relevancy of human actions— which work as criteria for tacit knowledge attributions—are uncontroversial and taken for granted; as though knowledge is some sort of real “stuff”, isolable as such and independent of any specified subject.

34 The different attitude that Collins takes toward tacit knowledge in TEK, as compared with his earlier work, has puzzled some readers (see notably Pinch’s and Lynch’s contributions to this thematic issue). In response to his critics, Collins attempts to clarify the relation between the stance adopted in TEK and that adopted in his earlier works, and explains the position endorsed in TEK as a shift of attention.

3 Fundamental background features of Collins’s worldview

35 Collins’s recent realist approach and three-fold ontological mapping of tacit knowledge operates within a particular worldview. To conclude our presentation of Collins’s perspective on tacit knowledge, we shall introduce some core elements of this worldview.

36 According to Collins, reality is made of two radically different kinds of beings and processes. On one side, there are interpretative beings and processes, that is, beings

Philosophia Scientiæ, 17-3 | 2013 10

that possess language, produce meaning, have culture, use symbolic entities and make translations between them. On the other side, we have beings and processes that do not interpret—in Collins’s terminology they are determined by strings and transformations of strings: cause-effect processes, mechanical transformations, nature as opposed to culture, inert things as opposed to meaningful entities. This dual perspective is at the heart of Collins’s crucial and pivotal “transformation-translation” thesis [TEK, 25ff.].

37 Only human beings are interpretative beings—or more exactly human-like beings, since, as Collins insists, where to draw the dividing line exactly is unimportant: the essential point is the existence of the ontological divide. Human-like beings are unique in their interpretative activities. Moreover, human beings are social beings in an essential sense [TEK, 116]. Consequently, all aspects that are on the side of interpretative realities, such as language, meaning, culture, translation and so on, are social realities. They would not exist in the absence of a collectivity—a collectivity whose members present certain kinds of somatic and non-somatic affordances that enable interpretative activities. As a true collectivist, Collins claims that all aspects of interpretative realities are properties of collectives, and that they cannot be reduced to individuals or to interactions between individuals. Individuals acquire language, meaning, and culture, as well as the ability to translate and interpret creatively and relevantly, by means of their immersion in society through a mysterious “socialization process”.

38 Human-social beings and human societies are characterized by variability, flexibility, openness and unpredictable processes, or in Collins’s concepts, by “polimorphic actions” [Collins & Kusch 1995, 1998]: different members of a given society, as well as one and the same individual at different moments, act flexibly, creatively and relevantly according to the social contexts; societies themselves, and the tacit social rules that characterize them, differ from place to place and change rapidly through time.

39 In contrast, non-human, or non-human-like beings and groups of such beings—animal species, inert “natural kinds”, etc.—are characterized by relatively uniform, fixed, rigid, predetermined (i.e., mechanical) behaviors. Such behaviors do not vary from one member of the species to the others—neither in a given period of history, nor through time: dogs or stones do not behave today in an essentially different way than they behaved yesterday and will behave tomorrow. In Collins’s terminology, such behaviors correspond to “mimeomorphic actions”. Non-human-like beings do not adapt their way of behaving flexibly throughout the process of their immersion into the group they belong to.

40 This ontological duality between interpretative-translational and non-interpretative- mechanical realities is highly controversial, and far from being an easy matter. It is reminiscent of Descartes’ dualism, which still preoccupies contemporary debates on the relations between the natural and the social sciences. In spite of its intricacies, Collins manages to propose his dualist position in a clear and bold way.

4 Contents of the special issue

41 Beyond a common interest in tacit knowledge, the authors of articles in this thematic issue have different backgrounds, stemming from various research traditions, and displaying a variety of professional interests. Three of them, Trevor Pinch, Michael

Philosophia Scientiæ, 17-3 | 2013 11

Lynch and Stephen Turner are well-established sociologists of science and technology. The other contributing scholars, Tim Thornton, Léna Soler and Sjoerd Zwart, have backgrounds in philosophy. Tim Thornton is a Wittgenstein specialist; Léna Soler has been trained in the tradition of historical philosophy of science (more specifically philosophy of physics); and Sjoerd Zwart’s main area of interest is the analytical philosophy of technology. As has already been mentioned in section 1, the author of the final critical article, Arthur Reber, won his spurs in empirical psychology. The differences in background of the authors guarantee contrasting ways of framing on Collins’s analyses of tacit knowledge, and this alone already enhances the interest of the present thematic issue.

42 Let us turn to the structure of this issue. The first critical contribution is written by Trevor Pinch, who once was a student of Collins and has collaborated with him for decades [Collins & Pinch 1979, 1982, 1993, 1998], [Pinch, Collins, & Carbone 1996]. Pinch is therefore very well suited to place the treatment of tacit knowledge in TEK in the context of the rest of Collins’s work. By so doing, he discusses a tension “between the realism underlying his [Collins’s] notion of ‘tacit knowledge’ and the constructivism underlying his [...] ‘experimenters’ regress’”. In relation to this challenge, Pinch addresses—as Lynch and Turner also do—the question of the ontological status of tacit knowledge and the empirical criteria according to which we may decide that someone has the tacit knowledge in question.

43 While Pinch focuses on Collective Tacit Knowledge, and Lynch concentrates on Relational Tacit Knowledge, both authors ask the same kind of question: should we take literally Collins’s “knowledge-as-stuff” metaphor and conclude that some subject possesses tacit knowledge as though it is some kind of “stuff”? Or should we treat tacit knowledge as an “actors’ category” used in relation to various aims, for example, to reinforce a cultural identity, secure the autonomy of science, enhance the status of a profession, or protect labor practices from managerial surveillance and control? Note that if, as the editors claim (see section 3.2), the attribution of tacit knowledge is under-determined, the two options are always possible with respect to the interpretation of a given situation (i.e., the analyst can either conclude that some real tacit knowledge is indeed possessed by the actors, or that the possession of tacit knowledge is claimed by the actors in order to achieve certain aims); moreover, the two interpretations are not necessarily mutually exclusive. Both Pinch and Lynch seem more sympathetic to the second option. In particular Lynch, as a student of Harold Garfinkel’s ethnomethodology in sociology, and as a well-known ethnographer of laboratory practices [Lynch & Sharrock 2010], pursues the second interpretation and suggests that an empirical sociological alternative [to developing a typology of tacit knowledge] is to investigate pragmatic and polemical uses of the tacit/explicit distinction in particular circumstances of action and conflict.

44 The next article is written by Stephen Turner. In his 1994 book, The Social Theory of Practices: Tradition, Tacit Knowledge, and Presuppositions [Turner 1994], Turner argues that the notion of practice conceived as a tacit stuff “shared” by a group, is beset by difficulties and fails to serve its purpose. In the present thematic issue, Turner criticizes Collins’s notions of “collectivity” and “collective” along the lines he set out in his 1994 book and in his 2011 review of TEK [Turner 2011]. Like Pinch, Turner concentrates on Collective Tacit Knowledge, but his aim is different. It is to examine the crucial idea behind the notion underlying Collins’s treatment of tacit knowledge, namely, the notion of “collective” upon which Collins has built his category of

Philosophia Scientiæ, 17-3 | 2013 12

Collective Tacit Knowledge. Turner distinguishes between the social and the collective. Whereas the former explains society using bottom-up explanations in terms of processes occurring between individuals, so that the social is reducible to relations among individuals, past and present, the latter is more ontologically laden. Like “Durkheim’s sociologization of Kant” it implies “collective, shared, psychological” irreducible contents, that are “necessary to account for social life.” Turner offers an analysis of the specific assets and difficulties of each approach, which he calls a “Scorecard”. The resulting “score” should, according to Turner, leave us unconvinced that “Collective Tacit Knowledge fills an explanatory need in a simple, or single cause, and in this sense ‘parsimonious’ way”. Turner arrives at this conclusion following an interesting line of thought. He starts with transcendental arguments of the kind “conditions for the possibility of” used by Kant and the neo-Kantians, and subsequently focuses on the long tradition of transcendental reasoning in which causal claims are deduced from transcendental arguments. He judges the latter to be vague and elaborates on two reasons for this vagueness: the problems of transmission and underdetermination. This brings him to question—like Pinch and Lynch but via a different route—the ‘realist status’ of tacit knowledge. When one reconstructs a position like ‘endorsing some tacit presupposition’, is this presupposition actually there, ‘in the head’ of some individuals or somewhere in the collective? Turner’s answer is negative, and he proposes an alternative interpretation.

45 The next two contributions come from scholars who were trained as philosophers, but in different traditions. In his paper, Thornton—who is himself the co-author of a recent book entitled Tacit Knowledge [Gascoigne & Thornton 2012]—criticizes Collins’s notion of “string” in relation to the issue of tacit knowledge. The culprit of Collins’s problematic use of strings is, according to Thornton, the way Collins has responded to (and misunderstood) Wittgenstein’s rules regress. Both Thornton’s and Soler-Zwart’s papers consider Collins’s four senses of “explicable” and ask the question of the relation between explicable in senses (1) or (2) on the one hand, and explicable in senses (3) or (4) on the other hand (see section 3.1 above). Thornton describes the success of the explication in senses (3) and (4) at a given time as an “action at a distance” with respect to explication in senses (1) and (2). According to his account, mechanization and scientific explanation of some knowledge K by some engineer or scientist, for example the construction of a robot able to ride a bike, would make the knowledge of people who know how to ride bike explicit “at a distance”. Yet, the people in question need not know anything about what the scientist or engineer has achieved, and remain as incapable as before of explicating how they ride a bike. Soler & Zwart in their article instead urge that we should always specify the sense in which some knowledge K has been explicated instead of just saying that K has been explicated ‘tout court’. The attributions “explicit” and “explicable” should always carry an index that specifies which of Collins’s four senses is meant. This would be sufficient to avoid confusion and the impression of counterintuitive consequences such as the one put forward by Thornton under the telling expression of “action-at-a-distance”.4

46 The main aim of Soler & Zwart is to elaborate and expand upon Collins’s tacit knowledge framework. The authors first argue that the identity of the RTK category and, as a corollary, what unifies its sub-cases, and makes them pertain to one and the same category, is not well-characterized in TEK. They suggest that a better characterization of RTK would be knowledge ‘not explicated for reasons that are

Philosophia Scientiæ, 17-3 | 2013 13

contingent with respect to the nature of K’, rather than ‘not explicated for relational reasons’. According to this suggestion, RTK is knowledge that is explicated or not for contingent reasons; thus in particular, it can be explicated or not by one and the same subject depending on the circumstances; or it can be explicable for one subject and not the other. Taking these possibilities into account, Soler & Zwart recommend that the terms “explicit”, “tacit” or “explicable” should always be accompanied by the specification for whom and when. This would help to avoid confusions or misunderstandings when using Collins’s distinctions. Moreover, the authors identify an ambiguity in the interpretation of RTK, STK, and CTK, and indicate how to avoid the related possible confusion. Using some new notations, they explicitly formulate a profound asymmetry between RTK on the one hand, and STK/CTK on the other, and introduce some possible sub-types of RTK. Finally, they consider dynamical aspects of Collins’s framework, and discuss the possibility—not considered by Collins—of transformations over time of STK to RTK and CTK to RTK.

47 The penultimate contribution of the thematic issue, preceding Collins’s response to the articles, is the ‘Collins-Reber piece’, the specific character and originality of which have already been stressed in section 1. To some extent, this piece is a report of a journey in which two well-established scholars who have studied tacit knowledge for decades, discuss that topic from their different disciplines and individual perspectives. It beautifully illustrates the difficulties scientists encounter when they try to understand each other’s intentions, use of terms, meaning of concepts, and main research questions on by and large the same subject. Aside from the particular topic of tacit knowledge, we recommend this article to anyone who is interested in framing, paradigms and incommensurability. Combining imagery from two articles of this thematic issue, we can say that Collins and Reber, after having passed for decades like “ships in the night”, finally came to see each other at the horizon, found a way to come closer without crashing, and finally navigated side by side, on the “wide river of tacit knowledge with islands of explicit knowledge popping up here and there and slowly developing into more extensive territories”.5

Acknowledgments

48 We would like to express our gratitude to a number of people and institutions for their indispensible roles in making possible the publication of this thematic issue. First of all, we are grateful to Harry Collins and to the members of the PratiScienS team who contributed both to the December 2011 conference and to the reports on the papers submitted for this thematic issue following the call for papers. Thanks also to some of the contributors of this issue who helped us to improve the English of this introduction, namely Harry Collins, Trevor Pinch, Tim Thornton, and especially Michael Lynch for the time and efforts he invested in compensating our lack of tacit linguistic knowledge and immersing us in Anglo-Saxon society. Additionally, we are grateful to the Editor in chief Gerhard Heinzmann and to the Managing Editor Manuel Rebuschi for accepting the proposal for this thematic issue, as well as to the Editorial Assistant Sandrine Avril for her keen help with realizing this issue.

49 In addition, Léna Soler would like to acknowledge the institutions that currently support or have supported the activities of the PratiScienS team: the Agence Nationale de la Recherche (ANR), the Maison des Sciences de l’Homme Lorraine (MSH), the Région

Philosophia Scientiæ, 17-3 | 2013 14

Lorraine, The Henri Poincaré Archives (Laboratoire d’Histoire des Sciences et de Philosophie), and the University of Lorraine.

50 Finally, Sjoerd Zwart would like to thank PratiScienS for the financial support that enabled him to visit Nancy and Paris to work on the present thematic issue.

BIBLIOGRAPHY

BLOOR, DAVID 1976 Knowledge and Social Imagery, London: Routledge & Kegan Paul.

CHOMSKY, NOAM 1990 Language and problems of knowledge, in The Philosophy of Language, edited by MARTINICH, A. P., Oxford: Oxford University Press, 275–298.

COLLINS, HARRY 1974 The TEA set: Tacit knowledge and scientific networks, Science Studies, 4, 165–186. 1975 The seven sexes: A study in the sociology of a phenomenon, or the replication of experiments in physics, Sociology, 9, 205–224. 1984 When do scientists prefer to vary their experiments?, Studies in History and Philosophy of Science, 15(2), 169–174. 1985 Changing Order: Replication and Induction in Scientific Practice, London: Sage, 1st ed. 1990 Artificial Experts: Social Knowledge and Intelligent Machines, Cambridge, MA: MIT Press. 2001a Tacit knowledge, trust, and the Q of Sapphire, Social Studies of Science, 31, 71–85. 2001b What is tacit knowledge?, in The Practice Turn in Contemporary Theory, edited by SCHATZCHI, T. R., KNORR CETINA, K. & VON SAVIGNY, E., London; New York: Routledge, chap. 7, 107–119. 2004 Gravity’s Shadow: The Search for Gravitational Waves, Chicago: University of Chicago Press. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

COLLINS, HARRY & KUSCH, MARTIN 1995 Two kinds of actions: a phenomenological study, Philosophy and Phenomenological Research, 55(4), 799–819. 1998 The Shape of Actions: What Humans and Machines Can Do, Cambridge, MA: MIT Press.

COLLINS, HARRY & PINCH, TREVOR 1979 The construction of the paranormal: Nothing unscientific is happening, in Sociological Review Monograph 27, On the Margins of Science, The Social Construction of Rejected Knowledge, edited by WALLIS, R., Keele: Keele University Press, 237–270. 1982 Frames of Meaning: The Social Construction of Extraordinary Science, Henley-on-Thames: Routledge & Kegan Paul. 1993 The Golem: What You Should Know About Science, Cambridge: Cambridge University Press. 1998 The Golem at Large: What You Should Know About Technology, Cambridge: Canto.

DAVIES, MARTIN 1989 Tacit knowledge and subdoxastic states, in Reflections on Chomsky, edited by GEORGE, A., Oxford: Blackwell.

Philosophia Scientiæ, 17-3 | 2013 15

DREYFUS, HUBERT 2004 La portée philosophique du connexionnisme, in Introduction aux sciences cognitives, edited by ANDLER, D., Folio/Essais, 2nd ed., 448–469, revised version of [Dreyfus & Dreyfus 1985].

DREYFUS, HUBERT & DREYFUS, STUART 1985 From Socrates to expert systems: The limits of calculative rationality, in Philosophy and Technology II: Information Technology and Computers in Theory and Practice, edited by MITCHAM, C. & HUNING, A., Studies in the Philosophy of Science Series, Boston: Reidel.

FERGUSON, EUGENE 1992 Engineering and the Mind’s Eye, Cambridge, MA: MIT Press.

GARFINKEL, HAROLD 1967 Studies in Ethnomethodology, Englewood Cliffs: Prentice Hall.

GARFINKEL, HAROLD, LYNCH, MICHAEL & LIVINGSTON, ERIC 1981 The work of a discovering science construed with materials from the optically discovered pulsar, Philosophy of the Social Sciences, 11(2), 131–158.

GASCOIGNE, NEIL & THORNTON, TIM 2012 Tacit Knowledge, Durham: Acumen.

KNORR-CETINA, KARIN 1979 Tinkering toward success: Prelude to a theory of scientific practice, Theory and Society, 8, 347–376. 1981 The Manufacture of Knowledge: An Essay on the Constructivist and Contextual Nature of Science, Oxford; New York: Pergamon Press.

KNORR-CETINA, KARIN & MULKAY, MICHAEL (EDS.) 1983 Science Observed, London: Sage.

KUHN, THOMAS S. 1970 The Structure of Scientific Revolutions, Chicago: University of Chicago Press, 2nd ed., 1st edition: 1962.

LATOUR, BRUNO & WOOLGAR, STEVE 1979 Laboratory Life: The Social Construction of Scientific Facts, London; Beverly Hills: Sage.

LYNCH, MICHAEL 1982a Closure and disclosure in pre-trial argument, Human Studies, 5(1), 285–318. doi:10.1007/ BF02127682. 1982b Technical work and critical inquiry: investigations in a scientist “re laboratory”, Social Studies of Science, 12(4), 499–533. 1985 Art and Artifact in Laboratory Science, London: Routledge & Kegan Paul. 1993 Scientific Practice and Ordinary Action: Ethnomethodology and Social Studies of Science, New York: Cambridge University Press. 2007 Expertise in action: Presenting and attacking expert evidence in DNA ‘fingerprinting’ cases, Villanova Law Review, 54(4), 925–952.

LYNCH, MICHAEL & SHARROCK, WES 2010 Ethnomethodology, Sage Benchmarks in Research Methods, London: Sage.

MACKENZIE, DONALD A. & SPINARDI, GRAHAM 1995 Tacit knowledge, weapons design, and the uninvention of nuclear weapons, American Journal of Sociology, 101, 44–99.

Philosophia Scientiæ, 17-3 | 2013 16

PINCH, TREVOR, COLLINS, HARRY & CARBONE, LAWRENCE 1996 Inside knowledge: Second order measures of skill, Sociological Review, 44(2), 163–186.

POLANYI, MICHAEL 1958 Personal Knowledge, Chicago: University of Chicago Press. 1967 The Tacit Dimension, Chicago: University of Chicago Press, 1st ed.

REBER, ARTHUR S. 1993 Implicit Learning and Tacit Knowledge: An Essay on the Cognitive Unconscious, New York: Oxford University Press. 2002 Tacit knowledge, in The International Encyclopedia of the Social and Behavioral Sciences, edited by SMELSER, N. & BALTES, P., Oxford: Elsevier.

RYLE, GYLBERT 1945-1946 Knowing how and knowing that, in Proceedings of the Aristotelian Society, vol. XLVI, 1–16.

SEARLE, JOHN 1995 The Construction of Social Reality, New York: The Free Press. 2001 Rationality in Action, Cambridge, MA: MIT Press.

SOLER, LÉNA 2009 Introduction à l’épistémologie, Paris: Ellipses, 2nd edition revised and completed. 2011a Tacit aspects of experimental practices: analytical tools and epistemological consequences, European Journal of Philosophy of Science, 1, 393–433. 2011b Les expérimentateurs sont-ils substituables les uns aux autres? (opacité des pratiques expérimentales, fragmentation de l’expertise, désaccords irrésolus entre praticiens: quelles conséquences épistémologiques ?), Le Philosophoire, 35 (La science), 65–113.

STICH, STEVEN 1978 Beliefs and subdoxastic states, Philosophy of Science, 45, 499–518.

TURNER, STEPHEN 1994 The Social Theory of Practices: Tradition, Tacit Knowledge and Presuppositions, Cambridge: Polity Press. 2011 Starting with tacit knowledge, ending with Durkheim?, Studies in History and Philosophy of Science, 42(3), 472–476. doi:10.1016/j.shpsa.2011.03.002

WITTGENSTEIN, LUDWIG 1953 Philosophical Investigations, Oxford: Blackwell, translated by G.E.M. Anscombe.

NOTES

1. PratiScienS stands for “Rethinking Sciences from the Standpoint of Scientific Practices”. The aim of the project is to develop a systematic analysis of the consequences of the so-called “practice turn” in the science studies and to investigate philosophically significant issues about scientific practices. One of these issues is the role of tacit aspects in the constitution of scientific results (understood in the broad sense of result: theoretical claims and systems of claims, experimental facts, technological achievements, mathematical theorems, etc.). Other central questions are the issue of how something acquires the status of a robust result in the empirical and the formal sciences or how we should assess the contingency or inevitability of robust scientific achievements. The project started in 2007 and is pursued by a small

Philosophia Scientiæ, 17-3 | 2013 17

interdisciplinary group of France-based historians, philosophers and sociologists of science (where ‘science’ must be understood here including mathematics and logic besides the empirical sciences). Beyond the core of the French-based researchers of the PratiScienS group, other researchers from many different countries are associated to the group. Besides Harry Collins, they include Hasok Chang, Karine Chemla, Peter Galison, Thomas Nickles, Andrew Pickering, Claude Rosental, Jean-Paul van Bendegem and William C. Wimsatt. 2. See in particular [Collins 1974, 1975], [Garfinkel 1967], [Knorr-Cetina 1979], [Latour and Woolgar 1979], and for the early 1980s, [Garfinkel, Lynch, & Livingston 1981], [Knorr-Cetina 1981], [Knorr-Cetina & Mulkay 1983], [Lynch 1982a,b]. In this context, [Bloor 1976] is also worth mentionning, since his and Barry Barnes’s “strong program” has been very influential in the social studies of science in practice, although Bloor did not himself conduct detailed empirical studies. 3. Beside Collins’s publications mentioned in note 1 and the references given note 2, here is a selection of works that deal with the tacit dimension of human knowledge (human knowledge broadly understood: from basic, everyday cognitive activities like linguistic or mnesic tasks, to very specialized theoretical and technical scientific practices): [Chomsky 1990], [Davies 1989], [Dreyfus 2004], [Ferguson 1992], [Lynch 1985n 1993, 2007], [MacKenzie and Spinardi 1995], [Pinch, Collins, & Carbone 1996], [Reber 1993, 2002], [Searle 1995, 2001], [Soler 2009, chap. IX], [Soler 2011a,b], [Stich 1978], [Turner 1994]. 4. Collins (private communication, 30 April 2013) agrees with Soler & Zwart but also thinks that the idea of explication at a distance is illuminating. 5. The first expression is borrowed from the Collins-Reber’s article, and the second from Collins’s response to the contributions.

AUTHORS

LÉNA SOLER Laboratoire d’Histoire des Sciences et de Philosophie, Archives H. Poincaré (UMR 7117), CNRS, Université de Lorraine, Nancy (France)

SJOERD D. ZWART Delft & Eindhoven Universities of Technology (The Netherlands)

Philosophia Scientiæ, 17-3 | 2013 18

Building an Antenna for Tacit Knowledge

Harry Collins

1 Introduction: The problem

1 I will try here to make the ideas in my book Tacit and Explicit Knowledge [Collins 2010], which I will refer to hereafter as “TEK”, more readily accessible.1 I am not going to write a summary but to try to explain the book to myself while “speaking aloud” as it were. Sometimes arguments and ideas become clearer, even to their author, when they are explained for a second or third time and I believe I am still finding out things about the book, especially the first part.

2 I have spent a large part of my academic life investigating the area of physics known as gravitational wave detection. The founder of the field was an electrical engineer and introduced the term “antenna” to describe the detectors. Like radio aerials, the detectors were meant to capture the vestigial influence of some otherwise invisible field and make its properties palpable. In the early days these antennae were metal cylinders on rubber mounts surrounded by liquid gases in pipes. The modest physicist who built the most reliable of these early machines often told me that the esoteric nature of gravitational waves did not concern him; he was “just a plumber”. I think that description fits me and TEK. TEK is an antenna built to capture and make palpable aspects of the mysterious tacit knowledge and I am just the plumber, albeit working with words and concepts, not metal and rubber. What we are both trying to do is make a machine that can enable us to catch aglimpse something ephemeral.2

3 TEK, I now realise, is based on a crude metaphor. Knowledge is a kind of “stuff” which you either have or do not have. If you have it you can do certain things, or understand certain things which you could not do or understand before; you are also enabled to make good judgments in respect of those things. For example, before you had some knowledge you could not build a working TEA-laser;3 once you have the knowledge you can build a working TEA-laser [Collins 1974]; [Collins 1985/1992]; before you had some knowledge you could not make good judgments about whether an interferometer

Philosophia Scientiæ, 17-3 | 2013 19

would be able to detect gravitational waves with certain waveforms; once you have the knowledge you can make such judgements [Collins & Evans 2007]; [Giles 2006].

4 A critic might say that the idea that there is some “stuff” that is passed along is redundant—at one time you cannot build lasers and make judgements about interferometers and at another time you can and that is all there is to it; no “stuff” is involved—after all, you cannot see or smell the “stuff” so it is an “idle wheel”. The response is that physical metaphors are still as useful as they have been through the history of physics: you can’t see, or smell, gravity or magnetic lines of force (or even, as Hume pointed out, causes in general) but they are still useful hooks on which to hang your thinking.4 The “knowledge-stuff” metaphor is an especially useful hook in two ways: one can ask about the kinds of knowledge that exist and, noting that certain machines and devices and certain animals and plants can do the things that humans learn to do only by being infected by knowledge, one is forced to ask whether what the non-humans have is the same knowledge-stuff as the humans have or, if not, why not and what is the difference.

5 Secondly, knowledge-stuff is divided into two kinds: there is “explicit knowledge-stuff” and there is “tacit knowledge-stuff”. It is tempting to say that you can see and smell explicit knowledge. It is tempting to say that this published paper is explicit knowledge and you can see it, smell it, and carry it around. None of this is quite true because the books and papers are not themselves explicit knowledge. They are, rather, the means by which explicit knowledge is transmitted. They are like the germs which cause a disease rather than being the disease itself. But at least you can see and smell the germs when explicit knowledge is involved.5 But in the case of tacit knowledge you can’t see the germs—not being able to capture the means of transmission of knowledge is what causes us to think of it as tacit knowledge. Where I argue in TEK that certain kinds of tacit knowledge can be made explicit, what I am doing is revealing how its transmission can be made visible and/or its implementation handled. Thus the most important difference between tacit and explicit knowledge is that explicit knowledge can be deliberately brought into existence at distant locations—the germs can be isolated and sent through the post or some such—while the germs of knowledge that is tacit cannot be sent around in this way because they cannot be seen or captured. Tacit knowledge is like a disease with invisible or as yet undiscovered germs; you cannot infect people through the post but it can be “caught” by those around you.

6 Thus we know of only one way to acquire tacit knowledge and that is to hang around with others who already have it.6 Explicating tacit knowledge is showing how it can be passed on in ways other than hanging about. The process which passes tacit knowledge around, so long as it has not been explicated—so long as the germs have not been made visible—is called “socialisation”. If I want to acquire the tacit knowledge pertaining to group “X” I need to spend time in their company interacting with them as much as possible; that way I increase my chances of catching their knowledge-disease. It is the way one learns to become fluent in a foreign language.

7 One can immediately see that the difference between tacit and explicit knowledge has enormous economic importance. Explicit knowledge is cheap to spread around—you can “broadcast” it and lots of people can buy it or get it for nothing. Tacit knowledge is much harder to distribute because to pass it on everyone has to be in the same location for a period. This means building schools and universities, laboratories, arranging for apprenticeships and so on—creating institutions in which people are brought together

Philosophia Scientiæ, 17-3 | 2013 20

so they can catch knowledge-diseases off other people.7 Businesses, if they want to steal tacit knowledge from other businesses, have to capture some of the people from the other business so they can hang around and infect everyone; it is often cheaper and easier to buy the entire business along with the workers’ tacit knowledge. Unsurprisingly, then, there is a lot of discussion about how and whether tacit knowledge can be converted into explicit knowledge. There are books in the management literature that claim to show how it is to be done.8 The question is also important to those who want to put human knowledge into machines—first it has to be captured in symbolic, that is explicit, form. As I will explain below, the answer to the question of whether tacit knowledge can be turned into explicit knowledge is that a little of it can, some of it can be converted in principle but not in practice, and a good bit of it cannot be converted in any way we can think of.

8 It was said above that if you have new knowledge you can do or understand new things. So suppose I enter the Amazon jungle, find an isolated tribe, and give them a copy of this paper. They won’t be able to do or understand new things—at least, not by using the written content of the paper—they won’t catch the disease. So it was an oversimplification to say that explicit knowledge can be “broadcast”. Explicit knowledge can be broadcast but only if the circumstances are right—the receivers of the knowledge have to be receptive to the infection.9

9 The detail of how it is that sometimes the transmission of explicit knowledge works and sometimes it does not is complicated but something general can be said at the outset. Where humans are involved, the receivers of explicit knowledge have to be fluent in the language of the transmission medium and fluency in language is acquired as tacit knowledge. This is one reason why Polanyi was right when he said: “[...] all knowledge is either tacit or rooted in tacit knowledge. A wholly explicit knowledge is unthinkable” [Grene 1969, 144]. Explicit knowledge is a lot more complicated than it looks at first sight.

2 The explicit and explication

10 Thought about more deeply, it is the fact that there can be explicit knowledge that is really difficult to understand.10 Thought about more deeply, there is nothing hard to understand about tacit knowledge. Since the creation of the first life forms, things have being doing things without being able to make explicit how it is they do them. Thus trees grow leaves and cats hunt mice without being able to explain how they do it and without being able to pass on those abilities in symbolic form. Likewise, we do not think it mysterious that we humans can pump blood round our bodies without being told how to do it nor learn to walk and speak without following a rule-book. What is exceptional is that sometime humans acquired the ability to talk about some of the things they could do and pass on these abilities by transferring materials with patterns on them. In the book I call such materials “strings”.

11 By the middle of the Twentieth Century, probably because of the optimism associated with the early computers, people thought it so normal to be able to represent and transmit abilities via strings that they had to invent a special term associated with abilities that could not be so transmitted. This term was “tacit knowledge”. Because all human abilities were talked of as manifestations of knowledge it was taken that those abilities that could not be passed around in the form of strings were manifestations of

Philosophia Scientiæ, 17-3 | 2013 21

tacit knowledge. Remarkably, tacit knowledge—not being able to represent what you could do with strings—came to seem to be something extraordinary and mysterious rather than the perfectly ordinary state of affairs.

12 But it is explicit knowledge that is the astonishing thing. Material things, such as vibrations in the air or marks on paper, pass around and, as a result, people can do or understand new things; there seems to be no causal relationship between the material scraps that are passed around and the abilities and understanding that they induce. There was never anything like this in the universe before the invention of language— the real “big bang”. The opening of Stanley Kubrick’s film, 2001: A Space Odyssey, shows early humans transformed into tool-users by coming into contact with a mysterious black obelisk. The development of the ability to use scraps of material to transmit knowledge and understanding should be thought about in grand terms like this; there must have been some equivalent of the black obelisk that made it possible for those strange scraps of material that I call strings to have such remarkable effects.11

13 Chapters 1-3 of my book and some of the Introduction are about explicit knowledge— how people pass on knowledge by passing around materials carrying patterns—strings. The term “string” has been used in other discourses, such as that of computer programming and, of course, physicists’ “string theory”.12 To save confusion I would probably have been better off inventing a new word such as “grom”. A string, as it is used here, is about as meaningless as a “grom”. A string is simply any material thing made of anything that is not completely plain or random. Vibrations in the air are strings, patterns of dirt on rocks are strings, and printed words are strings. The very non-specificity of the notion of string is important as it helps with understanding how strange the transmission of explicit knowledge is. It can be done with any material thing provided the conditions are right. The notion of “symbol” is far too confusing—a symbol is a string in the right circumstances. To ask the deep question about what are the circumstances that enable some scrap of material to act as a symbol you have to start with these essentially meaningless strings and struggle to work out how it is that sometimes they appear to convey meaning.

14 String transformations are fundamental to understanding how the world works. All languages, including purely spoken languages, have to be represented in some approximate way by strings. But languages are not strings. Strings are meaningless— just bits of stuff—whereas languages are meaningful. Using languages involves lots of string transformations so string transformation can easily be confused with language. In spoken language, for example, a set of mouth, throat and chest movements is transformed into a set of vibrations in the air, which are transformed into movements of the ear, which are transformed into a set of electric currents and so on. Usually the elements of the transformation process are transient; Edison invented a way to transform some usually temporary strings into more permanent strings.13

15 Strings carry “information” not meaning. The information that they carry has to do with the number and arrangement of their components. This is what “information theory” is all about. “Information theory” can tell you how to remedy losses in information transmission or at least how to measure them. It is because strings are without meaning that “information theory” is a success; there can be no comparable “meaning theory”. There can be no equally successful meaning theory because while information is a property of strings, meanings are a property of the societies in which languages live and continually evolve.

Philosophia Scientiæ, 17-3 | 2013 22

16 In spite of having written three chapters in TEK about the relationship between language and strings and the way knowledge is passed on using strings I do not understand how it works. All I manage to explain in TEK is some of the differences between occasions when knowledge transmission using strings does work and occasions when it does not work—I describe some of the “solutions” that can turn a failed attempt to transmit knowledge using strings into a successful attempt to transmit knowledge using strings. This does not solve the deep problem of explicit knowledge—I still do not understand the mechanism of those successes—I can only explain why the successes work better than the failures. But this does allow one to describe the process of explication even if it does not explain how there can be “an explicit” in the first place.

17 Tacit knowledge is knowledge that cannot be explicated or is not explicated—it is not or cannot be passed on using strings. Some of it may be passable by using strings in the future and some of it may be passable using strings in other places. A lot of what is thought to be irredeemably tacit can be explicated once we work out how and in other places it may not seem so irredeemable. If it is the case that tacit knowledge is knowledge that cannot be explicated then there are as many kinds of tacit knowledge as there are reasons that stop it becoming explicated. There are three kinds of reasons why tacit knowledge cannot be explicated so there are three kinds oftacit knowledge.

18 We still have to know a bit more about what “explicated” means. I argue in Chapters 1-3 of TEK that explicated can mean four things. One can be illustrated very simply. Somewhere in the text that you have read is a full-stop and printed within that full-stop in microscopic lettering is a message that is important to you and will empower you in certain ways. The full stop is what spies call a “microdot”. You did “see” that full-stop but it had no impact on you. Now, so long as the context is right, we can render the message in that full-stop explicit by physically transforming it. If we simply transform the full-stop and its message into something much larger, what was not explicit will become explicit and may now have the desired empowering effect upon you. If you knew where the full stop was and you had a powerful magnifying glass you could transform the string into something that could convey meaning to a fluent English speaker. You would see: “Drinking coffee makes your hair fall out”. This is an example of explication by physical transformation of a string.

19 That simple example explains a lot. When I push a CD into my computer’s drive and some words, or pictures, come up on the screen, physical transformation is what is going on. We say the computer is “reading” the disk but actually it is taking the marks on the CD, which you can’t read, and transforming them into marks you can read—just like looking at the full-stop through a magnifying glass; the computer is a kind of complicated magnifying glass. Both computers and magnifying glasses do physical transformations on strings. String transformation is the only thing that computers and magnifying glasses do.

20 Another kind of “explicating” is exemplified in TEK with an old joke about people who regularly go to a certain pub (sometimes it is the long-term inmates of a prison), who tell each other jokes by just shouting out numbers. Each number refers to a joke that everyone in the pub (or prison) already knows. For the rest of us the same joke has to be explained at full length if we are to understand it. The pub or prison situation is sometimes reproduced in ordinary life when long-standing members of bureaucratic organisation talk almost entirely in acronyms which outsiders cannot understand [TEK,

Philosophia Scientiæ, 17-3 | 2013 23

22–23]. The point is that sometimes a short string will have no knowledge-inducing effect but a longer string can induce the knowledge. Lengthening a string is the second kind of explication.

21 There are two other ways in which strings are transformed to the point they can be interpreted and which seem to fit into the way we use our language as examples of things being made explicit. The first of these is when we transform some causal/ behavioural sequence that we can accomplish without knowing how, into the form of some artefact we have created, such as a machine. A machine is a permanent string that produces temporary strings. (One should begin to see that the definition of string is so broad that it encompasses causal sequences and artefacts as well as patterns—the point is that causal sequences and artefacts are patterns.) Thus, when we make a machine that roughly reproduces some activity of ours that we are trying to understand (like balancing on a bike or playing chess), we can say that the working of the activity has been made explicit. For example, many people now think they understand chess because grand-master chess-playing machines have been built. Others say that these machines do not play chess like humans so human chess-playing has not been understood. We do not need to settle this argument to see that if the machines had reproduced human-like chess we would no longer say that chess always depends on tacit knowledge. Therefore humans’ ability to play chess must have been explicated. The point is that the workings of the machine and the behaviour of the machine can be readily interpreted: the machine is a string that can be easily interpreted by (certain) humans (that is, in this case, those who live in Western societies rather than the Amazon jungle)—it is, therefore, properly referred to as explicated knowledge.

22 The final way of talking about explication is closely related: it is when some sequence has been represented in the language of science—some set of equations or words—a string—which can be interpreted as explaining how something works. These four meanings of explicable are set out in Table 1, which is a reproduction of Table 4 on page 81 of TEK.

Table 1: Four meanings of explicable

1. Explicable by A longer string affords meaning when a short one does not elaboration

2. Explicable by Physical transformation of strings enhances their causal effect and transformation affordance

3. Explicable as A string is transformed into mechanical causes and effects that mechanisation mimic human action

4. Explicable as Mechanical causes and effects are transformed into strings called explanation scientific explanations

Philosophia Scientiæ, 17-3 | 2013 24

3 The tacit

23 So now I think I begin to see even better what is in my book. I make no attempt in TEK to explain why it is possible that certain things can be interpreted meaningfully by humans. It is just assumed that there are some things that mean something in one place and another thing in other places and sometimes things that are meaningful in one place have no meaning in others; the only question is why those meaningful things sometimes can, and sometimes cannot, be transmitted via strings.

24 Having set the problem up in this way we can get to the heart of the matter. The heart of the matter is found in Chapters 4, 5 and 6 of TEK. Ironically, the difficult part of the book is finding a way to understand the question and setting it up so that it is possible to answer it. That was the process of designing the tacit-knowledge antenna. Answering the question—using the antenna—is a comparatively straightforward business. There are three kinds of reason why meaningful knowledge is not, or cannot be, transmitted by strings and this means there are three kinds of tacit knowledge. It remains only to set out the three kinds of tacit knowledge: they are Relational Tacit Knowledge, for which I will use the acronym, RTK, Somatic Tacit Knowledge (STK) and Collective Tacit Knowledge (CTK). These three kinds can also be thought of as “weak”, “medium” and “strong” tacit knowledge.

3.1 Relational Tacit Knowledge (RTK)

25 Relational Tacit Knowledge is a ragbag category but the simplest of all once one has the idea. In RTK knowledge is tacit, for the time being, because social ‘r’elations are what they are. Change the social relations between people and what was once tacit can become explicit.

26 “Secrets” are an everyday example of RTK. Because their social relations can be competitive, scientists sometimes do not transmit strings to others even though they know they will be meaningful and empowering. A more subtle element of RTK is “mismatched salience”. In that case the “transmitter” has an incorrect model of the “receiver’s” knowledge and does not realise that some piece of information that seems perfectly obvious to the one party is not obvious to the other so it does not occur even to try and transmit it as a string; likewise, the receiver does not know what to ask. Closer social relations can resolve the problem.

27 Thirdly, there is knowledge that one party is unaware of possessing. The classic case concerns the position of the capacitor in the early days of the construction of the TEA- laser, see note 3 [Collins 1985/1992]. Here successful laser builders placed their heavy capacitors in the awkward “traditional” position—elevated and inverted—whereas those working from circuit diagrams would more likely place them on the bench, in which case the laser would not work. It was only later that the successful laser builders realised that the awkward positioning was essential because it shortened the top lead and lowered its inductance. So, in the early days the successful laser-builders could not have constructed the relevant empowering string however much they wanted to; later they could. Here the idea that the knowledge is tacit in consequence of social relations is strained but we can still find an excuse to use the ‘R’ in RTK by saying that new ideas are in-part expressed by changing social relations among scientists.14

Philosophia Scientiæ, 17-3 | 2013 25

28 RTK includes things like secrets which it seems gratuitous to refer to as anything as mysterious as tacit knowledge. A secret is something that is perfectly explicable but someone chooses not to explicate to someone else—why mystify it? RTK also includes things that are tacit for now but can become explicit in time—for example as scientific knowledge increases. But the category just about holds together as a species of tacit knowledge because in every case the knowledge, whether it secret or just currently not understood, can be passed on by “hanging around”—for example, hanging around successful laser builders. Secrets can be revealed by just watching what is going on—by spying; mismatched saliences will become matched, and traditional methods of doing things will be passed on just by hanging around without anyone realising their significance or even that it is happening. Hence RTK possesses the crucial characteristic that it cannot be passed on by strings—at least, not until or unless it becomes explicated—but it can be passed on through social contact. This is what ties it to the other categories of tacit knowledge though sometimes it seems a bit of a strain.

29 Note, that in principle there is nothing to stop every piece of RTK from becoming explicit: secrets could be told; if one had a better model of the recipients’ knowledge one could pass on bits of knowledge in the form of strings that one had not initially realised needed to be passed on; and as more science is done (category 4 of explicability) the importance of things like the inductance of the top lead become clear enough to be expressed in strings. Of course, this does not mean that all the elements of RTK can be told here and now or even that all the elements can ever be explicated; what it means is that any one element can be explicated given the will, the understanding and the effort. That said, there will always be RTK because there will always be secrets; we will never have complete knowledge of the contents of others’ heads; and science will never be finished. The continued existence of RTK is more a matter of social relations and of logistics than of epistemology but, like death and taxes, social relations and logistics are always with us.

3.2 Somatic Tacit Knowledge (STK)

30 Somatic Tacit Knowledge is tacit because it is stored in the arrangements of the body. The classic case of STK is the ability to ride a bicycle—the case discussed by Polanyi. As Polanyi famously said: If I know how to ride a bicycle [...], this does not mean that I can tell how I manage to keep my balance on a bicycle [...] I may not have the slightest idea of how I do this, or even an entirely wrong or grossly imperfect idea of it, and yet go on cycling [...] merrily. Nor can it be said that I know how to bicycle [...] and yet do not know how to co-ordinate the complex pattern of muscular acts by which I do my cycling [...] I both know how to carry out (this performance) as a whole and also know how to carry out the elementary acts which constitute (it), although I cannot tell what these acts are. [Polanyi 1966, 4]

31 This is an iconic description of the meaning of tacit knowledge and it fits well with much of the philosophy that has to do with the body. For example, Dreyfus and Dreyfus’s “five stage” model of skill acquisition is to do with the “internalisation” of skills such as car-driving: it is only novices who rely on trying to follow explicit rules whereas true skills involve unconscious processes of far greater complexity which cannot be captured by any set of rules [Dreyfus & Dreyfus 1986]. True skills of this sort

Philosophia Scientiæ, 17-3 | 2013 26

are embodied and not transformable into strings. This view also fits well with that of philosophers such as Heidegger and Merleau-Ponty.

32 And yet there is something badly wrong with Polanyi’s understanding of bicycle-riding. One can begin to see why by looking only a couple of pages further on in the source from which the above quotation is taken. Polanyialso says: [I]n order to compensate for a given angle of imbalance α we must take a curve on the side of the imbalance, of which the radius (r) should be proportionate to the square of the velocity (v) over the imbalance r ∼ v² / α. [Polanyi 1966, 6–7] 33 In other words, Polanyi says there is a set of rules for bicycle riding—it is just that we humans cannot follow them. To see what is being claimed here, imagine the following: we are trying to ride a bicycle on the surface of a small asteroid so that the pull of gravity is only a millionth of what it is on Earth and the bike, when it becomes unbalanced, falls so slowly it is, as the English expression goes, like watching paint dry. In such a case there would be no reason why we might not carry a booklet with Polanyi’s instructions for bike-riding and a few geometrical instruments and keep the bike balanced in rather the same way as we currently assemble flat-pack furniture—by reference forward and back to the instruction book while using a few tools. Now put the bike back on Earth and imagine that our brains and bodies worked a million times faster—then balancing on the bike would still be like balancing on the asteroid; it would be possible to manage via explicit knowledge—interpreted strings. So while Polanyi may be perfectly right about how we humans actually balance on bikes, and it involves “somatic tacit knowledge”, it does not follow that it is impossible to make this knowledge explicit and usable by different kinds of creatures with faster processing and reaction times—and it should come as no surprise that there are machines that can balance on bicycles. So Polanyi himself has made the knowledge of bike-balancing explicable in the sense of Table 1 category 4 and clever engineers have made it explicable in the sense of category 3.

34 “Somatic limit tacit knowledge” is knowledge that remains tacit in use solely because of the limited capabilities of human brains and bodies. I argue in the book that all somatic tacit knowledge—which is the staple of the phenomenological literature that is concerned with the body—is like this. It is much less mysterious than it has been made out to be and each bit of it could, in principle, be rendered explicit in the way just demonstrated in respect of the bicycle. The same goes for the abilities of cats, dogs, trees and sieves though, as I show, the case of something as simple as a sieve sorting stones, it can be extremely complicated actually to make some physical process explicit. Here, however, the limit to the explicability of the knowledge is, again, not epistemological, but a matter of the limits of our scientific abilities. Though we would struggle to reproduce the sorting method of a sieve—which depends, for example, on the exact shape of the stones, many coefficients of friction and just how the sieve is shaken—a much more accomplished scientific race, for whom the sieve problem would be trivial, might have explicated much of even the STK of our bodies and brains. We humans, of course have to continue to use this knowledge as tacit knowledge— something stored in the materials of our bodies (and how we do this remains an interesting topic for analysis), but, again, this is just a matter of how we happen to be constituted.

35 TEK also argues that certain elements of STK may never be explicable in terms of category 3—that is, we may never be able to reproduce what we do mechanically—

Philosophia Scientiæ, 17-3 | 2013 27

because of the very properties of the physical materials of which we are made. That is to reproduce us mechanically, the materials and mechanism would have to be identical —so it would be like making a three-dimensional photo-copy of us rather than reproducing in another form. That is not explicating, only copying (though one might need to explicate in order to understand what it would be to make an identical copy since there are an indefinite number of dimensions of possible difference). When it is the case that reproduction of the materials of the human body are essential to the reproduction of the actions they can carry out I call it “somatic affordance tacit knowledge”—it is the materials of body itself, or of the brain, which is part of the body, that affords the possibility of the action in question.

3.3 Collective Tacit Knowledge (CTK)

36 Collective Tacit Knowledge is the most profound case. To see what it is we can begin, once more, with Polanyi and the bicycle. What has been dealt with in the last section is what can be called “bicycle-balancing”. It is all about how one keeps the bike upright on the asteroid, on Earth, or whatever. But there is an entirely different aspect of bicycle-balancing which Polanyi does not deal with and for which there truly is no formula. This is bike-riding in traffic. Dreyfus and Dreyfus do discuss car-driving in traffic but they do not seem to notice the difference between this and controlling the car—the gears, the steering, and so forth; for Dreyfus and Dreyfus there is no discontinuity between the STK of the gear-shift and the very different kind of tacit knowledge needed to drive “in social concert”.

37 To start to see the difference one can note that balancing on a bike, or mechanically controlling a car, is the same process wherever it is executed whereas riding/driving in traffic is hugely different in different places and in different circumstances. Riding my bike in Cardiff I know I can ride on the pavement (sidewalk) without anyone bothering me—pedestrians will get out of my way with a smile so long as I am polite. I noticed the same thing in Tokyo recently. But there are other towns even the UK where it would be unthinkable to ride your bike on the pavement. I know that in the UK I can cross a road on my bike against a red light so long as I can “catch the eye” of the drivers who are waiting to go as soon as the green light comes on in their favour and they indicate their agreement with the tiniest acknowledgement. But I also know that if I ride in the UK at night I would be well-advised to have lights and that if I am driving my car I am entitled to be furious at any bike-rider who does not have lights. In contrast, in Beijing bike riders ride everywhere at night without lights and no-one gets furious. I notice that car-drivers in China do things, like cutting corners, and driving on the wrong side of the road, that would be unimaginable in Europe or America. But I notice that you can do many more outrageous things in Italy when you drive a car than you can in the UK and no-one gets angry [Collins 2004, 398–399].

38 So what I am doing here, so it seems, is providing strings that will enable you to understand how to ride your bike or your car in different countries. But this is misleading. First, my description of Chinese driving was of a few years back—perhaps it has all changed. And which are those towns where you can and can’t ride on the pavement? The recipe—the set of strings—is never fixed and never predictable because it is the property of entire societies and the way societies change is unpredictable. To know what the rule is I have to go to the society and find out by being there—there is

Philosophia Scientiæ, 17-3 | 2013 28

no rule about how frequently and how fast the rules change. What can be done is to say “in that frozen moment of past time, 2005, these were some rules for riding bikes in those parts of Beijing that I visited”, but that does not tell you how to ride on your next visit.

39 Much more important, I have provided no string for telling you how to “catch the eye” of a car driver nor how to recognise whether the look they return is an acknowledgement of the fact that they are not going to run me over when I cross in front of them or not. And I cannot provide such strings. Such things are only acquired through immersion in the society.

40 Riding-in-traffic has the same structure as language fluency. Fluency in languages cannot be acquired from the most exhaustive study of dictionaries, grammar books and guides to foreigners’ habits, it can only be acquired by spending time in the language- speaking community. The way to think about this kind of knowledge is as belonging to the collectivity, not the individual. That is why it is called “Collective Tacit Knowledge”. Language is not under anyone’s control—what happens in the whole community is what determines what can be said and what cannot be said at this time and place while even the rules for a complete description of a frozen moment are indefinitely long. All this is true of every kind of collective tacit knowledge.

41 What we as individuals do is borrow the knowledge of collectivities for a short time. We borrow it in the typical way of catching tacit knowledge—which is the typical way of catching diseases: we hang around with people who already have the knowledge/ disease. But if we stay away from the communal “host” on which we are a “parasite” our disease will begin to mutate. Our language will degrade and depart from the host language—it will become archaic, or come to resemble a cargo-cult language that no longer reflects the meanings of the same terms in the host. It is the same with all CTK. It is only CTK which we can think of no way to explicate it in any of the four senses of explicate—not even in principle. All we parasites can do, to change the metaphor, is suck continually on our collective host’s CTK-blood.

Conclusion

42 The term tacit knowledge, as it is used in academia, applies to three very different things that are usually not distinguished. This is not surprising because in most cases of the acquisition of tacit knowledge, all three kinds are being acquired at the same time in a very confounded way. Because of this, all manner of false claims have been made about the way the tacit can be turned into the explicit or the tacit cannot be turned into the explicit. Only by keeping the three types of tacit knowledge separate will we understand what is really going on when knowledge is transferred.

Philosophia Scientiæ, 17-3 | 2013 29

BIBLIOGRAPHY

COLLINS, HARRY 1974 The TEA set: Tacit knowledge and scientific networks, Science Studies, 4, 165–186. 1985/1992 Changing Order: Replication and Induction in Scientific Practice, Chicago: University of Chicago Press, 2nd ed., 1st edition Sage publications, 1985. 2004 Gravity’s Shadow: The Search for Gravitational Waves, Chicago: University of Chicago Press. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK]. 2012 Drei Arten impliziten Wissens, in Implizites Wissen – Epistemologische und handlungstheoretische Perspektiven, Weilerswist-Metternich: Velbrück Wissenschaft, 91–107.

COLLINS, HARRY & EVANS, ROBERT 2007 Rethinking Expertise, Chicago: University of Chicago Press.

DREYFUS, HUBERT L. & DREYFUS, STUART E. 1986 Mind Over Machine: The Power of Human Intuition and Expertise in the Era of the Computer, New York: Free Press.

GILES, JIM 2006 Sociologist fools physics judges, Nature, 442, 8.

GRENE, MARJORIE (ED.) 1969 Knowing and Being: Essays by Michael Polanyi, London: Routledge & Kegan Paul.

POLANYI, MICHAEL 1966 The logic of tacit inference, Philosophy, 41(155), 1–18.

WINCH, PETER 1958 The Idea of a Social Science, London: Routledge & Kegan Paul.

NOTES

1. This introduction takes off from my “Three Kinds of Tacit Knowledge”, which was commissioned as an introduction to Tacit and Explicit Knowledge for a German audience. It was published as [Collins 2012]. In re-writing the piece for this volume I have made so many additions, changes and reorganisations that, though the overall framing of the piece remains the same, it merits a new title. To save confusion for readers who might encounter both versions, I have indicated many of the substantive changes with footnotes. I thank the editors of this volume for triggering the transformation of the piece. 2. The antenna metaphor and this entire paragraph are new. I will probably include the antenna metaphor should there ever be a second edition of TEK. 3. TEA-laser stands for “Transversely Excited Atmospheric pressure carbon dioxide laser”. This was a kind of very powerful gas laser that was invented in the late 1960s and was very difficult to reproduce—at least at first. 4. None of this is spelled out in the book so I am already understanding it better. 5. In the original I talked of the transmission of explicit knowledge rather than the transmission of the means to engender explicit knowledge. I apologise to readers of my work for sometimes referring loosely to speech, writing, and so forth, as explicit knowledge rather than as the means of transmitting explicit knowledge. I don’t think any serious conceptual problems have resulted from this loose talk but corrections need to be made when exactness is an issue.

Philosophia Scientiæ, 17-3 | 2013 30

6. The steadily increasing “bandwidth” of the internet makes it tempting to think that one can “hang around” with people just by interacting with them electronically—hence the popularity of the idea of distance-learning and the like. But this is a mistake recognised by all those who are serious about communicating—we still have universities and schools and the airways are still clogged with businessmen and academics travelling to meet their colleagues. 7. Ordinary diseases spread through distributed societies pretty fast but the tacit knowledge disease is caught very slowly so the institutions that keep groups together for a long time are necessary. 8. Methods of converting tacit knowledge tend to deal with tiny areas of the problem with no grasp of how little it is that is being accomplished. 9. This context-boundedness creates a terrible mess exemplified by literature on semiotics which really struggles to work out what is a sign, what is an icon and so forth because the meanings of those terms are always shifting with context. I try to start from more basic elements, not signs or icons with their inherent implication of meaning, but with marks and patterns that might or might not be meaningful. 10. A vast amount has been written about how messages become “distorted”, how “the author is dead” and how meanings are “socially constructed” but here we are dealing with a different subject—the fact that sometimes a message does empower its receiver in roughly the ways one might expect. 11. This paragraph is entirely new to this publication—I will include the sentiment should there ever be a new edition of TEK. 12. This passage is new to this version of the paper. 13. Are the indentations on a gramophone record symbols? They are strings. 14. “The social relations between men and the ideas which men’s action embody are really the same thing considered from different points of view [...]” [Winch 1958, 121].

ABSTRACTS

My book, Tacit and Explicit Knowledge, is introduced. The introduction is also helpful in explaining the book to me, the author.

Mon livre, Tacit and Explicit Knowledge, est introduit. L’introduction est également utile pour m’expliquer le livre à moi-même, l’auteur.

AUTHOR

HARRY COLLINS

School of Social Sciences, Cardiﬀ University, Wales (UK)

Philosophia Scientiæ, 17-3 | 2013 31

Tacit Knowledge and Realism and Constructivism in the Writings of Harry Collins

Trevor Pinch

1 I want to begin this paper by praising Collins’s latest book on tacit knowledge Tacit and Explicit Knowledge [TEK] [Collins 2010]. I should declare a certain degree of self-interest. Collins was my mentor and I have ended up becoming his friend, colleague, and coauthor. This of course only makes my task harder! TEK overall is vintage Collins. It is his best book since Changing Order [Collins 1992], not only because it is one of his shortest (and hence liable to be read as opposed to his massive and impressive tome on gravitational radiation, Gravity’s Shadow: The Search for Gravitational Waves [Collins 2004]), but also because as with the earlier book, Changing Order, Collins shows great ingenuity in the examples he conjures up. Changing Order brought us not only a rich array of fieldwork instances, but also funny creative ideas like Popper Mice to illustrate the problem of induction. In TEK we also find numerous tantalizing ideas and examples culled not only from his own fieldwork, but also from his own experiences, reading, imagination, and his earlier books. Some gems include: a sociological version of string theory with analog and digital strings (to be used in thinking about explicit knowledge); husbands grunting at wives (is it a language?); saluting machines (we would be mad to build them); vegetarian dogs (could they even exist?), how androids learn to dance on Star Trek (can a robot carry out improvisation in the same manner as learning the basic steps of dancing?) and my personal favorite, Chinese cars exiting highways by the entrance ramp and the oncoming cars splitting neatly into two lanes to let the exiting vehicle pass (an example of what he calls Collective Tacit Knowledge). I have never been to China to witness such a terrifying spectacle.

2 TEK also has a nice counter-intuitive feel to it in that Collins spends the early part spelling out what explicit knowledge is, arguing that in a way the deeper issue is exploring what explicit knowledge consists of as opposed to tacit knowledge. Who is TEK written for? It is written for a wide audience who use and ponder the idea of tacit knowledge. This includes not only Science and Technology Studies scholars,

Philosophia Scientiæ, 17-3 | 2013 32

philosophers and psychologists, but also people in business schools, educators, and organizational sociologists. The book should also be of wide interest to researchers working in robotics, HCI, and Artificial Intelligence. The main theoretical protagonists are, the chemist turned philosopher, Michael Polanyi—who first introduced the term tacit knowledge—and the philosopher Hubert Dreyfus, who agrees with Collins on most issues but who according to Collins gets it wrong mainly on the issue of Somatic Tacit Knowledge. To be added to this list is Bruno Latour and his many followers and acolytes (and especially the “lazy” followers!) whom Collins disagrees with. Theoretically TEK builds on the three inspirations for much of Collins’s work; the later philosophy of Wittgenstein, Peter Winch’s little book The Idea of a Social Science and Thomas Kuhn—all, of course, given a sufficiently radical reading.

3 The main idea in the section of the book on tacit knowledge is to show that the term has been used to cover three different sorts of knowledge that need to be separated. Collins calls these three forms of tacit knowledge, Relational, Somatic, and Social. Relational Tacit Knowledge is knowledge that could be spelt out in principle but often stays implicit because of the way we organize tasks and pass on knowledge. In short whether it is made explicit or not depends on our relation to other communities or to the individual to whom the tacit knowledge is being passed on to. For instance, a crucial component in the replication of a scientific experiment, such as the length of leads on an electronic component in building early lasers, might not be specified by the scientists even though getting its length correct may be crucial to other scientists trying to gain the skills to repeat the experiment. The length could in principle be made explicit. Somatic Tacit Knowledge is tied to the corporeality of our bodies; it is in a way a form of bodily knowledge, such as learnt in playing a musical instrument. Your hands will correctly glide over the keys of the piano to play a Scott Joplin piano rag even though your mind may not be explicating each note you need to hit. Social Tacit Knowledge, the third form, is clearly for Collins the most important and is unlike Relational Tacit Knowledge which Collins calls “weak tacit knowledge”, and Somatic Tacit Knowledge which Collins calls “medium”. It is the sort of knowledge which is very specific to the living of a particular form of life, such as acquiring the rules to follow the rules—a point made by Wittgenstein and Winch. Because it depends upon the meaningful interpretive dimension of social life as lived within a very specific social context, even in principle Social Tacit Knowledge cannot be explicated in full. It is the sort of knowledge one needs to exit the highway safely in China.

4 This new three-fold distinction between, Relational, Somatic and Social Tacit Knowledge, reveals that Collins’s own thinking on tacit knowledge is itself in movement. To make the new categories work he has to reinterpret his own earlier studies, such as the classic study of scientists building and attempting to replicate the first TEA Lasers [Collins 1974], the work of UK scientists trying to reproduce a measurement carried out on small sapphire crystals in the Soviet Union, known as the Q of Sapphire [Collins 2001], and parts of his earlier book, Artificial Experts: Social Knowledge and Intelligent Machines [Collins 1990] on the debates over expert systems and their role in Artificial Intelligence. Collins also uses ideas developed in earlier books such as the distinction between mimeomorphic and polimorphic actions outlined in the book with Martin Kusch, The Shape of Actions: What Humans and Machines Can Do [Collins & Kusch 1998], which is itself a refinement of the distinction between what he called “behavioral specific actions” and other sorts of actions first delineated in Artificial Experts. Collins also adds to the mix by introducing the distinctions between different

Philosophia Scientiæ, 17-3 | 2013 33

kinds of expertise, such as interactional and contributory expertise outlined in the book with Robert Evans, Rethinking Expertise [Collins & Evans 2007].

5 I want to focus here upon the sort of evidence Collins claims in support of his ideas of tacit knowledge. Tacit knowledge is famously a difficult thing to show empirically because due to its nature it cannot be articulated. It is in a way a negative (in the sense of something absent) and showing a negative empirically is always hard. One way to show it in the case of science is to point to the difficulties scientists have in repeating experiments and measurements. At one time, T(1), these measurements or experiments cannot be performed successfully, but at a later time, T(2), they can be successfully carried out. What has changed between T(1) and T(2)? The answer is that the scientists have acquired tacit knowledge. It is thus necessary to show that something was missing from the experimental activity which occurred at T(1). But the exact tacit knowledge that was missing can only be inferred because if what was missing could be spelt out it would mean that tacit knowledge could itself be explicated and it is the defining feature of tacit knowledge that it is knowledge which cannot be explicated (except of course in the case of Relational Tacit Knowledge—see below). Collins makes great use of examples where we can infer the missing knowledge. He draws especially upon his own earlier TEA laser study and the later Q of Sapphire study. These are both examples where scientists fail to do something earlier but are successful later. By interacting with and studying the scientists who struggle and fail and then who later succeed, Collins uses a kind of ethnographic fieldwork to elicit the missing tacit knowledge. His fieldwork is very rich, such as the follow up TEA laser study completed in 1975 with Robert Harrison where Collins actually worked alongside Harrison as he struggled to make his laser work [Collins & Harrison 1975]. The later “success” in replications forms as it were a control experiment as Collins and the scientists ponder over what exactly has changed and what they didn’t grasp quite right in their earlier failed attempts. These studies seem quite convincing at pulling out the inferred tacit knowledge and are some of the most careful studies of this type in the sociology of science. But note that these two studies mainly show what Collins now calls “Relational” Tacit Knowledge— that is tacit knowledge that could in principle be made explicit given the right circumstances. This gives the studies in hindsight a certain persuasiveness because some of the missing tacit knowledge can actually be explicated because it is relational in character.

1 A demonstrable example: broken text and repair

6 Now I move on to examine some of the newer different examples Collins brings to the table in his new book TEK. Collins’s examples at first blush seem compelling. One of the most striking come in a section where Collins wants to show how we as humans are good at repairing passages of speech we mishear or words which are misspelled. We do this routinely by trying to make sense of the words. Collins highlights two paragraphs of text which both contain “broken” words (words that are misspelled) [TEK, 115]. One paragraph (the first one) appears as gibberish with the words not only misspelled but also seemingly rearranged randomly whilst the second paragraph, although full of misspellings, keeps the words in the correct order. The reader is invited to read the text him or herself. The example is used to demonstrate “our everyday ability to make good something broken while hardly thinking about it as long as it makes sense to us”

Philosophia Scientiæ, 17-3 | 2013 34

[TEK, 115]. I participated in this demonstration myself as a reader. Sure enough, when instructed to read the first passage, I could not make sense of it and quickly moved on to the second passage which initially looked equally baffling and soon I had that “Ah hah” moment as I found the second passage, although containing the same broken words,started to make sense.

7 Collins uses this as a “demonstration” to argue for three things: (1) that because most readers can read the second passage so easily it must mean that reading is accomplished by searching for meaning, (2) that although Collins’s own spell-checker highlighted each word with a jagged red line, the copy editor at Chicago University Press will not even think about correcting either paragraph, and (3), that the reader will know when to give up reading the first passage but will persevere when reading the second.

8 Now it might seem churlish to try and cast doubt upon the power of this demonstration but I will try to do so. The first thing to note is that readers, especially reading books quickly, do not all read in a linear manner. In other words Collins has no control over how the reader actually reads the text and this may make a difference. For example readers often scan texts that they are about to read. When confronted with the two texts which appeared both at the scanning level to be gibberish I noticed my eyes were drawn to the misspelled word “Cmabrigde” which appears in both texts as this happens to be how I regularly mistype the word Cambridge (I often invert the order of letters when I type quickly, for example frequently typing “teh” for the word “the”). In the second text the misspelled word “Cmabrigde” is followed by the word “Uinervtisy” which is sometimes how I mistype the word “University” and which may have made me think that reading the second text would be more rewarding as it contained the meaningful (but misspelled) pair of words “Cambridge University”. This all of course all happens in the flash of an eye, but like subliminal cueing in advertizing experiments, this may make a real difference to how the task is carried out and what the results mean. One could also point out that the statement (2) above is not strictly correct as there are actually 41 words correctly spelt in both paragraphs, so Collins’s spell- checker will not highlight “each word” as is claimed. There are also plenty of copy editors who would try to correct or at least highlight the misspelled marked words despite it being part of an exercise. In my empirical work as a sociologist I sometimes use transcribed speech excerpts with phonetic spellings and, despite explicit instructions, proofreaders and editors are always trying to clean them up. One could also point out on (1) that some subset of people (including my father and sister) suffer from a reading disability where they have to process each letter in a word at a time and hence would find it extremely difficult to read the second passage as well, and hence some small subset of readers will not as claimed in (3) persevere when reading the second passage. Also I noticed that the sentences in the two paragraphs are of different lengths. The first paragraph starts with a 7-word sentence and the second with a 12- word sentence. I am assuming this does not matter but it is a difference. Another difference is that the first paragraph contains a grammatical clue as to it being gibberish which is not present in the second paragraph—a question mark appears in the middle of the word “huh?and”. Maybe this difference provides an added sign that the first paragraph is likely to be “gibberish” and not worth pursuing. Knowing this— when to stop reading a text because it is likely to be gibberish—is what Collins calls part of “Collective Tacit Knowledge”. So the example would still work, but it would work to

Philosophia Scientiæ, 17-3 | 2013 35

make a different point, namely demonstrating our ability to recognize gibberish rather than our abilityto do repair.

9 These are, of course, quibbles—overall the demonstration works, it is convincing in the way that demonstrations should be according to Collins’s own writing on the difference between experiments and demonstrations. In his book The Golem at Large he claims, “demonstrations are designed to educate and convince” [Collins & Pinch 1998]. The reader is not being asked to experiment with the text in the sense of finding out new things about how reading broken text might be achieved but rather we are being asked to be convinced of “our everyday ability to repair”. And at first blush I was convinced.

2 A canonical example: riding a bike

10 I chose the broken text example because I believe that the “demonstration” character of many of the examples gives them a different warrant than the fieldwork cases. It is the very “demonstrable” quality of the examples which makes them powerful. This is the beauty of the most famous example of all in the tacit knowledge canon: Michael Polanyi’s example of riding a bicycle. We do not have to describe the detailed process of learning to ride a bicycle when using this example because nearly everyone knows from their own experience how they learnt to ride a bicycle. The key part of learning to ride a bicycle—the ability to balance and ride independently—seems to have been acquired in the same manner by everyone at least since bicycling became a common part of our shared culture. Nearly every child has shared that thrilling moment of “going solo”—riding on their own without training wheels or a parent or sibling pushing them and knows they have accomplished something. Bicycle riders also readily agree, as soon as it is pointed out to them, that they didn’t acquire this knowledge by reading books or by learning about the physics of balance or by studying in a master class for bike riders! There is no need to actually stage a demonstration of all this because everyone is familiar with it already from personal experience. In a way the riddle of how to show that something that is tacit and hence missing can be made convincing is by appealing to a canonical example based on something even more compelling than demonstration—personal experience—a move which appeals to “what everyone knows” and which is therefore pointless to question.

11 Many of the examples given in TEK are backed up with Collins’s own fieldwork where he teases out exactly how the knowledge is acquired and what sort it is. But nearly all these examples are given in Chapter 4 on “Relational Tacit Knowledge”. It turns out that Chapter 6, “Collective Tacit Knowledge and Social Cartesianism”, the aspect Collins considers the “irreducible heartland of the subject”, has the fewest examples from fieldwork. Indeed the only boxed example is the famous quote from Wittgenstein about if a lion could speak we would not be able to understand him. He does say that fuller treatments are available elsewhere and that he will only elaborate on issues on how Collective Tacit Knowledge relates to Somatic Tacit Knowledge (discussed in Chapter 5). But I worried about this lack of detailed case studies in the heartland of the subject.

12 There is also something very different about the science examples Collins discusses from his fieldwork and other examples in the book. In the science examples there are very clear criteria for knowing when the scientists have acquired sufficient tacit knowledge to be successful. In the TEA laser example this is famously when the laser will vaporize a block of concrete. The Q of Sapphire example is messier because there is

Philosophia Scientiæ, 17-3 | 2013 36

no dramatic outcome to show that the correct outcome has been achieved, but, as Collins points out in his fieldwork on that case, it seems clear when the scientists are able to correctly measure the Q factor, obtaining the same high values in the UK as were obtained in similar circumstances as earlier by the Russian group. The successful measurement, unlike, say, results in parapsychology and cold fusion, does not contradict existing knowledge. Thus the Western scientists when they first learned of and later watched the Russian scientists at work in their laboratory in Moscow successfully measuring the Q of Sapphire did not doubt the measurement could be performed. They came to trust the Russian scientists enough to believe that the high Q could be measured, even though at that stage they were not able to repeat the measurement themselves. Thus when the Q of Sapphire was finally measured in the UK it was immediately recognized that it had been done so successfully. Of course as Collins points out, accomplishing a task such as building a laser successfully once does not guarantee that you can do it again—the measure of the skill being acquired is being able to reproduce the result of interest. This is an important point and an issue we will return to later.

3 Bicycle riding revisited

13 This lack of examples for the existence of Collective Tacit Knowledge puzzled me. But perhaps there is something about the very notion of Collective Tacit Knowledge which makes it hard or impossible to illustrate with “knock down examples” or “demonstrations” or even with convincing fieldwork. Take for instance again the famous bike riding example which Collins now recasts as demonstrating Somatic Tacit Knowledge. Recently I observed my neighbors’ five-year old son, Sky, learn to ride a bike. He learnt in the standard way by his mother pushing him and letting go on the quiet suburban dead-end street on which we live. As we stood around on our driveway celebrating Sky’s success on the road, Sky turned into our driveway at high speed and, much to our horror, we realized he had no clue as to how to stop or, as Collins would put it, “navigate pedestrian traffic”. We leapt out of way as he skidded off his bike on the tarmac drive—he burst into tears, his pride hurt more than his body. This example is on the border line of Sky mastering bicycle riding as part of Somatic Tacit Knowledge and dealing with traffic (according to Collins part of Collective Tacit Knowledge). But how would we convincingly demonstrate how and when Sky had achieved Collective Tacit Knowledge as a competent bike rider? In other words how do we tell that he has learnt how to deal with traffic? We would have to follow him as he negotiated the main road at the end of our street, turning in various configurations, riding with other riders, riding in traffic of various sorts, negotiating Ithaca’s one and only round-about, and so on. But how would we be able to show that he had acquired such knowledge? One way to proceed empirically might be to find more instances such as the “crash” in our driveway where he had not acquired such skills. My point is this: it is much harder empirically to show what is at stake and whether tacit knowledge has been acquired with respect to Collective Tacit Knowledge. Another personal example of bike riding can be used to make the point.

14 I learnt to ride a bike as most British kids do. Because our family did not have a car I rode everywhere most of my childhood including to and from school every day along busy roads. I used to think of myself as a fairly accomplished bicycle rider. That was

Philosophia Scientiæ, 17-3 | 2013 37

until I spent six months in the Netherlands! Collins claims that the Collective Tacit Knowledge needed for bicycle riding is different even for different cities, such as Amsterdam. I was in Maastricht. We had no car and since I had to get my daughter to school quite far away I borrowed a bicycle and rode with her on the back to and from school every day. Certainly bicycling in the Netherlands required me to learn new skills, negotiating bicycle traffic lights, knowing how to avoid young Dutch men and women on motor scooters who scooted along nonchalantly with cigarette in hand sharing the same bicycle paths as the much slower bicycles, and so on. One day my Dutch collaborator, Karin, passed me on her bike with her own daughter. Later that day over drinks she said she had to tell me that although I was managing OK on my bicycle I was not riding in “the Dutch way”. If TEK had been available then, and I had read it, maybe I would have left the conversation at that and said: “Duh, of course, that is to be expected I haven’t yet mastered the Collective Tacit Knowledge necessary.” My own reaction was that of a typical Brit. I was upset at this assault on my bicycle riding prowess and quizzed her as to my inadequacies. The confidence and speed of Dutch riders and their elan in riding is notoriously high, but I felt that I had managed some of that. As she discussed it more and more I realized that her position was “that she just knew” I wasn’t doing it the “Dutch way”. Casting aspersions on the behavior of the non- natives is a familiar trope from anthropology and it is often used in everyday matters to do with driving or riding behavior. If you drive around Boston long enough with a Boston driver you will hear talk about “out-of-town drivers”. Brits arriving in my town in America will talk about “lousy American drivers”. It seems part of common folk wisdom that there are ways of riding/driving which develop in particular national contexts or even cities—they are part of the identity of living in such cities. Such ways of riding/driving are according to Collins part of Collective Tacit Knowledge.

15 But let us invoke for a moment another part of sociology—“impression management” [Goffman 1959], the social construction of identity, and so on. Supposing the claims that Maastricht bike riders have a particular style which “damn foreigners” can never achieve are merely part of identity building, how would we prove it? Where would the borderline between identity building and Collective Tacit Knowledge be placed and how would we as sociologists study it? One could imagine collecting video footage of bike riding and doing double-blind tests; one could imagine putting such a matter to the tests of Collins’s imitation game, but Collins does not do this in this book. He merely asserts it as something we all know. Here the politics of what is at stake when particular claims to hold tacit knowledge are made are relevant [Doing 2007]. I may claim that I have the Collective Tacit Knowledge to pass as a Maastricht rider and Karin may deny it. Who is right and how do we tell, especially given that we both have an “interest” in being right?

4 Judging skill in practice

16 So how do we actually know in cases of skill acquisition that someone has acquired the requisite tacit knowledge? Here is another obvious point—many skills require that judgment of the neophyte be made by someone who possesses the skill already and hence is capable of recognizing that the skill has been acquired. This is how many somatic skills are assessed. For example, one of the most difficult skills veterinary surgeons need to acquire is how to use the scalpel to open the skin of the animal on the

Philosophia Scientiæ, 17-3 | 2013 38

surgeon’s table. Too much pressure could damage delicate internal organs, not enough pressure could lead to several cuts each of greater depth risking infection, more trauma and so on [Pinch, Collins & Carbone 1996]. In this case skilled surgeons are used to assess and guide the progress of the neophyte. Most such skills are learnt gradually and credentialed in some way—but for many ordinary tasks where Collins claims Collective Tacit Knowledge is in operation, credentials are not applicable. Karin has no credentialed expertise that she is a proper Dutch bike rider. Perhaps she is an outlier— an over-meticulous Dutch rider whom other Dutch riders would never take seriously?

17 But then assessing skill is in many cases not straightforward because there are often multiple audiences for the skill who possess different levels of skill themselves. Take piano playing for instance. My daughter will play beautifully to my ears and I will be as fully moved as hearing any possible piano player, but then I am not a piano player and she will report back to me that her teacher is not satisfied with her playing. This is something performing musicians face all the time. I play in two musical groups both of whom perform live. I cannot tell you the number of times after a performance when we in the band say to each other “Well we really sucked tonight” only to receive extravagant plaudits from the audience. The truth of the matter is that it takes a lower level of skill to convince most audiences in clubs (and here the genre of music, location, and quality of the sound system, are also important as well as the skill set expected of the audience—playing in Berlin with the Berlin Philharmonic Orchestra before audience members reading the score is very different in terms of the skill set of the audience). The point is this, if we take skill as something which is staged and performed with possibly an audience present then whether the skill has been accomplished or not becomes a much less straightforward question. In many of Collins’s examples there is no audience or issue of people judging a skill—it seems that the skill is either carried out or not carried out. In short by removing the context or any possibility of disagreement this encourages an over-realist notion of skill.

18 I will use yet another example to show how this could change the analysis. Take again piano playing. The way kids learn to play piano in England and the US is that they have specialized teachers and their skills are assessed periodically by a system of tests which enables skilled practitioners to delineate the exact “level” your child is at. In New York State (where I live) your child has to play every year for an outside judge (at an event misnamed “festival”) who will grade the performance in terms of different musical criteria such as “interpretation”, “technique”, and “dynamics”. This then is a fine- grained skill assessment system based upon skilled practitioners doing the judging. But why should we accept this at face value? There are many ways to deconstruct such results—often parents don’t get to do so but I would like to relate an incident where the deconstructive work was done for me. My daughter was having her last such exam. Usually she obtains the grade of “superior” and three superiors in this system equals a gold cup. She had already won one gold cup and this was her last chance to win a second one—she just needed one more superior grade. She would graduate high school that year and would no longer be eligible. She was naturally nervous; she played her pieces—a rather tricky Rachmaninov piece as I recall—and eventually came out crestfallen to learn she had obtained only an “excellent” grade (one grade below superior). No gold cup, alas! But later that evening her piano teacher called to say that the examiner had thought she was of “superior” quality, but felt that she was just holding back for some reason during this one performance. The examiner said she had

Philosophia Scientiæ, 17-3 | 2013 39

wished she had known it was my daughter’s last chance because if she had known that she would have given her a “superior”. In short what I am pointing to is that there is a whole sociology of judging skill—judging skill itself is a skill—in Collins’s terms it involves Collective Tacit Knowledge: as to what stage the pupil is at, when to be harsh or generous and so on. The assessment of skill, the system of grading skills, and how this works for multiple audiences seems to be missing from Collins’s account.

5 Constructivist and realist tensions in Collins’s work

19 Why does all this matter? It matters crucially because of Collins’s other important early contribution to the sociology of scientific knowledge, his delineating of what has become known as the “experimenter’s regress” [Collins 1992]. Unlike the TEA laser case, there are a set of cases of experiments in science where the correct outcome of an experiment is ambiguous. These are the sorts of cases dealt with in the Golem series of books [Collins & Pinch 1994], [Collins & Pinch 2006]. For instance, claims to find cold fusion, large fluxes of gravity waves, or Vitamin C as a cure for cancer are all cases where experimental outcomes are highly contested. In such cases for a while it is not clear who are the skilled experimenters—the ones claiming the phenomenon of say cold fusion, or their detractors who find no evidence of cold fusion. There is no independent way of measuring the skills or competence of experimenters in such cases. There is no equivalent to, say, vaporizing the block of concrete as in the TEA laser case. Who has the requisite tacit knowledge in the cold fusion case? Is it Pons and Fleishmann who claim to manifest the phenomenon of cold fusion or the skeptics at MIT whose experiments fail to show cold fusion? In short there seems in these sorts of cases no way to “grade” tacit knowledge (for a similar point with a more meticulous working through of examples, see [Soler 2011]). Which of course is the very point of the notion of an experimenter’s regress. How does this constructivist side of Collins’s work fit with his realist account of tacit knowledge (realist in the sense that the individual or the community either have such knowledge or do not)?1 The sociologist of science trained in the sociology of scientific knowledge might want to say that the outcome of the cold fusion controversy has been constructed in such a way that we now know Pons and Fleischmann did not have the genuine tacit knowledge to produce cold fusion. In short the consensus moved against them. But before the consensus has been reached how could we tell? My answer would have to involve a form of methodological relativism whereby I would say that for one community Pons and Fleischmann did have the requisite tacit knowledge but that for another community they did not. This immediately makes tacit knowledge much less realist and something that is also socially constructed with the phenomenon. We do not know who has the requisite tacit knowledge until the controversy is settled. Indeed with a nod to Collins we can frame this as the TACIT KNOWLEDGE REGRESS. Q: How do I know in a case at the research frontiers of contested knowledge who has the requisite tacit knowledge? A: It depends on whose experiments are deemed to be the correct ones. Q: How do I know whose experiments are the correct ones? A: It all depends on who has the requisite tacit knowledge. Q: How do I know who has the requisite tacit knowledge? And so on...

Philosophia Scientiæ, 17-3 | 2013 40

20 Even if Collins does not want to buy into the tacit knowledge regress it would seem that his work could be enriched by considering more how tacit knowledge is performed, the audience for such performances, and how institutions are set up (such as schools which grade and credential training in musicianship) and operate to assess and grade skilled performances. It would provide at the least a fuller picture of tacit knowledge.

21 My argument in this paper is that with the publication of TEK Collins has provided us with the most complete account of tacit knowledge to date. His arguments and examples concerning Relational Tacit Knowledge seem thoroughly convincing. I have not here entered into the debate over the importance or not of Somatic Tacit Knowledge. Collins plays that aspect down in favor of what he calls Collective Tacit Knowledge. Although Collins maintains that he has provided plenty of examples of Collective Tacit Knowledge elsewhere, in TEK we are left mainly with anecdotes from Collins own experience of social life to drive home the point. I have offered some counter-examples from my own experience to at least interrogate the veracity of such examples and to draw attention to what might be at stake in the sorts of examples Collins provides. By pushing on the realist versus constructivist nature of Collins’s work I am pointing to a fundamental paradox of which Collins of course is well aware. Collins likes to do what he calls “compartmentalization” whereby he can avoid arguments he generates in one area being applied to what he deems as a separate compartment of his endeavor. For example, he claims he is a realist about the social and that this allows him to say without contradiction that he has shown in a realist mode how experimental results are constructed. He thus dodges or side-steps the issue of whether his own realist claims are constructed or not. The same tension appears in his work on tacit knowledge. During moments of experimenters regress tacit knowledge is constructed but at other moments it is “really” there for participants to possess. I suggest that rather than relying on strategies such as compartmentalization which make the tensions go away, we should embrace an idiom of performativity which turns the questions of how tacit knowledge is staged and performed for different audiences, and how such assessments of tacit knowledge are institutionalized (such as at music schools), into an empirical endeavor which might give us an even more complete understanding of the deep mystery of tacit knowledge.

BIBLIOGRAPHY

COLLINS, HARRY 1974 The TEA set: Tacit knowledge and scientific networks, Science Studies, 4, 165–186. 1990 Artificial Experts: Social Knowledge and Intelligent Machines, Cambridge, MA: MIT Press. 1992 Changing Order: Replication and Induction in Scientific Practice, Chicago: University of Chicago Press, 2nd ed., 1st edition Sage publications, 1985. 2001 Tacit knowledge, trust, and the Q of Sapphire, Social Studies of Science, 31, 71–85. 2004 Gravity’s Shadow: The Search for Gravitational Waves, Chicago: University of Chicago Press. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

Philosophia Scientiæ, 17-3 | 2013 41

COLLINS, HARRY & EVANS, ROBERT 2007 Rethinking Expertise, Chicago: University of Chicago Press.

COLLINS, HARRY & HARRISON, ROBERT 1975 Building a TEA laser: The caprices of communication, Social Studies of Science, 5, 441–450.

COLLINS, HARRY & KUSCH, MARTIN 1998 The Shape of Actions: What Humans and Machines Can Do, Cambridge, MA: MIT Press.

COLLINS, HARRY & PINCH, TREVOR 1994 The Golem: What You Should Know About Science, Cambridge: Canto. 1998 The Golem at Large: What You Should Know About Technology, Cambridge: Canto. 2006 Dr Golem: How to Think About Medicine, Chicago: the University of Chicago Press.

DOING, PARK 2007 Give me a laboratory and I will raise a discipline: The past, present, and future politics of laboratory studies, in The Handbook of Science and Technology Studies, edited by HACKETT, E., AMSTERDAMSKA, O. & LYNCH, M., Cambridge: MIT Press, 279–295.

GOFFMAN, ERVING 1959 The Presentation of Self in Everyday Life, New York: Doubleday.

PINCH, TREVOR, COLLINS, HARRY & CARBONE, LAWRENCE 1996 Inside knowledge: Second order measures of skill, Sociological Review, 44(2), 163–186.

SOLER, LÉNA 2011 Tacit aspects of experimental practices: Analytical tools and epistemological consequences, European Journal of Philosophy of Science, 1, 393–433.

NOTES

1. This tension between realism and constructivism seems different to earlier debates about reflexivity and Collins’s claims that he is realist about the social but constructivist about the scientists’ claims. In this case it would seem that claims made by the participants to have particular sorts of tacit knowledge are themselves subject to interpretative flexibility.

ABSTRACTS

In this paper I examine Harry Collins’s influential writing on tacit knowledge. In particular I turn my attention to his recent book, Tacit and Explicit Knowledge [Collins 2010], or TEK, which is arguably the most complete and systematic statement of what he means by the term “tacit knowledge”. As well as examining tacit knowledge as elaborated in this contribution, I draw out an underlying tension in Collins’s major contributions to the sociology of scientific knowledge in general between the realism underlying his notion of “tacit knowledge” and the constructivism underlying his other well-known concept, “the experimenters’ regress” (as for instance, elaborated in his well-known book Changing Order [Collins 1992]). In order to make this argument I pay particular attention to an aspect of his writings on tacit knowledge which I think is worthy

Philosophia Scientiæ, 17-3 | 2013 42

of closer examination: namely the sorts of empirical support claimed for the features and properties of tacit knowledge to which he attends. In short I ask questions concerning some of the specific empirical examples and the conclusions he draws from them.

Dans cet article, j’examine les écrits influents de Harry Collins consacrés à la connaissance tacite. Je me penche en particulier sur son récent livre, Tacit and Explicit Knowledge [Collins 2010] ou TEK, qui est sans doute l’exposé le plus complet et le plus systématique de la manière dont Collins conçoit la connaissance tacite. Tout en examinant la connaissance tacite telle qu’elle est développée dans cette contribution, je dégage, au sein des contributions majeures de Collins à la sociologie de la connaissance scientifique en général, une tension sous-jacente, entre d’un côté le réalisme qui sous-tend sa notion de « connaissance tacite », et, de l’autre, le constructivisme qui sous-tend son concept célèbre de « régression de l’expérimentateur » (tel qu’élaboré par exemple dans son fameux livre Changing Order [Collins 1992]). Pour construire cet argument, j’accorde une attention particulière à un aspect des écrits de Collins sur la connaissance tacite qui, je pense, mérite un examen plus approfondi : à savoir les types de support empirique qui sont invoqués en faveur des caractéristiques et des propriétés de la connaissance tacite visée. En bref, je pose des questions à propos de certains des exemples empiriques spécifiques invoqués et des conclusions qui en sont tirées.

AUTHOR

TREVOR PINCH Department of Science and Technology Studies, Cornell University (USA)

Philosophia Scientiæ, 17-3 | 2013 43

At the Margins of Tacit Knowledge

Michael Lynch

1 Ever since Michael Polanyi wrote on the subject more than a half-century ago [Polanyi 1958], tacit knowledge has proved to be one of the most resilient themes in history, philosophy, and social studies of science. It also has an enduring place in psychology and cognitive science, and in business and management studies, among other fields. Although the most original and provocative discussions of tacit knowledge were published decades ago, interest in the topic has remained strong and has intensified recently with the publication of a new edition of Polanyi’s The Tacit Dimension [Polanyi 2009], Harry Collins’s Tacit and Explicit Knowledge [Collins 2010], hereafter [TEK], and Mary Jo Nye’s biography of Polanyi [Nye 2011].

2 No doubt, there are many reasons for the persistent interest in tacit knowledge. The most obvious reason is that the theme resonates with the discourse and interests of practitioners as well as scholars (and of scholars as academic practitioners). Scientists and medical practitioners consistently testify to the indispensible role in their practices of tacit knowledge, craft knowledge, and variations on those themes, such as “golden hands”, Fingerspitzen, bedside manner, and other informal skills gained through on-the-job training rather than formal education. Scholars in many fields have embraced the theme to challenge philosophical and historical conceptions of science that use textbook accounts of method and autobiographical reflections by prominent scientists as their sole means of access to scientific practice.

3 Attempts to come to terms with tacit knowledge often contrast it with what it is not: it is not explicit; it cannot be reduced to rules or formulae; it cannot be conveyed through written or verbal instructions; and computer programs cannot replicate it. Nevertheless, as Stephen Turner points out, assertions about tacit knowledge often resort to analogies with explicit rules, perceptions, and interpretations [Turner 2012]. Polanyi uses such an analogy when he defines a “skilful performance” as “the observance of a set of rules which are not known as such to the person following them” [Polanyi 1958, 50]. This analogy with rules appears to be shot through with paradox: the “rules” are not written or recited; they are not even “known” by the person who skillfully follows them. To transpose Donald Rumsfeld’s category of “known unknowns” [Rumsfeld 2005], these tacit rules are “unknown knowns”. There are, of course, ways

Philosophia Scientiæ, 17-3 | 2013 44

around the paradox. Polanyi’s way was to drive a logical wedge between knowing how to do something and being able to verbalize how one does it. Depending on one’s disciplinary preference, Polanyi’s tacit rules can be assigned to the rules of a language, to cognitive organizations and neurological processes, or to a social ideology or “discourse” that is unrecognized or misrecognized by the docile subjects who act in accord with it [Turner 1994].

4 Because tacit knowledge is defined as being beyond the reach of rules and explicit formulations, many analysts (including some laboratory practitioners) associate tacit knowledge with non-rational domains of luck, magic, superstition, and mysticism [Cambrosio & Keating 1988], [Jordan & Lynch 1992], [Lynch 1985]. However, while he insists that the tacit dimension is essentially “inexpressible”, Polanyi places that dimension squarely in the real world. Indeed, it is so down-to-earth that rules and generalizations are too abstract to be able to grasp it. He also endows tacit knowledge with political significance, as it is a reality that shatters the dreams of ideologues and the plans of social engineers. And yet, this reality eludes all efforts to come to terms with it. At best, it can be mentioned, indicated, or alluded to. For Polanyi tacit knowledge is not and cannot ever be made explicit. It is not enough to say that tacit knowledge is unstated but capable of being put into words, should the need arise and the relevant information and communication technology be available—it simply cannot be conveyed in words to someone who does not already “dwell” in the tacit dimension.

5 Unlike Polanyi, Collins portrays the relationship between tacit and explicit knowledge as more of a continuum from readily explicable to ever-more-recondite forms of knowledge [TEK]. His typology is not simply an array of categories of explicit and tacit knowledge, as the types of tacit knowledge are ordered in terms of “strength” along a core-periphery axis. Collins diagrams “the terrain of tacit knowledge” as a concentric series of circles and squares [TEK, 158], with the innermost circle representing the “strongest” type of tacit knowledge, and the outermost (most marginal, weakest) type representing knowledge that is tacit only as “a matter of how particular people relate to each other” [TEK, 86]. Such “relational tacit knowledge” lies just within the margin between tacit and explicit knowledge in Collins’s typology. This “relational” category includes various types of knowledge that can be formalized and transmitted by someone who possesses the relevant communication technology and technical capacity. For example, secrets—knowledge withheld from some persons or groups but not from others—can be made explicit, in principle, though there may be very powerful reasons not to reveal them in some relational circumstances. Collins makes clear that he finds “relational tacit knowledge” less interesting, philosophically, than his other two major categories: “somatic tacit knowledge” (physically embodied skill) and “collective tacit knowledge” (knowledge embedded in forms of social life). “Collective tacit knowledge” is his core category [TEK, 158]—“the irreducible heartland of the concept” [TEK, 119]—whereas “relational”, and to a lesser extent “somatic”, tacit knowledge is, in principle if not in practice, reducible toexplicit knowledge.

6 In this paper, I question Collins’s equation of “relational” with “weak” and uninteresting tacit knowledge. My questioning extends to the very coherence and reality that Collins attributes to the overall “terrain” of tacit knowledge. However, I do not mean to suggest that tacit knowledge is unreal. To paraphrase Ian Hacking’s question about social constructionism [Hacking 1999], my question is “Real in relation to what?” My argument is that the very notion of tacit knowledge is inseparable from the

Philosophia Scientiæ, 17-3 | 2013 45

contingent relations in which it is bound. Treating it as a concept—or as part of a “vocabulary” to borrow C. Wright Mills’ classic account of motives—differs from treating it as a substantive terrain or a pervasive force [Mills 1940]. As I mentioned earlier, tacit knowledge often is defined in relation to what it is not: it is not formal (or even formalizable); it is not explicit; it is not reducible to rules, algorithms, written instructions and protocols. Consequently, the “what” of tacit knowledge is located in relation to “what it is not”. This does not mean that it is unreal, nor does it mean that it is a kind of epistemic anti-matter or dark matter. Instead, I mean to suggest that the reality of tacit knowledge is bound up in uses of language; and, specifically, various conceptions of explicit knowledge that define what it is not. Far from being uninteresting, its marginal relations to conceptions of explicit knowledge are crucial for understanding the meaning (the uses) of tacit knowledge—such relations constitute the relevance of tacit knowledge. Among the various uses of the concept of tacit knowledge are polemical uses that serve to enhance and defend the autonomy of academic and non-academic professions. Consequently, my aim in the remainder of this paper is to examine what is done with the notion of “tacit knowledge” rather than to define and classify what it is as a substantive domain. I will leave it to readers to decide if my treatment simply differs from Collins’s or whether it reveals a basic misconception of the phenomenon on his (or my) part.

7 In my view, a treatment of tacit knowledge as relational and residual may lead us to question the very unity and coherence of the “tacit dimension” that Polanyi postulates, or the “terrain” on which Collins proposes to “map” tacit knowledge. Moreover, such a treatment can encourage us to focus instead on particular academic and vernacular uses of the concept as well as other, arguably related, terms. This relational turn is a shift away from philosophical debate about what can or cannot be made explicit, in principle or in practice, and toward studies of what counts—and, equally important, what does not count—as tacit knowledge in specific historical and contemporary circumstances. Such research examines explicit claims about tacit knowledge, and can lead us to explore what such usage hides as well as what it reveals.

8 When shifting focus from a notion of “terrain”, which suggests a coherent and substantive ground, to an examination of situated usage, my aim is not to reduce tacit knowledge to mere words or attributions. I mention this in light the way Collins, in a book co-authored with Robert Evans, contrasts “relational theories” with a “substantive” theory of expertise [Collins & Evans 2007, 2]. Collins and Evans elaborate upon this contrast by saying that acquiring expertise involves an arduous social process of learning the knowledge (including, especially the tacit knowledge) possessed by an “expert group”. Such expertise is possessed—as skill—and it can be lost through disuse. Contrary to “relational theories”, which Collins and Evans argue would treat expertise strictly as a contingent “attribution” made by others, their theory treats expertise as something that “individuals may or may not possess [...] independently of whether others think they possess expertise” [Collins & Evans 2007, 3]. Their dichotomous separation of linguistic “attribution” from substantive possession of expertise ignores complex, empirically evident practices and institutions that define, certify, and test what counts as expertise in particular professions, games, and other activities performed in specific historical and cultural circumstances. Importantly, when they theorize about the topic and attempt to demarcate “real” expertise from non- or pseudo-expertise, Collins and Evans participate (if only vicariously) in the social practice of constituting what counts as expertise. The problem with their theorizing is

Philosophia Scientiæ, 17-3 | 2013 46

that it is abstracted from the political and professional circumstances in which claims and counterclaims about expertise are aired, as well as from the activity-specific procedures and forums that demonstrate and test expertise. The concept of “attribution” does not do justice to the substantive, and yet highly variable and situated, organization of such practical tests and professional institutions. In the remainder of this essay, I will make a similar argument about Collins’s treatment of tacit knowledge. In contrast to Collins, I shall argue that tacit knowledge is not simply a real domain or dimension; instead, I shall argue that tacit knowledge is not a unified “terrain”, but that it takes distinct forms depending upon local claims and counterclaims about what it is and is not. Finally, I shall suggest that Collins’s theory is limited by the way he locates tacit knowledge on an abstract “terrain”, rather than (as some of his own early work exemplifies) in relation to discursive situations, historical labor conflicts, and efforts to secure and protect professional autonomy.

1 How to locate tacit knowledge

9 Theoretical and philosophical writings are by no means the only discursive means for locating (defining, identifying, exhibiting) tacit knowledge. Numerous practical situations and conundrums are a fertile source of reflection on how practical and discursive relations are central to the very idea of tacit knowledge. Although Polanyi postulates an abstract “tacit dimension”, his writings on the subject locate tacit knowledge by drawing contrasts with explicit rules, instructions, and plans. And, while Collins also writes abstractly of a “terrain” on which tacit knowledge can be mapped [TEK], in his early work on the subject [Collins 1974, 1975], he locates tacit knowledge by reference to the practical difficulties and polemical contentions attending efforts to replicate experiments and experimental equipment. In his work on “artificial experts” [Collins 1990] he locates tacit knowledge by reference to the practical difficulties with designing artificial intelligence (AI) programs and expert systems that proponents claim can (or someday will) reproduce embodied and communicative activities.

10 Though Collins has rarely acknowledged any indebtedness to ethnomethodology (the study of practical actions and practical reasoning in ordinary as well as professional circumstances), his procedures for locating tacit knowledge were not unlike those that Harold Garfinkel had earlier used for locating what Garfinkel called ad hoc practices in many fields of action within and beyond the sciences [Garfinkel 1967]. Difficulties with developing early machine translation programs, also were a source of insight into the “indexicality” (the contextual sensitivity and variability of linguistic intelligibility [Bar- Hillel 1954]) that became central to Garfinkel’s ethnomethodology [Garfinkel 1967]. Garfinkel later used the term “instructed actions” to cover relations between explicit instructions and the situated actions of using them [Garfinkel 2002]. Instructed actions are not simply a kind of tacit knowledge; they involve a pairing of explicit directions, maps, and other verbal and written formulations with the practices of using them on particular occasions. Tacit knowledge can be made relevant in connection with that pairing; consequently, the relational organizations of instructions to their situated uses provide means (certainly not the only means) for locating tacit knowledge (i.e., making it practically relevant). Garfinkel frequently used exercises for his students that required them to describe their efforts to enact a particular protocol, assemble an object from plans, or follow a set of instructions or directions. His aim was not to train

Philosophia Scientiæ, 17-3 | 2013 47

the students to become adept at following instructions; instead, it was to attune them to the situated difficulties and contingencies that arise when they attempt to enact plans, act in accord with rules, or follow directions. And, while he required his students to describe those difficulties and contingencies in fine detail, the aim of the exercise was not to develop complete descriptions of tacit knowledge. Instead, the aim was to demonstrate the impossibility of ever developing complete descriptions in an absolute sense of completeness, and to encourage the realization that practical standards of completeness are themselves bound up with relational contingencies. Lucy Suchman’s Plans and Situated Actions is an innovative and lucid elaboration of Garfinkel’s procedure for investigating how even the most elaborate plans and instructions for using novel technologies require users to deploy ad hoc procedures for getting tasks done [Suchman 2007]. Suchman uses insight from her investigations to critique the determinative role ascribed to explicit plans in AI and related Silicon Valley belief systems.

11 According to these procedures, tacit knowledge can be located prospectively or retrospectively, even though it can never be described exhaustively. Prospectively, rules, recipes, formal instructions, plans, directions, maps, or protocols implicate unspecified knowledge that enables one to work out just how to follow them in a future course of action. Retrospectively, tacit knowledge is revealed through the deficiencies of written records for gaining access to actual work practices. In many cases, tacit knowledge is implicated by a retrospective-prospective effort to find what explicit instructions are “saying” in light of interpretative and practical problems and temporal relations that emerge over the course of efforts to follow them. While, following Collins, we may simply be inclined to place instructed actions in the category of “relational” tacit knowledge, to do so is to miss the point of Garfinkel’s exercises: the situational, interactional, and temporal relations that arise in the course of the practices of instructed action constitute the very relevance of tacit knowledge [TEK]. As Garfinkel’s students discovered when performing an exercise to follow directions for getting from a starting point to a destination in a city, the difficulties and contingencies they encountered were particular and mundane: they did not enter a “dimension” of tacit knowledge, instead they ran into a stream of difficulties during their ongoing attempts to reconcile the sketchy details of their directions with the unfamiliar and overwhelmingly dense “lived” course of the journey. The fact that the directions, when read in isolation from the journey, did not tell the students just what they needed to know in order to follow them successfully, might seem to implicate a qualitatively different kind of knowledge that enabled such success. But, when investigated in detail, the knowledge in question appeared to be commonplace, mundane, and tied to familiar situations and routines. This knowledge did not make up a coherent cognitive dimension or terrain; instead, it was an assemblage of ecologically embedded judgments, actions, and improvisations that were performed along the course of the unfolding journey. Garfinkel preferred the heading “instructed actions” to cover these various practical maneuvers and reactions. He did not contrast a tacit dimension to the explicit knowledge embodied in rules, written instructions, recipes, and other formal devices. Instead, he pointed to the way such formal devices were constituents of instructed actions; they were not causes of such actions, but instead were tools used in the course of them.

12 In all of these cases, instructed actions are located in the relationship between an explicit rule, directive, instruction, formula, or report, and a particular course of actions that makes use of a rule prescriptively as a guideline and/or retrospectively as

Philosophia Scientiæ, 17-3 | 2013 48

a source of evidence, or as a basis for criticizing or justifying the actions taken. This relationship is not limited to an analogy between, for example, explicit rules and “rules” that are tacitly followed. In such cases, rules, formulated instructions, and plans are not analogies for another kind of knowledge; instead, they feature concretely in practices, where they are tools rather than exhaustive representations of the actions in which they are used. In such cases, what is not formulated in rules, plans, and so forth, often can be described in detail, but the details are highly particular—tied to singular situations, local histories, ongoing relationships, and personal idiosyncrasies. An analyst can generalize about these particulars, but they are “tacit” by virtue of their particularity in reference to generalized accounts and instructions.

2 A continuum or Cartesian divide?

13 In his investigations of tacit knowledge, dating back to the mid-1970s, Collins, like Polanyi before him, deploys a contrast with various forms of “explicit” knowledge to define what tacit knowledge is not. In his recent book on the subject, he goes much further than Polanyi to develop an entire cosmology in which “strings” (including but not limited to the sequence of code in a digital program) provide the material and causal basis for transmitting “explicit” knowledge and programs of action. His argument locates tacit knowledge through a process of elimination: if a practice or domain of knowledge can be encoded and transmitted by a string of words, signs, or electronic pulses, then it is not tacit. He also develops a taxonomy of various orders of tacit knowledge and reserves the most central place in his scheme for “collective” tacit knowledge.

14 Unlike Hubert Dreyfus, for whom the concept of tacit knowledge is among the various philosophical resources that he uses to make the case for what computer programs “can’t do” [Dreyfus 1979, 1992], Collins in his recent work uses what computer programs (or, more generally, abstract strings of “instructions”) cannot possibly encode and transmit as a way to define tacit knowledge. This analytical strategy differs from the procedure described earlier for locating tacit knowledge by describing the contingencies that arise during efforts to follow instructions and rules (a methodological strategy that Collins himself used in his early work). Instead of providing detailed historical or ethnographic descriptions of actions that necessarily go beyond what available rules and instructions specify, Collins now deploys a strategy for considering and classifying a series of candidate types of tacit knowledge. In this scheme, when a candidate type of knowledge or action can be made explicit, it no longer counts as deeply and irrevocably “tacit”—it is tacit only as a matter of happenstance. His categories of tacit knowledge are thus ordered according to their “strength” at resisting eventual re-allocation to the “explicit” side of the ledger. The general category of somatic skill is more difficult to explicate, and the core category of socially founded knowledge defies the very possibility of explication in a foreseeable future in his version of tacit knowledge.

15 Collins faces a formidable challenge in his attempt to specify types of knowledge that are beyond explication in any foreseeable future. He recognizes that there is no possibility of drawing a bright line between tacit and explicit knowledge, but he nevertheless insists upon a “Cartesian” distinction, however fuzzy at the margins, that demarcates core tacit knowledge from whatever can be reduced to the effects of digital

Philosophia Scientiæ, 17-3 | 2013 49

codes, genetic sequences, and other material/semiotic “strings”. Although Collins has a running debate with Dreyfus, like Dreyfus he tries to indicate what cannot possibly (or, at least, foreseeably) be expressed in human or machine languages, regardless of whether the relevant expressions take the form of written or oral instructions conveyed from expert to novice, or algorithms designed to “instruct” a computer to play chess, recognize a face, or convert speech to writing. Both Collins and Dreyfus recognize that computer programs are now able to model, perform, or simulate far more than what was possible decades ago when Dreyfus initiated his critique of AI’s ambitions. However, as Dreyfus argued with a memorable line, to claim that incremental advances in computer programs support the idea that there is no essential limit to the potential expansion of programmed systems into previously unfathomed domains of human conduct, is like saying that “the first man to climb a tree could claim tangible progress toward reaching the moon” [Dreyfus 1992, 100]. However, as early critics of Dreyfus were keen to point out [Pappert 1968], a reciprocal difficulty attends the effort to define a boundary between what can or cannot ever be explicated through instructions or programs.

16 The debate about what machines can or cannot do is complicated by the fact that the computers, calculators, printers, and typewriters that now bear names that once were used for human job titles neither simulate nor duplicate the actions of the humans they put out of work, any more than an automobile duplicates or simulates the actions and capacities of a horse. By and large, efforts to develop novel information and communication technologies seek novelty (just enough to warrant a favorable decision by a patent examiner), without concern for simulating or duplicating how humans think or act [Voskuhl 2004]. Such technological innovations have little bearing on debates about a possible qualitative limit to what can ever be explicated in a foreseeable future. From the pragmatic point of view of many engineers, the question of whether their innovations will ever run up against an essential barrier to explication is a “philosophical” question that can be deferred for as long as there is the immediate promise of incremental success in the effort to design technologies that do jobs differently from the way humans do(or once did) such jobs.

17 Changes in communicative technologies (and not just computer programs) reconfigure instructed actions and challenge any effort to draw stable distinctions between explicit and tacit knowledge. Consider the example of recipes. As noted earlier, one way to locate tacit knowledge is through a contrast between explicit instructions and immediate efforts to follow those instructions. Sometimes, though not always, a written text is treated as the prototype of explicit knowledge. In the case of a written recipe, there are well-known problems. The recipe calls for one type of cooking oil, and we only have another type. It is vague in its description of many practical details, and it fails even to mention many practical requirements, perhaps because the recipe writer assumed that some practices should be elementary for virtually anyone who would set foot in a kitchen. By closely examining a novice’s difficulties, we are bound to find that many situations occur that the recipe fails to mention or anticipate, and that a mastery of the recipe can only be gained through repeated practice and hands-on instruction.

18 But now consider the difference between a cookbook and an online video or televised cooking show. The video can demonstrate “tricks” that are glossed over in written recipes, such as how to hold or manipulate a particular utensil, what “stir vigorously” might look like, and so forth. The voice-over instructions can be closely coordinated

Philosophia Scientiæ, 17-3 | 2013 50

with the visible demonstration of the sequence of actions. Much can be shown as well as said. An effort to enact what the video instructs may present its own problems, and moving quickly from one online version to another of the “same” recipe can be a source of confusion. However, what turns out to be “explicit” and “tacit” varies with the media through which the instructions are conveyed. When a video is used to supplement a written recipe, it seems that the explicit materials have now incorporated at least part of what the written recipe failed to specify.

19 Now consider an interactive forum, such as an online exchange or hotline [Lynch 2002, 208ff.]. These forums have the advantage of allowing users to raise questions geared to singular problems that arise in the course of an attempt to follow particular kinds of recipe. Again, while such forums present their own limitations, they also occupy an intermediate position between a written recipe and a hands-on tutorial. Such communication technologies mediate the user’s relation to the portrayed practice in a different way than a written recipe. Each case differently configures just what is explicit and just what is unstated. What is made explicit, and how it is made explicit, reciprocally defines, and is defined by, what is presumed to be tacit. Above all, a consideration of such cases allows us to see that the very meaning of “explicit” (and, by implication, “tacit”) is not at all clear.

20 Collins proposes a “Cartesian” discontinuity between explicit (programmed and programmable) knowledge and a core domain of collective (human) tacit knowledge. Proponents of “Strong AI” such as Paul and Patricia Churchland propose the opposite [Churchland & Churchland 1990]: that there is no essential divide between the two domains. However, there is no dispute on either side over the fact that there is continuity at the margins, and that new communication and information technologies can convert or translate previously tacit knowledge into novel modes of explicit instruction. For the most part, the AI debate is about the ultimate possibility of making tacit knowledge explicit, and marginal cases are of interest mainly as bases for extrapolating what the current state of the art might or might not promise for the future. For different reasons, I shall stay at the margins to examine what Wittgenstein called “intermediate cases”—cases at the margins of Cartesian distinctions [Wittgenstein 1953 § 122]—while remaining non-committal about any essential differences between tacit and explicit knowledge.

21 The AI debate imposes an artificial condition upon discussions of tacit knowledge—the condition being the possibility of total explication of a human practice, so that a machine can do the practice, as though on its own and with no residue of tacit knowledge needed to complete what the machine leaves undone. The debate narrows the focus of those, like Collins and Dreyfus, who rise to the challenge of specifying that and how tacit knowledge cannot ever be articulated in the way a program would require. Their charge is indifferent to knowledge that is “tacit” in the sense of being unspoken, implicit, not openly expressed, but capable of enunciation. There can be many reasons for why such knowledge is not enunciated besides an essential inability to formulate it [Cambrosio & Keating 1988, 246]. One, apparently clear, example of such a marginal case is secret knowledge.

Philosophia Scientiæ, 17-3 | 2013 51

3 Secret knowledge: What tacit knowledge is almost not

22 As noted earlier, Collins highlights the role of “collective” tacit knowledge as his core ideal type [TEK], while assigning weaker or lesser status to other categories of “relational” tacit knowledge. One of the categories he consigns to the margin is “secret” knowledge. This category includes knowledge that can be made explicit, but which is deliberately withheld or concealed [TEK, 91]. Such knowledge is relational, because the extent to which it is explicit or concealed is selective, often strategic, and related to circumstances. Collins adds that there is no “deep” philosophical puzzle about such knowledge. In the remainder of this essay, I will argue that the category of (possibly) concealed knowledge is “deeper” and more interesting than Collins lets on. Moreover, ambiguities about what is truly tacit (or essentially incommunicable) and about what may or may not be deliberately withheld or disguised, can be of strong interest for historical and sociological investigations, as well as for philosophical analyses and political conflicts.

23 In principle, secret, deliberately concealed, or undisclosed knowledge is potentially explicit knowledge that just happens to be withheld from some potential recipients on some occasions. A person who keeps a secret knows more than she tells, but not necessarily more than she can tell. To be more precise, we need to distinguish between two possible senses of the word “can” in this formulation (Collins distinguishes no less than eight senses of “cannot” [TEK, 84]). One sense has to do with interpersonal situations, interactional norms, and local agreements: a speaker might say, “I can’t tell you” to a particular person from whom she is keeping a secret, but this doesn’t mean that she is unable to tell it to anyone else. Another sense of “can” and “cannot” has to do with the very possibility of articulating what remains unsaid. When Polanyi says that “we can know more that we can tell” [Polanyi 2009, 4], he makes clear that what we cannot tell is incapable of being articulated to anyone, and it remains beyond the reach even of the solicitations of a skilled psychoanalyst, interrogator, or torturer. Such knowledge is not simply tacit—in the sense of being understood but withheld—it is incommunicable.

24 The distinction between secret and incommunicable knowledge becomes more complicated when we recall that efforts to conceal knowledge frequently involve strategies for concealing that such knowledge is being concealed. For example, pretexts and alibis used by intelligence agencies when they conduct covert operations not only enable “plausible deniability”, they forestall suspicion by deflecting attention away from the fact of concealment. In contrast to Polanyi’s line about “the observance of a set of rules which are not known as such to the person following them”, covert agents conceal a set of practices from being evident, as such, to an outside observer. Even when an outside investigator is convinced that particular agents are concealing or disguising what they easily could articulate if they chose to do so, it can be dauntingly difficult to expose their secrets or even that they are withholding secrets [Lynch & Bogen 1996]. Even in less politically charged circumstances, a group of agents may present a public front that systematically conceals their own understandings of what they are doing, on the assumption that outsiders would be prone to misunderstand and disrupt the actual practices if they were revealed to them [Hilgartner 1990]. Even during an earnest attempt to instruct a novice, an instructor may withhold understandings of the

Philosophia Scientiæ, 17-3 | 2013 52

practice that the novice cannot possibly “absorb” without getting confused and discouraged. An expression, such as “you don’t need to know” is exquisitely ambiguous —it may hint at a dark secret, or simply mark that a more complex or subtle understanding is being deferred until the novice has sufficient competence to grasp it. In such contexts, ambiguities between what Collins classifies as relational and non- relational categories of tacit knowledge become strategically useful.

25 In situations of labor conflict, workers use (and are suspected of using) techniques for holding productive capacity in reserve in a way that is not visible to managerial or administrative authorities. Secrecy in this sense is not a matter of withholding specific statements. It is a more elaborate art of complying with formal requirements, and masking the extent to which such requirements are undermined in practice. A classic example is a phenomenon that Frederick Winslow Taylor dubbed “soldiering” among workers in a steel plant [Taylor 1911]. “Soldiering” was his term for the collective guild-like maintenance of a barrier between insiders’ and outsiders’ (and, specifically, managers’) knowledge of their work-practices (playing off of Collins’s terminology, we could say that it is an instance of “collective tacit metaknowledge”). Taylor suspected that workers under his supervision were conspiring to reduce the pace of work while maintaining the appearance of high productivity.

26 For Taylor, the tacit knowledge of manual workers offered a practical challenge to his efforts to manage labor and improve its efficiency. Initially, he faced a seamless alliance between workers’ secrecy and their possession of tacit skills, and he aimed to break the workers’ “soldiering” in order to formalize, standardize, and control the labor process. Taylor designed his program in scientific management in order to wrest control of the pace of work from the laborers, so that it could be managed and optimized by the industrial engineers [Braverman 1974, 64ff.]. His challenge was to devise a way to make explicit what workers hid from his view through their “soldiering”. Taylor did not aim to describe the workers’ practices; instead, he sought to reconfigure those practices by changing the incentives to reward individual productivity at the expense of worker solidarity. He also developed schemes for measuring and regimenting embodied motions that previously had been left to the workers’ discretion. His method was not simply an effort to duplicate, or even simulate, the workers’ practices; rather, it became a means to bypass, supplement, and rationalize their practices. Taylor’s notion of “soldiering” assumed that the workers’ practices were tacit only because they were collectively concealed, and yet they appeared to an outsider to be indistinguishable from the embodied skills and collectively cultivated craft knowledge that a non- practitioner would be unable to grasp. Taylor’s use of slow-motion film and his system of time-and-motion study produced an intervention that (in his view) revealed what the workers concealed, though in Braverman’s critical view the intervention reconfigured the labor practice and systematically destroyed the communal solidarity that supported it [Braverman 1974].

27 “Soldiering” strategically deploys the difference between an insider’s and an outsider’s perspectives on a practice. Another practice used in labor conflicts that also deploys that difference, though in an inverted way, is “working to rule”. Workers, such as air traffic controllers, have used this practice to slow down production during labor actions. It is a kind of strike, but instead of leaving the job, workers conspicuously follow all of the formal protocols and safety rules that, ordinarily, they would bypass with informal shortcuts in order to maintain the workflow. It seems likely that, even

Philosophia Scientiæ, 17-3 | 2013 53

when conducting a “work to rule” operation, the workers do not simply “follow the rules”, since they would still need to deploy their informal know-how in order to perform situated tasks. Consequently, as in cases of “soldiering”, the pace can be slowed surreptitiously while an ostentatious effort is made to adhere to formal protocols. In both scenarios, tacit knowledge has an ambiguous and contentious role. Not only do the opposing parties in such conflicts hold contrary assumptions about what is or is not concealed, as opposed to deeply embedded in the individual skills and collective form of life of the workers, the ambiguity is a substantive and strategic constituent of the “relational” situation.

4 Tacit knowledge as a rhetorical theme and professional resource

28 In addition to being an attractive theme and explanatory resource for philosophers and social scientists, tacit knowledge and other themes related to it are vernacular idioms. As is the case for so many other keywords in philosophy and social science, analytical uses of the term have an ambiguous and confusing relationship to ordinary, situated vocabularies [Mills 1940], [Winch 1958]. A prime source of confusion has to do with the way ordinary words such as “skill” and “knowledge” are terms of praise as well as analytical concepts. In many situations, skill and knowledge are valued as possessions of persons and professions, and as such they are commodities that can command a high price and provide the basis for privilege, authority, and esteem. The concept of tacit knowledge aids and abets such authority. Professions that lay claim to tacit knowledge, not only reserve it as a source of expertise, they also imply that others cannot possess or even fully comprehend it without first undergoing a lengthy apprenticeship—an apprenticeship that is controlled by the profession. As long as relevant non- professionals assume that the tacit knowledge is legitimate, they must trust what the experts say because they have no basis for evaluating it themselves. The lack of public transparency associated with expert knowledge has long been viewed as a problem for evaluating expert evidence in public forums such as jury trials [Hand 1901]. This property of tacit knowledge also presents an analytical problem for those like Collins who take a realist position toward expertise [Collins & Evans 2007]. In practical situations, the reality of tacit knowledge can be difficult to establish. This difficulty is not just a methodological challenge for a sociological study of tacit knowledge, it is a sociological phenomenon in its own right—and, moreover, it constitutes social relations as well as reflects them. Indeed, how the claimed reality of tacit knowledge plays out may be more interesting sociologically than the “fact” of whether tacit knowledge is or is not “real” in a particular case.

29 Polanyi, Collins, and others who have written on the topic of tacit knowledge express great respect for its role in the acquisition of technical authority. Polanyi, especially, treats tacit knowledge as a basis for opposition to centralized efforts to “manage” scientific production [Polanyi 2009]. Unlike Frederick Winslow Taylor, Polanyi lines up on the side of the workers (in this case, a more highly-educated and privileged class of worker than Taylor’s factory hands), and he invokes tacit knowledge in support of their autonomy and authority. Like prominent contemporaries of his, whose arguments also supported the autonomy of science [Merton 1942], [Bush 1945], Polanyi stresses that autonomy pays off, not only for the scientists who are granted authority over their

Philosophia Scientiæ, 17-3 | 2013 54

enterprise, but also for the society as a whole which benefits from the innovations that emerge from free scientific inquiry. Others have suggested that the idea that public outlays for scientific research should be combined with a hands-off policy toward how the public funds are spent is a highly convenient ideology for the profession [Mulkay 1976], [Gieryn 1999]. In that light, Polanyi’s arguments about the autonomy of science can be viewed as an expression of a professional ideology. Polanyi was a chemist when not moonlighting as a philosopher of science, and in many ways his writings express strong commitment and admiration for the craft, integrity and “connoisseurship” associated with the best scientific work. Though laudable in many respects, Polanyi’s (and increasingly, Collins’s) orientation to tacit knowledge in the sciences is what sociologists of scientific knowledge call “asymmetric” [Bloor 1976]. To understand what such a treatment implies, I will give two examples of a “symmetrical” treatment of tacit knowledge in medicine.

30 In an article on the theme of “incommunicable knowledge” in late 19th century British medicine, historian of medicine Christopher Lawrence describes efforts by Victorian gentlemen-doctors to resist the introduction of diagnostic instruments such as the sphygmomanometer and the stethoscope [Lawrence 1985]. Fearful that these items of equipment would open up the more elite ranks of the medical profession to mere specialists, the gentlemen-doctors emphasized the “incommunicable knowledge” necessary to perform the bedside work of examining the patient’s pulse, heart rate, and blood pressure. In addition to referring to the cumulative embodied knowledge that comes with long experience, the physicians mentioned the refined nerves and sensibilities of men raised in the gentle classes. To defend the autonomy of clinical medicine, these physicians invoked an epistemology of individual experience which, by definition, defied analysis. A similar distrust of applied science, an approval of craft skills and the praise of rule- of-thumb practice, can be found in the language of many other contemporary professions, such as chemistry or engineering. But, in medicine the account of the nature of clinical skill was linked to other things. It was used to show that only the gentleman, broadly educated, and soundly read in the classics, could be equipped for the practice of medicine. [Lawrence 1985, 505]

31 Lawrence makes clear that the autonomy of the medical profession was at stake, and the Victorian physicians’ stress on the importance of experience, cultivated sensibilities and intuitive judgment, was as much a defensive rhetoric and professional ideology, as it was evidence for a tacit dimension that resisted efforts to convert it to mere technical routines.

32 Warwick Anderson discusses a more recent clash in ongoing efforts to turn clinical medicine into a science [Anderson 1992]. More explicitly than Lawrence, he also relates the case to themes in social studies of science. The episode he describes took place in Australian Hospitals in the early 1970s, and involved a computerized diagnostic system (a precursor to evidence-based medicine). Like Lawrence’s Victorian physicians, Anderson’s clinicians perceived the new “scientific” technology as a threat to their authority, and they resisted it by stressing the “artisanal” and “craft” knowledge cultivated through bedside practice. After having characterized the two sides in this contest, Anderson proposes to treat [...] the “craft” and “scientific” representations of diagnosis symmetrically, as discursive resources used in a hospital context to legitimate the divergent competences of the two occupational subgroups. [...] I do not intend to engage in the attempt to establish whether medical diagnosis is indeed, as the traditional

Philosophia Scientiæ, 17-3 | 2013 55

physicians argued, inarticulable in principle and practice: rather, what I do hope to show is that it served their interests to represent their work as such. [Anderson 1992, 655]

33 The diagnostic system that billed itself as “scientific” was computerized, but it was not an attempt to represent human clinical judgment. Like Taylor’s method, it was designed to replace and improve upon, rather than to reproduce, the clinician’s diagnostic methods. Anderson goes on to describe a clash between the competing “rhetorics” and “grammars” of clinical diagnosis, and he makes clear that how the discourse played out was a tangible constituent of the social-historical organization and re-organization of clinical practice.

Conclusion

34 Lawrence’s and Anderson’s treatments of “incommunicable knowledge” and “clinical craft” produce a kind of gestalt switch on a realist treatment of tacit knowledge. Instead of treating physicians’ accounts as evidence of a resilient type of tacit knowledge that strongly resists rationalization, Lawrence and Anderson treat those accounts as rhetorical efforts to secure authority in specific historical labor conflicts (for related cases in other fields, see [Gieryn 1999], and [Doing 2004]). Their analytical approach requires a detachment that is difficult to attain (or even to want to attain) in an academic profession that values autonomy and resists ubiquitous efforts to manage it.

35 Although Polanyi’s and Collins’s arguments are aligned with other efforts to support and defend the autonomy of particular professions in the face of real or imagined administrative encroachments, their realist treatments of tacit knowledge miss an important phenomenon: they fail to take account of how that concept and other concepts akin to it are used to claim and allocate authority and thus to constitute domains of expertise in contested situations. Collins is forthright in pursuing a programmatic effort to demarcate legitimate instances of expert knowledge from their Doppelganger (the various pseudo “theories” and dubious forms of skepticism that vie for public attention in the “marketplace of ideas”) [Collins & Evans 2007]. In line with that ambition, it is possible that he will come up with a set of “social” criteria for demarcating legitimate tacit knowledge from its rhetorical doubles. However, even if we grant the possibility that he may succeed in that endeavor (a possibility that I doubt), the point of this paper is not to prove that Collins is wrong, but to suggest that Lawrence’s social-historical investigation, and Anderson’s social-anthropological analysis provide alternatives to Collins’s efforts to construct conceptual typologies. The point I have made is a very simple one: tacit knowledge (along with its close cousins incommunicable knowledge, craft knowledge, etc.) is, first of all, a vernacular concept; a concept or theme that is used to do rhetorical work in concerted efforts to manage work and to resist such management. Consequently, one way to study tacit knowledge is to examine how it is used as a tool for constituting epistemic and political authority in particular historical conflicts. We might go so far as to say that this is an empirical sociological alternative to Collins’s ideological project.

Philosophia Scientiæ, 17-3 | 2013 56

BIBLIOGRAPHY

ANDERSON, WARWICK 1992 The reasoning of the strongest: The polemics of skill and science in medical diagnosis, Social Studies of Science, 22(4), 653–684.

BAR-HILLEL, YEHOSHUA 1954 Indexical expressions, Mind, 63(251), 359–379.

BLOOR, DAVID 1976 Knowledge and Social Imagery, London: Routledge & Kegan Paul.

BRAVERMAN, HARRY 1974 Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century, New York: Monthly Review Press.

BUSH, VANNEVAR 1945 Science, the Endless Frontier. A Report to the President, Washington, DC: U.S. Government Printing Office.

CAMBROSIO, ALBERTO & KEATING, PETER 1988 ‘Going monoclonal’: Art, science, and magic in the day-to-day use of hybridoma technology, Social Problems, 35, 244–260.

CHURCHLAND, PAUL & CHURCHLAND, PATRICIA 1990 Could a machine think?, Scientific American, 262(1), 32–39.

COLLINS, HARRY 1974 The TEA set: Tacit knowledge and scientific networks, Science Studies, 4, 165–186. 1975 The seven sexes: A study in the sociology of a phenomenon, or the replication of experiments in physics, Sociology, 9, 205–224. 1990 Artificial Experts: Social Knowledge and Intelligent Machines, Cambridge, MA: MIT Press. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

COLLINS, HARRY & EVANS, ROBERT 2007 Rethinking Expertise, Chicago: University of Chicago Press.

DOING, PARK 2004 ‘Lab hands’ and the ‘Scarlet O’. Epistemic politics and (scientific) labor, Social Studies of Science, 34(3), 299–323.

DREYFUS, HUBERT 1979 What Computers Can’t Do, New York: Harper and Row. 1992 What Computers Still Can’t Do: A Critique of Artificial Reason, Cambridge, MA: MIT Press.

GARFINKEL, HAROLD 1967 Studies in Ethnomethodology, Englewood Cliffs: Prentice Hall. 2002 Ethnomethodology’s Program: Working out Durkheim’s Aphorism, Lanham: Rowman & Littlefield.

GIERYN, THOMAS 1999 Cultural Boundaries of Science: Credibility on the Line, Chicago: University of Chicago Press.

HACKING, IAN 1999 The Social Construction of What?, Cambridge, MA: Harvard University Press.

Philosophia Scientiæ, 17-3 | 2013 57

HAND, LEARNED 1901 Considerations regarding expert testimony, Harvard Law Review, 15, 40–58.

HILGARTNER, STEPHEN 1990 The dominant view of popularization: Conceptual problems, political uses, Social Studies of Science, 20, 513–539.

JORDAN, KATHLEEN & LYNCH, MICHAEL 1992 The sociology of a genetic engineering technique: Ritual and rationality in the performance of the plasmid prep, in The Right Tools For the Job: At Work in Twentieth-Century Life Science, edited by CLARKE, A. & FUJIMURA, J., Princeton: Princeton University Press, 77–114.

LAWRENCE, CHRISTOPHER 1985 Incommunicable knowledge: Science, technology and the clinical art in Britain 1850-1914, Journal of Contemporary History, 20(4), 503–520.

LYNCH, MICHAEL 1985 Art and Artifact in Laboratory Science, London: Routledge & Kegan Paul. 2002 Protocols, practices, and the reproduction of technique in molecular biology, British Journal of Sociology, 53(2), 203–220.

LYNCH, MICHAEL & BOGEN, DAVID 1996 The Spectacle of History: Speech, Text, and Memory at the Iran-Contra Hearings, Durham: Duke University Press.

MERTON, ROBERT K. 1942 Science and technology in a democratic order, Journal of Legal and Political Sociology, 1, 115– 126.

MILLS, C. WRIGHT 1940 Situated actions and vocabularies of motive, American Sociological Review, 5, 904–913.

MULKAY, MICHAEL 1976 Norms and ideology of science, Social Science Information, 15, 637–656.

NYE, MARY JO 2011 Michael Polanyi and His Generation: Origins of the social construction of science, Chicago: University of Chicago Press.

PAPPERT, SEYMOUR 1968 The artificial intelligence of Hubert L. Dreyfus: A budget of fallacies, Memo 154, Computer Science and Artificial Intelligence Laboratory, MIT, Cambridge, MA.

POLANYI, MICHAEL 1958 Personal Knowledge, Chicago: University of Chicago Press. 2009 The Tacit Dimension, Chicago: University of Chicago Press, first edition, Garden City: Doubleday, 1966.

RUMSFELD, DONALD H. 2005 The unknown, in Pieces of Intelligence: The Existential Poetry of Donald H. Rumsfeld, edited by SEELY, H., New York: Free Press.

SUCHMAN, LUCY 2007 Human Machine Configurations: Plans and Situated Actions, Cambridge: Cambridge University Press, revised and retitled edition of Plans and Situated Actions, Cambridge: Cambridge University Press, 1987.

Philosophia Scientiæ, 17-3 | 2013 58

TAYLOR, FREDERICK WINSLOW 1911 Principles of Scientific Management, New York: Harper and Row.

TURNER, STEPHEN 1994 The Social Theory of Practices: Tradition, Tacit Knowledge and Presuppositions, Cambridge: Polity Press. 2012 Making the tacit explicit, Journal for the Theory of Social Behavior, 42(4), 385–402.

VOSKUHL, ADELHEID 2004 Humans, machines, and conversations: An ethnographic study of the making of automatic speech recognition technologies, Social Studies of Science, 34(3), 393–421.

WINCH, PETER 1958 The Idea of a Social Science, London: Routledge & Kegan Paul.

WITTGENSTEIN, LUDWIG 1953 Philosophical Investigations, Oxford: Blackwell, translated by G.E.M. Anscombe.

ABSTRACTS

Michael Polanyi and H.M. Collins contrast tacit knowledge with explicit knowledge. For Collins, secrets and other forms of “relational tacit knowledge” are tacit, but only in relation to specific circumstances and relationships. Collins treats such relational knowledge as less interesting theoretically than collective knowledge that is essentially difficult and perhaps impossible to convey through explicit formulations. In this paper I focus on relational tacit knowledge, despite its marginality in Collins’s typology, because it draws attention to conceptual ambiguities in the relationship between tacit and explicit knowledge. More importantly, such ambiguities come into play as strategic resources in historical labor conflicts and in efforts to secure and preserve professional autonomy. Instead of treating tacit knowledge as a substantive possession of individuals and groups, I suggest that an empirical sociological alternative is to investigate pragmatic and polemical uses of the tacit/explicit distinction in particular circumstances of action and conflict.

Michael Polanyi et H.M. Collins opposent la connaissance tacite à la connaissance explicite. Pour Collins, les secrets et les autres formes de « connaissance tacite relationnelle » sont tacites, mais seulement du fait de circonstances et de relations humaines spécifiques. Collins traite cette connaissance relationnelle comme moins intéressante d’un point de vue théorique que la connaissance collective qui est par essence difficile, voire peut-être impossible à exprimer par des formulations explicites. Dans cet article, je me focalise sur la connaissance tacite relationnelle, malgré sa marginalité dans la typologie de Collins, parce qu’elle permet d’attirer l’attention sur les ambiguïtés conceptuelles de la relation entre connaissance tacite et connaissance explicite. Qui plus est, ces ambiguïtés entrent en jeu, au titre de ressources stratégiques, dans l’histoire des conflits relatifs au travail et dans les efforts visant à sécuriser et préserver l’autonomie professionnelle. Au lieu de traiter la connaissance tacite comme une substance possédée par des individus et des groupes, je suggère qu’une alternative empirique et sociologique serait d’enquêter sur les utilisations pragmatiques et polémiques de la distinction tacite/explicite dans des circonstances particulières d’action et de conflit.

Philosophia Scientiæ, 17-3 | 2013 59

AUTHOR

MICHAEL LYNCH Department of Science & Technology Studies, Cornell University (USA)

Philosophia Scientiæ, 17-3 | 2013 60

Taking the Collective Out of Tacit Knowledge

Stephen Turner

1 In Tacit and Explicit Knowledge, [Collins 2010], hereafter TEK, Harry Collins argues against the use of a certain model of tacit knowledge, associated with the standard example of knowledge of how to ride a bicycle, as a model for all tacit knowledge. He distinguishes this kind of embodied knowledge, which he calls “Somatic” Tacit Knowledge, from two other kinds, “Relational” and “Collective” Tacit Knowledge. Collective Tacit Knowledge as Collins conceives it falls into a familiar category: it is a collective object. Collective objects are tempting explanatory devices for various reasons, but they also have characteristic problems, one of which Collins recognizes but makes no attempt to solve, namely the problem of transmission, or what Collins calls “the deep mystery” of “how to make explicable the way that individuals acquire Collective Tacit Knowledge” [TEK, 138]. Put differently, this is the problem of how a collective object gets into individual heads. Collins does argue, and argued in the past, that at least some kinds of tacit knowledge can be acquired only through direct contact with others who possess it. This suggests an alternative explanation of the phenomenon of tacit knowledge itself, one that is “social” or interindividual rather than “collective”. My concern in this article will be with the differences between these two kinds of explanation and the status of social explanations as rivals to collective explanations, especially in connection with tacit knowledge and Collins’s examples.

1 Social vs. collective

2 The terms “social” and “collective” are often used interchangeably, or without distinction, and this lack of discrimination is especially evident in the cognitive science literature. In the literature of social theory and sociology, in contrast, the terms are sharply distinguished, at least in theoretically sensitive contexts, and typically denote explanatory alternatives with different ontological implications. Indeed, the terms represent an important dividing line between traditions. Émile Durkheim made a point of criticizing his rival Gabriel Tarde for replacing

Philosophia Scientiæ, 17-3 | 2013 61

the expression “collective psychology” by “interpsychology”. The first expression appeared to him to be tainted with ontology, because it seems to imply that there is a collective psychology proper. [Durkheim 1906, 133]

3 The issue marked a major fault line within French sociology. One of the cardinal points made by Tarde was that the mechanisms for explaining social phenomenon involved processes between individuals, notably imitation, which incidentally plays a large role in cognitive science discussions of social interaction, rather than anything “collective” [Tarde 1890]. This led to the use of the term “social” as a way of distinguishing from and contrasting to “collective”. Durkheim of course rejected this bottom-up approach to the explanation of social phenomena. Durkheim’s sociologization of Kant involved the idea that there were collective, shared, psychological contents, contents irreducible to the processes of individual psychology, and that appeals to these collective contents were necessary to account for social life. His approach was explicitly “collective”. This fault line has persisted. In American sociology, movements such as Symbolic Interactionism rejected any such explanations, and attempted to explain such things as “significant symbols”, to use the term of G.H. Mead [Mead 1934], in terms of social interaction, but also from the bottom up, as products of the interactive process.

4 Outside of these contexts, however, confusion in the use of these terms reigns. “Collective” is used, especially in the context of cognitive science, for processes that are not collective but social in the sense of interindividual. Nevertheless, “social” is routinely used in these literatures in a way that does not discriminate between “collective” and “social” in the sense of social interaction. The point may be illustrated by a text by Michael Tomasello. On the one hand, he uses the “collective” language of “shared intentionality”: Underlying these two singular characteristics of human culture—cumulative artifacts and social institutions—are sets of species-unique skills and motivations for cooperation [...]. [W]e may refer to the underlying psychological processes that make these unique forms of cooperation possible as “shared intentionality”. [Tomasello 2009 xiii]

5 On the other hand, when he explains the underlying psychological processes, they are social; they involve interaction between individuals, where one individual is engaged with other individuals in ways that transform the other individuals, First, humans actively teach one another things, and they do not reserve their lessons for kin. [...] Second, humans also have a tendency to imitate others in the group simply in order to be like them, that is, to conform. [Tomasello 2009 xiv]

6 Imitation is an imperfect one-on-one process which does not produce collectively standardized results; teaching might be designed to produce standardized results as external behavior, but what people take away from the actual interactive experience of being taught, the internal psychology, is individual. They may be good or bad at arithmetic. They are disciplined to be standard in the answers they give to arithmetic questions. Being disciplined and desiring to conform produce social results. The skills and the desires allow teams to work together. But “shared intentionality” implies something more, a common mental content together with a common motivation. This parallels the issue in Collins: the idea of Collective Tacit Knowledge implies something more than that which may be transmitted by the mechanisms involved in personal contact mentioned by Tomasello or described by Collins himself.

7 Some of the confusion about these terms is legitimate. It is not always obvious which term is the correct one. There are widely used but problematic notions, such as

Philosophia Scientiæ, 17-3 | 2013 62

“sharing”, that make sense primarily as a collective fact, but which might also be understood as part of interaction, for example in relation to joint attention, with “social” but not “collective” implications. Yet it is easy to slide from facts about interaction to collective claims, despite their very different implications. Moreover, there are definitional issues that confuse matters. One may of course redefine the notions of “sharing” and “intention” in a way that eliminates the oddity of saying that multiple people have the same intention. One might redefine “intention” in terms of some common external goal, for example, such as winning a war, and treat as “sharing” anything intentional directed toward that goal. But this merely has the effect of allowing a huge variety of very different intentional and psychological causal structures and backgrounds, involving, for example, different and conflicting beliefs about the world, to count as “the same” intention, and thus something that could be said to be shared. But this is not a “collective” notion of intention in the sense that there is significant shared or collective internal or psychological content. Indeed, we may have very different concepts even of an external goal such as winning a war.

2 Collins and the idea of Collective Tacit Knowledge

8 In this article I will try to sort out some of these distinctions in relation to the case of tacit knowledge, which presents them in a particularly clear form. It is one of the virtues of Tacit and Explicit Knowledge [TEK] that it sharpens this issue. My basic thesis will be a simple one. Claims about “collective” facts normally depend on transcendental, “conditions for the possibility of”, arguments; the actual empirical evidence only supports non-collective “social” claims. This argument has direct bearing on the problem of tacit knowledge. One role of the concept of tacit knowledge has been to account for mutual interaction and understanding. In this role tacit knowledge is assumed to be shared, a common possession of those interacting, for example as speakers of a common language, and that this shared tacit knowledge is a condition for the possibility of the kind of communication through meaningful speech that language permits. Another role, however, has been to account for highly individual skills and competences, such as the skill of using one’s own physically distinctive body to perform complex tasks such as riding a bicycle. These roles are very different, and not obviously congruent, a point made by Collins.

9 Collins argues that the “bicycle” model of tacit knowledge, based on the standard example of embodied but inarticulable tacit knowledge of how to ride a bicycle, is unable to account for certain facts that a full account of tacit knowledge should account for. A “collective” kind of tacit knowledge would explain what is needed. This is a paradigmatic argument from explanatory necessity, with a transcendental argument as a solution, though Collins puts his claim somewhat differently, as a claim about what the most parsimonious explanations is. Whether this self-explication is appropriate is a question to be dealt with later. But there are two problems with it. In the first place, Collins does not consider alternatives, so it is unclear what it is more parsimonious than. The other is that ontologically, it is anything but parsimonious. Collins joins a long line of historical figures as well as contemporary thinkers who believe that there is some sort of collective mental element that is out there, in some sense, is assimilated or acquired by the participants, and functions as a shared structure which is in turn a condition for certain performances that cannot be

Philosophia Scientiæ, 17-3 | 2013 63

explained or accounted for in any other way. A collective intention, for example, is a fact beyond individual intentions and irreducible to them. To assert an irreducible new thing is to make a new ontological commitment. Traditionally, doctrines of collective intention have struggled with the question of what sort of collective intender or collective fact is implied by the notion of collective intention [Roth 2012], [Searle 1995, 2010], [Turner 1999]. An individual intention can be recognized and acknowledged by others or made the subject of agreement between people. Intersubjective agreement of this sort does not raise these questions.

10 What makes these arguments transcendental arguments is that “Collective Tacit Knowledge”, “collective intentionality”, and so forth are the conditions for the possibility of outcomes that Collins and similar users of these arguments believe cannot be explained in any other way. We are supposed to accept the existence of Collective Tacit Knowledge and his characterization of it because it meets this explanatory necessity. It is this general kind of conclusion that will concern me here, as it has in many other places. The counterclaim is this: the leap to a “collective” solution is unwarranted and the characterization of things to be explained as themselves “collective” is also unwarranted. But here I propose to explicitly discuss the distinction between collective and social in general, with an eye to explaining its sources and the case to be made against appeals to collective facts, as well as their undeniable attractions.

11 My basic point will be this: “Collective” mental objects, of which tacit knowledge in Collins’s account is an instance, are accessible only through transcendental arguments or arguments that share the problems of transcendental arguments. What I will show here is that these arguments are, as a group, defective in important ways that are relevant for any empirical explanatory account of the subject matter of tacit knowledge. As I have noted, Collins himself concedes one of the central issues with these accounts, the problem of how Collective Tacit Knowledge is acquired. In contrast, “social” explanations, that is to say those that involve interpersonal interactions and interpersonal processes such as imitation, are not subject to these defects; however, social mechanisms of the kinds that are generally known and accepted as genuine causal and psychological processes, never produce, account for, or explain the kinds of facts that collective accounts proceed from. But social explanations have their own issues.

12 To keep track of the relevant issues, it is useful to think in terms of a scoreboard. The advantages of the “collective” model, in Collins’s case the idea of Collective Tacit Knowledge, are these: a. Fills an explanatory need in a simple, or single cause, and in this sense “parsimonious” way (but see below on ontological non-parsimony). b. Enables understanding of other people. c. Enables claims about the content of what is tacit.

13 The disadvantages are these: a. Underdetermination: as with all regress or transcendental arguments, the Achilles heel of these arguments is the fact that the same overt results can be accounted for in different and inconsistent ways. b. The problem of location: it is difficult to get a coherent answer to the question of where the collective stuff is located, and consequently how it interacts with ordinary causal and interpersonal processes.

Philosophia Scientiæ, 17-3 | 2013 64

c. Transmission: there is no good way of accounting for how it is that the collective stuff is acquired by individuals, especially of how it is that the “same” stuff becomes part of the individual psychological or cognitive processes that are directly involved in thought and action explanation. d. Circularity: much depends, in these arguments, on characterizing the thing to be explained in such a way that only collective explanations qualify as explanations. e. Ontological non-parsimony: the explanation requires ontological commitment to a novel collective object, in this case Collective Tacit Knowledge.

14 “Individual” or “social interaction” accounts have their own issues, but they come down to three major problems: a. How can individual content and interactive processes operating between individuals aggregate to produce collective outcomes, such as a common language, co-operation, teams, joint collective action, and so forth? b. How are we to understand meaning, language, the worldviews of others, and practices if we do not treat these as collective facts? c. How can we understand alien cultures and contexts of knowledge if not by reference to their shared presuppositions, assumptions, tacit knowledge, and meanings?

15 In short, despite the oddity of their claims, collective accounts seem to be about something, and serve some important cognitive purposes. The question is whether these purposes can be served by a better alternative explanation.

16 The focus of my discussion here will be with the undeniable intellectual attractions of collective notions as means of achieving understanding. To know that the people of the Middle Ages “assumed” something different than we do about God’s place in a hierarchical teleological order of the world, for example, illuminates and makes intelligible some of the otherwise puzzling things that they wrote and believed. So to vindicate non-collective social explanations, these achievements also need to be accounted for, and in a way that is consistent with a reasonable understanding of the mechanisms that these non-collective social explanations employ. So in much of this article I will be concerned with giving an account of what sort of understanding one gets by attributing presuppositions, Collective Tacit Knowledge, and the like to a group of people, or to the individuals that compose the “group”. This is also the focus of much of Collins’s book as it relates to Collective Tacit Knowledge. Collins deals with this problem under the heading of the problem of making tacit knowledge explicit, something he does through the introduction of a series of distinctions organized around the notion of strings and the distinction between the physical and the meaningful, which he discusses in terms of transformations of strings as distinct from translation. My account will dispense with the whole apparatus of presuppositions, strings, Collective Tacit Knowledge, and the like as misleading and unnecessary. My concern will be with Collins’s claims about what he calls “The Irreducibility of the Collectivity” [TEK, 124].

3 Some deep history

17 The idea of collective mental properties or possessions begins inadvertently, with Kant. Kant contributed the idea of a transcendental argument and the notion that our possibilities of experience were shaped by categories that were the conditions of the

Philosophia Scientiæ, 17-3 | 2013 65

pre-organized experience of the sort that we actually conversed and thought about. Such apparent features of the world as space and causality were the products of this mental pre-organization. The underlying organizing principles were tacit, though Kant did not use this language (in fact there is a problem in translating the term tacit knowledge into German in the first place, but the usual translation, Implizites Wissen, already transforms it into a quasi-Kantian notion) in the sense that they were normally not articulated, but could be.

18 Kant had no intention of introducing a “collective” account of these presupposed conditions of thought. And despite the psychological language of the mental of cognition that he employed, he insisted that he was not offering a psychology at all, but rather an account of the logical conditions of knowing. Nevertheless, someone had to possess these conditions, in some sense, to be a knower. Kant thought that the most basic conditions at least were common to all rational beings and indeed a condition of rationality. The philosophical argument for their presence and necessity worked with two notions. The first was the idea of a regress, in which premises needed to be justified by more basic premises. The second was the idea that this kind of analysis could produce a unique result that excluded the possibility of alternative premises for the same conclusions [Paulson 1992, 324–332]. If the claim to exclude alternatives worked, it would obviate any need to consider the possibility that, for example, different people had different presuppositions or implicit knowledge yet nevertheless managed to get around in the world, and even communicate; there would be only one possible set of presuppositions for everyone.

19 The neo-Kantians radicalized this argument by applying it not only to the certain knowledge of the physical world supplied by physics, but to a whole range of domains of thought in which objective knowledge was possible. Hermann Cohen, the key radicalizer, applied the basic idea of a domain of thought organized in a tacit but reconstructable logical hierarchy of concepts to ethics, which he argued had the character of a logically organized domain because jurisprudence, which was organized around a constitutive ethical concept, was itself a logically organized domain. The “fact of science”, the fact that there was an organized domain, implied that there was a set of presuppositions and a logical hierarchy of concepts that constituted it. The circularity is evident in the characterization of the domain: what else could account for a domain with this kind of conceptual order than an explanation of that order, and what would explain it other than a hierarchical arrangement of the concepts in terms of their logical relations?

20 Cohen interpreted this kind of analysis epistemologically as distinct from psychologically, like Kant. Kant had made a great point of claiming that, as Warren Schmaus put it, “he was not offering his theory of categories as an empirical, psychological account of the origin of experience” [Schmaus 2004, 30], but rather of what was presupposed by experience. Thus, as Schmaus paraphrases Kant, “to say, for instance, that the category of quantity is necessary for experience [...] is to say only that one could not experience objects without their having some quantity or another” [Schmaus 2004, 30–31]. But the application of these ideas to historically specific domains, such as the law, and then to the various academic disciplines, including theology, had the effect of localizing them to the individual members of these disciplines who were actually “knowers” who thought in terms of the organized concepts of their discipline. The trick of saying “one could not experience objects

Philosophia Scientiæ, 17-3 | 2013 66

without their having some quantity or other” does not work with the objects of specialized disciplines: one can experience the actions of agents of the law without presupposing anything specifically “legal”. And one can understand the theological beliefs of others without presupposing the existence of God. So the “conditionality” could not be bound to experience, and migrated to the knower. When the idea of a domain constituted by hierarchically organized concepts was extended to the “historical a priori”, that is to say to the presuppositions of a past epoch that distinguished it from our own, the localization was made even more apparent. To say that people in the Renaissance experienced Il Duomo differently than we do implies that the difference is in them, not in Il Duomo. For localized domains, the categories and concepts involved were not conditions of experience as such, but of the highly specific objectivity inducing experiences of people who in some sense possessed the constituting concepts. In the law, for example, this would be people trained in the law, who could agree on objective facts of legal science. Along with localization came another shift, more subtle, from “experience” to “meaning”: thus the objective facts of legal science were distinctive because they were legally meaningful, and meanings in general became a special realm that had to be constituted by presuppositions.

21 The presuppositions in these cases had to belong to the people who were of the time, or of the discipline. And because the function of the presuppositions was to assure or explain the “objectivity” of results in the domain, they had to be uniform or shared. None of this was argued for: it was a given in the Kantian lineage of the idea of objectivity and constitution. Because these were not thought of as matters of psychology, but as matters of the logic of concepts—Cohen, for example, claimed to be making inferences from the facts of universal jurisprudence, a mythical discipline unconnected to actual lawyers—there was no sense that a psychological account was needed. Eventually, various neo-Kantians saw the need to provide accounts that made sense of the facts of localization.

22 This contrast—epistemic or meaningful as distinct from psychological and local or historical—produced a peculiar and confused result. What was the relation between the fact of localization, the fact that only a set of related people were knowers of this kind, and the epistemic content, the presuppositions, themselves? Was it possible to avoid questions like “how were the presuppositions acquired by these knowers, and why did some people have them and others not have them?” In short, can one historicize, socialize or sociologize epistemology without at the same time psychologizing it?

23 For these localized presuppositions there was no avoiding the question of where they came from and what sort of reality they had. The more common and plausible solution to the problem of where the presuppositions came from and had their real existence was to place them into some sort of collective mind. The group mind is a solution to the problem created by localization. But it is a concept with its own history. Ontologizing the problem of collective mental life introduced a novel and problematic entity. Part of the Kantian legacy was to separate mind and brain, which facilitated the idea that groups could have minds without the causal accouterments of brains. But there were still nagging problems about cause. If the presuppositions of the group are not given in experience and thus accessible to individual reflection, they have to be acquired. And if they are presupposed by experience, they have to be acquired some way other than by ordinary learning, which operates through experience, with empirical inputs, and produces individual mental results, like habits and beliefs, rather than collective

Philosophia Scientiæ, 17-3 | 2013 67

possessions. This is the core of the transmission problem that Collins acknowledges but, like the neo-Kantians, cannot solve.

24 The larger problem here is about the distinction between epistemic and psychological inherited from Kant: the status of the presuppositions, concepts, and categories which are the conditions of explicit knowledge. If we grant that they are logically required, does this imply that they are therefore necessarily causally real in some sense? This kind of reasoning seems to generate causal claims from transcendental arguments. And there is a long tradition of transcendental reasoning which implies something like this, but it is vague or evasive about what is implied [Turner 2010, 14–30].

4 Separating the logical and the psychological

25 We think that there ought to be some sort of close relationship between the logical and the psychological or the social: when we say “we” or “possession”, as well as “knowledge”, we seem to be talking about something that is in people’s heads and accessible, to some extent, to them. The fact that we can sometimes self-explicate our presuppositions, or make our tacit knowledge explicit through explaining ourselves more fully, or do this for others by identifying their presuppositions, seems to indicate that we are saying something about what is already there in the mind. But the logical and the psychological work in quite different ways. So this sense may simply be an illusion.

26 Consider the following comment by Hans Kelsen about the interpretation of the League of Nations covenant, paragraph 1 of article 4. It reads “The Council shall consist of Representatives of the Principal Allied and Associated Powers [...]”. As he points out, it would have made the United States and Japan permanent members of the council, despite the fact that the United States never ratified the treaty and Japan withdrew. As Kelsen says: [...] a literal interpretation of paragraph 1 certainly does not correspond to the intentions of the authors of the statute. They assumed as a fact that all the “Principal Powers” would ratify the Treaty of Versailles at least and thereby become members of the League. They did not think at all of the case where one of these states, after having ratified the Treaty, would withdraw from the League. They therefore refrained from stipulation that only members of the League could be represented on the Council. [Kelsen 1939, 48]

27 The two middle sentences of this passage are related as follows: there is a claim about an “assumption” and another claim about something “they did not think at all”. The initial claim about what they assumed is not in conflict with the second claim about the psychological fact about not having noticed the possibility that countries would fail to ratify or leave the League. The second claim is, rather, a clarification or psychological explanation of the first. In this case the fact is not, however, an occurrent fact (i.e., something that the person is currently conscious of or is otherwise manifest), but an omission or absence of a psychological fact. Yet both the psychological claim and the “assumption” claim are true and they are consistent with one another. Indeed, as Kelsen’s passage reads, the assumption claim is in some sense supported and explained by the psychological one.

28 So what sort of fact is the fact about the assumption? One way of approaching this question is to ask for whom it is a fact. If it is not a fact in a psychological sense of

Philosophia Scientiæ, 17-3 | 2013 68

introspectability—one cannot introspect an omission—it was therefore not an introspectable fact for the signatories. Is it a fact for the interpreter, who is providing an explanation of the failure to make the logically possible inference that countries might withdraw from the treaty? Is the explanation the kind of explanation that makes the inference intelligible—intelligible to the interpreter and the audience of the interpreter?

29 The issues here become a little clearer if we consider some different ways of formulating Kelsen’s comment, and the relation between the “did not think” and the “assumption” parts of the sentence. If we treat the assumption part as an “as if” statement, namely that the writers of the document proceeded “as if” nations would not withdraw from the covenant, we capture the meaning without adding any term, such as assuming, that suggests something psychological or suggests that the “assumption” is a transcendental condition. The sentence becomes one in which the “as if” clause allows us to give an intelligible interpretation, though not necessarily a correct one. The notion of “as if” is explicitly an appeal to a fiction in Vaihinger’s sense [Vaihinger 1911]: it contradicts the actual psychological facts to say that the signatories actually “assumed” this. But it is a fiction that allows us to make sense of their reasoning, which would otherwise be unintelligible.

30 It is a fiction that, for a particular audience, makes sense of the conduct of the authors of the document. First, it is an audience that already understands the meaning of the terms of the fiction. But second, the need for this particular kind of sense-making is restricted to people who have noticed the problem of withdrawal from the covenant, and only to those people. The people who wrote the treaty did not need to think about, or use, the “assumption” because they all failed to notice the issue. They also did not understand each other in the fictive or hypothetical “as if” sense; they were living in a world in which this possibility had not come to notice.

31 The failure to notice was a “failure” only from a different perspective. From theirs, there was only agreement and mutual understanding, and nothing to explain. They all noticed, and attended to, the same things. If we had the power to appear from the future, and proposed a clause which resolved the problem, they might say “we did not notice that”, or they might say “you are assuming that nations will withdraw from the covenant”. But we wouldn’t have “assumed” anything either. We would simply have had a different experience—in this case the historical experience of withdrawals from the League. If we had not had that experience, we might also fail to notice this possibility. Indeed, as there was no legal procedure in the covenant for withdrawal, there is a sense in which the possibility of withdrawal was itself invented by those who wished to withdraw, and by their insistence on treating the covenant as merely a treaty which could be repudiated in the normal manner of formal repudiations of treaties, something that was not intended or even imagined by the original authors of the covenant.

32 This gets us to an odd result: claims about “assumptions” say no more than “as if” claims. The point of appealing to claims, in this practical context of interpretation, is to make sense of some belief or action that we do not understand. The aid to understanding what claims about assumptions provide is relative to the beliefs, noticings, and so forth of the persons interpreting the claims. The correctness of claims about assumptions as well as the content of the “as if” claims are relative to the specific

Philosophia Scientiæ, 17-3 | 2013 69

misunderstanding, without a misunderstanding there is nothing for the claim to be about.

33 This does not mean, however, that someone—the authors of the covenant, for example —would not assent to the attribution of the assumption as a clarification of what they meant, or even produce the “assumption” if they were faced with a question that implied that withdrawal was possible. The question itself, in this case, would have brought the possibility to notice, or at least brought to notice the fact that the questioner had imagined this possibility. So to answer “yes, that is what was assumed” tells us nothing about a state of mind in the past. It is merely a retrospective interpretation provided in response to a new question which implied a possibility that had not been considered. This, indeed, is a model for what happens when the tacit is made explicit: the person making the tacit explicit tells the hearer what he thinks would enable the hearer to make sense of the relevant inferences.

5 What happens when the tacit is made explicit?

34 The Kelsen example suggests that there is no stable object that corresponds to such things as “presuppositions”. In his discussion of Collective Tacit Knowledge, Collins is unequivocal that it is such as stable object: “knowledge that is located in society” [TEK, 138], a “shared background of tacit knowledge” that are necessary because they “work as conditions [of] communication” [TEK, 142], and so forth. The key to his argument against the bicycle model is precisely that Collective Tacit Knowledge, unlike bodily tacit knowledge, can be conveyed through language as well as through sharing physical activities, and indeed that “Collective Tacit Knowledge is, to a large extent, located in the language of the collectivity rather than its practices” [TEK, 135]. But according to Collins, the tacit knowledge of the individual who uses the language and the knowledge located in it is knowledge of how to decode strings. This creates a large puzzle, which cannot be pursued here: if the material of interaction is strings, how does “language” relate to this, and how can “knowledge” be found in language, if language in its external sense consists of strings?

35 One way of thinking of this is that the individual possesses presuppositions that permit him or her to use the language, presuppositions that are somehow transmitted along with the learning of the language itself. But are there such things as these presuppositions that are shared and are the condition of communication? Is there something collective shared in addition to the language itself? Collins provides us with no argument that there is. He merely asserts that the conditions of communication require it. But is this really so? If we look past the problematic language of “presupposition” to the interactive processes themselves, we get an image of social life and mutual understanding like this: we interact with one another in terms of gestures and statements of various kinds, which we interpret in ways that make sense to us. When they do not, we quickly generate alternative ways to make sense of what we are told. We can do this effortlessly and unconsciously. But on occasion we also do it consciously and explicitly, for example in trying to understand people who are very different from us and live in very different societies or hold very different beliefs.

36 Take a simple example: it is customary for the Chinese to respond to statements that are unwelcome or to which they object with silence—silence conveys the message that the statement is objectionable. But would the person who was raised with this custom

Philosophia Scientiæ, 17-3 | 2013 70

be able to find the appropriate functional substitute for it, when faced with a foreigner who was oblivious to the message being sent by silence? Would the Chinese who were not aware of the specificity of this custom, and for that matter the fact that it was a custom and not the normal human response to unpleasant speech, even understand that the person who was behaving badly by continuing to speak in an unwelcome way was in need of an explanation of the silence?

37 This case is a simple example of tacit knowledge being made explicit. It is found in business travelers books on how to behave in China. The behavior of the Chinese is interpreted as a rule which can be formulated as an explicit functional substitute, “silence means that the person is waiting for you to say something different and more welcome”. But this is a rule for someone in a culture in which there is an expectation that people signal and then explain their displeasure with an utterance. If the traveler was from a culture in which every act was interpreted in terms of status and therefore deference, and deference needed to be acknowledged, the advice might be “silence means that your hearer expects you to show deference by saying something acknowledging his displeasure as well as his status”.

38 They are functional substitutes only for people whose expectations are formed by those societies. “We” are unaccustomed to thinking in terms of prestige and deference, and would need to have the specific concepts of prestige and deference translated, explained, and illustrated with examples before we would find the second rule usable. Neither rule is in the head of the Chinese ready to be made explicit—indeed, treating these as “rules” at all is merely to employ a convenient analogy. They are rules for travelers to use to interpret behavior and act in ways that are functional substitutes for the way they would act within their own society. But they are substitutes only. Truly fluent interaction, of the kind that is possible in the situations in which one is most familiar, would involve the full use of our capacity to attribute intentions and beliefs to others, our capacity to repair and revise attributions, and our capacity to make ourselves understood by others who have misinterpreted us by providing repairs to their inferences.

39 This way of formulating the problem of what it is that we do when we articulate tacit knowledge has the effect of shifting the problem from the act of stating the tacit to the social interaction between the person doing the stating and the person for whom the stating is being done. In discussing the Kelsen quote, I have already stressed the sharp differences between the psychological facts and the use of the notion of “assumption”. Assertions about “assumptions”, this discussion suggested, are better understood as attempts to provide functional substitutes for a particular audience, often an audience different from the audience in the original setting; in the case of the signatories of the League of Nations, the original audience being the diplomats of the day and the other audience being the later observers who experienced the withdrawal of states and wondered why this wasn’t provided for in the treaty itself. But the use of the term “audience” raises a different problem, about “sharing”, a notion to which, as we have seen, Collins also appeals.

40 As I have noted, we are routinely called upon to explain ourselves to people who don’t understand us, and we do so by empathically imagining what it is that they have wrongly inferred or wrongly believe or “assume” that we can correct by providing information that will repair the relevant inference. This interactive process does not appeal, in any essential way, to the notion of rules, or to anything “collective” or

Philosophia Scientiæ, 17-3 | 2013 71

shared. “Functional substitutes” are invented on the fly as a normal part of interaction, and the ability to do this, which depends perhaps on our capacity to empathically think through or simulate the thinking of the other, applies to a huge variety of tasks, including the understanding of historical figures. When Kelsen explains the thinking of the original negotiators of the League of Nations treaty provisions, similarly, the explanation is one that could just as well be applied to a single individual: Woodrow Wilson, for example.

41 These attempts at explanation, like the rule about the meaning of silence for the Chinese, are audience relative and underdetermined, just as our unconscious supplying or inventing of presuppositions is. The fact that we interact with people who are similar to us in many respects means that our usual inferences about the actions of others, their intentions, the relations between their words and deeds, and the inferential relations between the different kinds of things they say become habitualized and automatic. Moreover, our interactions with others increase the extent to which our habitualized responses serve the purposes of understanding the people around us. To the extent that there is tacit knowledge, it is contained in these habitualized responses.

42 There is nothing “collective” or even “social” about this: the habits are ours as individuals. The social element is found in the capacity to guess what other people need to know in order to understand us when an interaction goes wrong, our capacity to invent and attribute inferences to others—to repair what has gone wrong when our habitualized responses fail, or those of the person we are interacting with fail. This capacity to read their minds and determine what they need to be told explicitly to repair the interaction may be rooted in such parts of the neuro-cognitive system as mirror neurons and simulation. But in any case these capacities are social rather than collective in their reach. The diplomats who signed the treaty establishing the League of Nations, like Woodrow Wilson himself, failed individually to anticipate withdrawals from the treaty. Saying “we” failed is simply to record this fact, not to invoke a collective mind or common mental processes leading to this failure. Similarly for the notion of audience I have employed here. It is merely a collection of individuals, not a body with some sort of shared collective mental content. If this is a correct account of what goes on when the tacit is made explicit, there is no explanatory need to appeal to Collective Tacit Knowledge, and therefore nothing for it to explain, parsimoniously or otherwise.

6 The scorecard

43 The “scorecard” with which I began was a list of issues with collective accounts and with their rival non-collective social accounts. How do these issues look after these considerations? It no longer looks as though Collective Tacit Knowledge fills an explanatory need in a simple, “parsimonious” way. It fills one need, the need for sense- making, by attributing presuppositions to a collective object; at the same time it ignores the problem of how this collective object gets into people’s heads. Moreover, it requires a new ontological commitment to the collective object itself. This can only be justified by a claim of explanatory necessity: we don’t have independent access to this object. But it is not clear that there is an explanatory need that the concept fills: we can account for understanding in an ontologically more parsimonious way by reference to social rather than collective processes. The understanding of other people enabled by

Philosophia Scientiæ, 17-3 | 2013 72

the idea of Collective Tacit Knowledge and the content it supposedly gives us access to turn out to be more readily accounted for by our capacity to invent inference repairs. It is of course possible to describe the things to be explained in ways that force us to appeal to collective facts. But it is always possible to describe these differently withoutloss of empirical content.

44 The idea of Collective Tacit Knowledge, like all similar ideas, faces some basic issues related to the idea that these were stable objects of some kind. One was underdetermination: there is no stable thing corresponding to “presuppositions” or tacit knowledge. There are, rather, multiple possible explications, some of which will be intelligible to certain audiences, none of which is uniquely determinate. One might insist that this inaccessibility through regress arguments is precisely the point: the shared tacit stuff is mentally inaccessible, tacit in the deepest sense, but still shared. But to say this creates another problem: if the stuff is inaccessible, how did it get into the heads of the people who supposedly possess it in the first place? This is the downloading problem, and it is insoluble: even if there was a stable determinate object that was shared, it would be impossible to explain how this stable object was related to the fact of activity and the constant flux created by normal social interaction. The fact that the location of this supposed object and the problem of its interaction with the causal world has been a problem ever since Kant, and indeed the sheer diversity of conceptions about what is supposedly out there or in the mind, and where it is located, suggests that there is no stable determinate object of this kind.

45 Can “individual” or “social interaction” accounts avoid these problems yet account for what needs accounting for: a common language, co-operation, teams, joint collective action, practices, and so forth? All of these involve mixtures of explicit beliefs, habits, mutual understandings, and sometimes common material objects. They all work through interaction, interaction that pushes inferences into the habitual and therefore tacit. The mystery elements in these phenomena are the aspects that are inaccessible to consciousness or accessible only through the attribution of presuppositions. If we understand presuppositions in the sense of David Lewis, there is no mystery. The constant invention of these things is a normal part of social interaction. There is nothing that needs to be located outside the mind, transmitted, downloaded, or to exist in some mysterious causal relation to ordinary cognitive processes. The creation of these repairs is an ordinary cognitive process. Cultures and the like are not things with essences or jointly motivating spirits. They are not causes at all. These terms are simply aggregate descriptions and typifications of what people do and believe, and they do habitually and therefore tacitly that is different from what we do.

46 Collins is a major representative of what could be understood as a naturalistic approach to the study of science, beginning with what he called the Empirical Programme of Relativism [Collin 2011, 83–108], [Collins 1981]. The most famous text on naturalistic epistemology, Quine’s “Epistemology Naturalized”, took the view that epistemology would eventually collapse into or be replaced by scientific knowledge of the process of knowing [Quine 1969]. Kant’s distinction between psychology and the epistemic was the target of this claim, just as Kant’s analytic synthetic distinction was the target of “Two Dogmas of Empiricism” [Quine 1951]. It is not a small irony that Collins proves to be the heir not of naturalism, but of Kant and these two problematic distinctions, as they were carried through in the equally problematic tradition of the collective mind.

Philosophia Scientiæ, 17-3 | 2013 73

BIBLIOGRAPHY

COLLIN, FINN 2011 Science Studies as Naturalized Philosophy, Dordrecht: Springer, chap. Harry Collins and the Empirical Programme of Relativism, 83–108.

COLLINS, HARRY 1981 Introduction: Stages in the empirical programme of relativism, Social Studies of Science, 11, 3– 10. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

DURKHEIM, ÉMILE 1906 Review of Gabriel Tarde, ‘L’Interpsychologie’, Bulletin de L’Institut Général Psychologique, juin 1903, 1–32, L’Année sociologique, 9, 133–135.

KELSEN, HANS 1939 Legal Technique in International Law: A Textual Critique of the League Covenant, Geneva: Geneva Research Centre.

MEAD, GEORGE HERBERT 1934 The vocal gesture and the significant symbol, in Mind Self and Society from the Standpoint of a Social Behaviorist,, edited by MEAD, G. H. & MORRIS, C. W., Chicago: University of Chicago Press, 61–67, cited according to the 1972 edition. URL http://www.brocku.ca/MeadProject/Mead/pubs2/mindself/Mead_1934_toc.html.

PAULSON, STANLEY 1992 The neo-Kantian dimension of Kelsen’s pure theory of law, Oxford Journal of Legal Studies, 12(3), 311–332. URL http://www.jstor.org/stable/764769.

QUINE, WILLARD VAN ORMAN 1951 Two dogmas of empiricism, The Philosophical Review, 60, 20–43. 1969 Epistemology naturalized, in Ontological Relativity and Other Essays, New York: Columbia University Press, 69–90.

ROTH, PAUL 2012 Searleworld, History and Theory, 51(1), 123–142.

SCHMAUS, WARREN 2004 Rethinking Durkheim and His Tradition, Cambridge: Cambridge University Press.

SEARLE, JOHN 1995 The Construction of Social Reality, New York: The Free Press. 2010 Making the Social World: The Structure of Human Civilization, Oxford: Oxford University Press.

TARDE, GABRIEL 1890 Les Lois de l’imitation: étude sociologique, Paris: Félix Alcan, Cited according to the English translation by Elsie Clews Parsons, The Laws of Imitiation, New York: Henry Holt & Co., 1903.

TOMASELLO, MICHAEL 2009 Why We Cooperate, Cambridge, MA: The MIT Press.

TURNER, STEPHEN 1999 Searle’s social reality. Review essay of John R. Searle, The Construction of Social Reality,History

Philosophia Scientiæ, 17-3 | 2013 74

and Theory, 38, 211–231. 2010 Explaining the Normative, Oxford: Polity Press.

VAIHINGER, HANS 1911 Die Philosophie des Als Ob: System der theoretischen, praktischen und religiösen Fiktionen der Menschheit auf Grund eines idealistischen Positivismus. Mit einem Anhang über Kant und Nietzsche, Berlin: Reuther & Reichard, cited according to the English translation by C.K. Ogden, The Philosophy of “As If”: A System of the Theoretical, Practical and Religious Fictions of Mankind, New York: Harcourt, Brace and Company, 1925.

ABSTRACTS

The concepts of “collective” and “social” are routinely confused, with claims about collective facts and their necessity justified by evidence that involves only social or interactional facts. This is the case with Harry Colllins’ argument for tacit knowledge as well. But the error is deeply rooted in the history of philosophy, in the notion of shared presuppositions popularized by neo- Kantianism, which confused logical claims of necessity with factual claims about groups. Claims of this neo-Kantian kind have difficulties shared by Collins’s argument for Collective Tacit Knowledge. The alternative, a fully social account of our capacities of understanding and interpretation, avoids these difficulties and allows for a model of tacit knowledge that serves the supposed explanatory needs that the “collective” part of Collins’s account purports to meet.

Les concepts de « collectif » et « social » sont couramment confondus, via des affirmations à propos des faits collectifs et de leur nécessité justifiées par des éléments de preuve mobilisant uniquement des faits sociaux ou interactionnistes. C’est notamment le cas dans l’argument de Harry Collins en faveur de la connaissance tacite. Mais l’erreur est profondément enracinée dans l’histoire de la philosophie, via la notion de présuppositions partagées popularisée par le néo- kantisme, lequel a confondu les énoncés logiques de nécessité avec les énoncés factuels à propos des groupes. Les affirmations néo-kantiennes de ce type se heurtent à des difficultés partagées par l’argument de Collins en faveur de la connaissance tacite collective. La solution alternative, un compte rendu complètement social de nos capacités de compréhension et d’interprétation, évite ces difficultés, et permet un modèle de la connaissance tacite qui satisfait les supposés besoins d’explication que la partie « collective » du compte rendu de Collins vise à satisfaire.

AUTHOR

STEPHEN TURNER University of South Florida (USA)

Philosophia Scientiæ, 17-3 | 2013 75

Tacit Knowledge and Its Antonyms

Tim Thornton

1 Introduction

1 In Tacit and Explicit Knowledge [Collins 2010a], hereafter [TEK], Harry Collins explores what is tacit by setting out what stands opposed to it, the various antonyms of “tacit” in the book. The antonyms include different ways in which things can be “explicated” and the way they can be encoded in or carried by “strings”. But, as I will argue in the second section of this paper, the subject matter of the contrast between what is tacit and what can be explicated turns out not to be knowledge but rather what is known: the nature of a worldly task or process rather than how one knows how to do that task (for example, tacitly or explicitly). One clue to this is the way that the tacit status of one subject’s knowledge of how to carry out a task or process can be affected by facts about the mechanization of that task or process elsewhere. I will argue that the presence in the account of such “action at a distance” is explained by the fact that Tacit and Explicit Knowledge is really a book about ontology rather than epistemology.

2 In the third section of the paper, I suggest a diagnosis of the blurring of epistemology and ontology. It results from the selection of strings—“bits of stuff inscribed with patterns”—rather than what can be expressed in linguistic signs as the key antonym for tacit. And that in turn, I suggest, stems from a failure correctly to follow the lessons of Wittgenstein’s discussionof rule following.

3 To put these, perhaps contentious, claims in context, it is helpful to compare Collins’s work with Michael Polanyi’s seminal text. Polanyi starts his book The Tacit Dimension with the following slogan: I shall reconsider human knowledge by starting from the fact that we can know more than we can tell. [Polanyi 1967, 4]

4 The broad suggestion is that knowledge can be tacit when it is, on some understanding, untellable. Tellable knowledge is a subset of all knowledge and excludes tacit knowledge. But as Polanyi—like Harry Collins more recently—immediately concedes,

Philosophia Scientiæ, 17-3 | 2013 76

the slogan is gnomic, e.g. [TEK, 4, 85, 88–91]. Does it carry, for example, a sotto voce qualification “at any one particular time”? Or does it mean: ever? Polanyi continues: This fact seems obvious enough; but it is not easy to say exactly what it means. Take an example. We know a person’s face, and can recognize it among a thousand, indeed among a million. Yet we usually cannot tell how we recognize a face we know. So most of this knowledge cannot be put into words. [Polanyi 1967, 4]

5 But this still leaves the question of the nature of the impossibility unclear. Can aspects of it be put into words but just not all of it at the same time? Or are the elements themselves ineffable? Polanyi’s slogan suggests an approach to understanding tacit knowledge which resembles the via negativa in theology: approaching the nature of God by describing what God is not, the (finite and limiting) properties God does not have. In the case at hand, it is characterising what is tacit by selecting and investigating a suitable antonym. Tacit knowledge is not explicit, for example, on a suitable understanding of “explicit”.

6 Harry Collins takes the same general approach in Tacit and Explicit Knowledge. He describes this strategy in a pithy summary at the very start of his book: “explain ‘explicit’, then classify tacit” [TEK, 1]. Thus the first three chapters address the nature of what is explicit and the next three examine three kinds of tacit knowledge which vary in accord with their impediments to explicitness. They are relational, somatic and collective tacit knowledge. Each of these later chapters ends by asking whether the form of knowledge concerned could be made explicit. The different kinds of tacit knowledge correspond to different possibilities for explicitness and hence different underlying antonyms. There are, he says, “three main kinds of reason for not being able to write things down” [TEK, 80].

7 As I will argue, however, his approach to the nature of what is explicit faces a dilemma common to any attempt to characterise tacit knowledge which stems from the fact that it has to be tacit and it has to be knowledge. But it is not easy to meet both conditions. Playing up the tacit status threatens the idea that there is something known. Whilst articulating a knowable content, that which is known by the possessor of tacit knowledge, risks making it explicit.

8 I can give an example of the danger by quoting Collins’s recent summary of some of his own earlier work on tacit knowledge (before his more recent distinction) set out originally in Changing Order. In the 1970s, he visited six of the seven UK laboratories that were attempting to build a working laser of a new design (a Transversely Excited

Atmospheric pressure CO2, or TEA, laser), although it had already been successfully built in other laboratories in the US. In one case, a scientist who had already built one working model aimed to replicate it so as to have two working models. Despite this limited problem—a clear case of Kuhnian “normal science”—and despite the availability of explicit instructions, Collins discovered a surprising difficulty. [N]o scientist succeeded in building a laser by using only information found in published or other written sources. Thus every scientist who managed to copy the laser obtained a crucial component of the requisite knowledge from personal contact and discussion. [Collins 1985, 55]

9 Returning to summarise his investigation in a recent popular article, Collins writes: [Y]ou may not know what you need to know and I may not know what I know. Thus, in the early days of TEA lasers scientists did not necessarily know that the inductance of the top lead was important but by copying existing designs they built in successful short top leads without knowing why. [Collins 2010b]

Philosophia Scientiæ, 17-3 | 2013 77

10 The first sentence echoes Polanyi’s slogan and thus suggests that what follows is supposed to be a case of tacit knowledge (and the article is called “Tacit knowledge: you don’t know how much you know”). But whilst it seems plausible to deny that the scientists had explicit knowledge of the requirements on the length of the top lead, and hence it might count as tacit in some sense of that term, it is not clear how it counts as tacit knowledge since no one seems to have had knowledge of the top lead. Successful short leads seem not to have been the result of sensitivity to the length but rather a featureof a copying process.

11 Outlining the general challenge and illustrating it with some of Collins’s earlier work is not to say that it is not met in his later work. But, as I will argue in the next section, Tacit and Explicit Knowledge does fall prey to this problem. In concentrating on the nature of tacitness by setting out various antonyms, Collins lets slip the knowledge status of tacit knowledge. He blurs epistemology and ontology.

12 Ascribing a precise view of the nature of tacit knowledge to the book is, however, more difficult than it might first seem because, as well as making some direct claims about tacit knowledge, the book also describes some conventional uses of that phrase against which it cautions. Thus the book says: We are now in a position to explain the tacit and the explicit. That which is not explicit knowledge is mostly just the way the world unfolds. Sometimes it is referred to as tacit knowledge. Much of it consists of the working out of mechanical sequences of greater (cats, dogs, humans-as-animals, paint sprayers, neural nets) or lesser (trees, sieves) complexity. For most of this the term “tacit knowledge” should not be used, the notion of mechanism being more appropriate [...]. [TEK, 80, italics added]

13 The first part seems to make a direct claim about the nature of tacit and explicit knowledge but later on there is a description of use and a warning about it. There is thus a standing possibility of misinterpreting a description aimed at making explicit an everyday sense of “tacit” with an endorsement of the view. To attempt to minimise this risk I quote the work more than is customary in such a paper.

2 The knowledge status of tacit knowledge in Tacit and Explicit Knowledge

14 The first sentence of the first chapter of Tacit and Explicit Knowledge says “Tacit knowledge is knowledge that is not explicated” [TEK, 1, italics added]. This might be terminologically anodyne if “explicate” were taken to mean make clear. If so tacit knowledge would be knowledge that is not, or perhaps could not be (a distinction to which Collins returns), made clear, which is close to Polanyi’s slogan, we can know more than we can tell. In fact, however, Collins expands his initial opposition of tacit and explicit to explicable and then to four distinct meanings of “explicable”. There are thus four distinct antonyms to “tacit” because there are four ways in which he characterises explication. These are explication by elaboration, transformation, mechanization and explanation [TEK, 81] (with elaboration perhaps the closest to “make clear”).

15 These four senses of “explicable” give four senses to “explicit” and since the tacit is “that which has not or cannot be made explicit” they provide four ways for things not to be tacit or four antonyms of “tacit” [TEK, 85]. One possibility would be to say that tacit knowledge is knowledge that is not explicit in any of these four ways. It seems

Philosophia Scientiæ, 17-3 | 2013 78

closer to the spirit of the book to say that we can understand different reasons to call things “tacit” by looking to these different antonyms and hence that there are different kinds of tacit status.

16 In fact, Collins seems to suggest that as well as there being different kinds of tacit knowledge there are also different degrees of tacitness. The first evidence of this is that he says that Relational, Somatic and Collective Tacit Knowledge are, respectively, “weak”, “medium” and “strong”: “these adjectives referring to the degree of resistance [...] to being made explicit” [TEK, 85, italics added]. But, second, he connects knowledge which has been made explicit in one sense with knowledge that is less (or “not quite so”) tacit as illustrated in the New Scientist article. In The Logic of Tacit Inference, Polanyi argues persuasively that humans do not know how they ride, but he also provides a formula: “In order to compensate for a given angle of imbalance α we must take a curve on the side of the imbalance, of which the radius (r) should be proportionate to the square of the velocity (v) over the imbalance r ∼ v² / α”. While no human can actually ride a bike using that formula, a robot, with much faster reactions, might. So that aspect of bike-riding is not quite so tacit after all. [Collins 2010b italics added]

17 The formula is part of an explanation of cycling and the idea of making a robot ride a bicycle looks to be an instance of mechanization. So such knowledge is made explicable and hence explicit in two of the four senses outlined and, by that fact, is, according to Collins, rendered less tacit.

18 Though the idea that tacitness admits of degrees is interesting in itself, the passage also suggests something more fundamental about the target of Collins’s analysis. The fact that humans cannot ride a bike by using the formula mentioned suggests that it is tacit. But the further fact that a robot with faster reactions might be programmed in accordance with the formula implies that it is not as tacit as it might otherwise have been thought to be (“not quite so tacit after all”). So the tacit status of human knowledge of how to ride a bike is affected—reduced—by facts about robot bike riding.

19 That is to say that, the fact that the task or process can be explained by others— whether or not they themselves have practical knowledge how to do it—counts against it being fully tacit for a different subject, however he or she thinks about or grasps riding a bike. There is a kind of “action at a distance” for the status of a human subject’s knowledge by facts about explication, for example by mechanization, elsewhere. This is surprising. Why should the question of whether a subject has tacit rather than explicit knowledge of how to carry out some task be affected by facts about whether some other agent or system could carry out the same task explicitly?

20 That it does, according to Collins’s analysis, follows from two features of the analysis. First, there are several antonyms of “tacit” in play giving rise to distinct kinds and gradations of tacitness. But, second and more fundamentally, it is an initial indication of the fact that the focus of Collins’s attention is not really knowledge, of a task or process, at all but the nature of the task or process itself. I will argue that this same view (from the New Scientist article) is present in Tacit and Explicit Knowledge and that it undermines the knowledge status of tacit knowledge. In the two-fold challenge of explaining how tacit knowledge can be both tacit and knowledge, Collins fails to account for it being knowledge.

21 To clarify my objection, it will help to contrast two claims Collins makes. He says, on the one hand, that:

Philosophia Scientiæ, 17-3 | 2013 79

[T]he idea of tacit knowledge only makes sense when it is in tension with explicit knowledge, and since cats and dogs and sieves and trees cannot be said to “know” any explicit knowledge, they shouldn’t be said to know any tacit knowledge either. In fact, they don’t “know” anything [...]. [TEK, 78]

22 The worry expressed in this quotation concerns limits on who or what can be a possessor of knowledge. Only those who can have explicit knowledge can properly be said to have tacit knowledge. But at the same time, Collins suggests, that one of the things that makes the very idea of tacit knowledge seem unduly mysterious is just the separation of such rational subjects from the rest of the animal, vegetable and mineral world. At the start of the book, he offers a kind of philosophical therapeutic diagnosis. Modernism in general, and the computer revolution in particular, has made explicit knowledge seem conceptually straightforward and tacit knowledge puzzling. But nearly the entire history of the universe [...] consists of things going on quite nicely without anyone telling anything to anything or anyone. [TEK, 7]

23 So whilst explicit knowledge presupposes tacit knowledge, the concept of tacit knowledge is parasitic on that of explicit knowledge. Thus it is helpful to his project of demystifying the concept of tacit knowledge to suggest the similarities between human and non-human cases. In all the ways that do not involve the way we intentionally choose to do certain acts and not others, and the way we choose to carry out those acts, the human, per individual body and brain [...] is continuous with the animal and physical world. We are just like complicated cats, dogs, trees, and sieves. [...] Sometimes we can do things better than cats, dogs, trees and sieves can do them, and sometimes worse. A sieve is generally better at sorting stones than a human (as a fridge is better at chilling water), a tree is certainly better at growing leaves, dogs are better at being affected by strings of smells, and cats are better at hunting small animals. [...] That teaching humans to accomplish even mimeomorphic actions is a complicated business, involving personal contact, says nothing about the nature of the knowledge, per se. [TEK, 104–105]

24 Aside from the fact that we can choose to do some things rather than others, and can choose to do them in particular ways, whilst cats, dogs, trees and sieves cannot, the performance of the tasks, which for us is expressive of tacit knowledge, is just the same. In that respect, we are just like those animals, plants and artefacts, according to Collins.

25 But whilst the therapeutic aim of this comparison is clear—inverting the recent assumption that tacit knowledge is more conceptually puzzling than explicit knowledge by emphasising that tacit knowledge is natural and longstanding—the details are less so. One problem is the point Collins himself makes: that cats and dogs, sieves and trees cannot be said to know anything, whether explicit or tacit. Intuitions may vary about whether cats and dogs can have knowledge but it is surely uncontentious that sieves and trees cannot. Given that fact, how can there be continuity between the tacit knowledge of human subjects and the dispositions of plants and artefacts?

26 A clue to how Collins addresses the apparent incompatibility—between the claims that cats, dogs, trees and sieves know nothing whilst the way they “do” things is just like the way we do things when we use and express our tacit knowledge—is his focus on what he calls (in the last quote) the “nature of the knowledge, per se”. This “nature of the knowledge, per se” does not, however, seem to mean the way humans know how to do the task, their knowledge, after all. Rather, it seems to mean the nature, not of the

Philosophia Scientiæ, 17-3 | 2013 80

knowledge, but of the task, or better the process, itself. The process or task can be a common element between humans and non-humans since the same process can be enacted, or brought about, or underpinned by human agency or non-human happening.

27 Let me illustrate. Consider one of Collins’s examples: the task or process of typing. This task can be carried out by skilled typists (who may intuitively be thought to have tacit knowledge of the location of the keys), beginners (who may have to scan the keyboard visually to find a key) or robots (which, intuitively have no knowledge at all, in virtue of lacking minds or rational faculties). One might expect that an analysis of the tacit knowledge of typing would have to distinguish between these cases and examine the way in which subjects think about, or knowingly grasp, the task at hand. (One possible analysis would say that the first group has tacit knowledge, the second explicit knowledge and the third no knowledge at all, merely instantiating a mechanism or process.) But because Collins wants to draw parallels between cats, dogs, trees and sieves and ourselves, his focus is on the mechanical process of typing not on the different ways that can be grasped.

28 Collins discusses this example in a chapter on one of the sub-species of tacit knowledge: Somatic Tacit Knowledge. In a discussion of Dreyfus, Collins notes that, for skilled typists, consciously following the rules they originally learned by slows them down. If his focus were on knowledge rather than process this would be important. But in fact he plays it down saying: this seems to bear on nothing but the way humans work; it does not bear on the way knowledge works. [TEK, 104]

29 “Knowledge” simpliciter does not denote the knowledge or know-how of human typists, then, but is rather a generalised account of the task or process of typing that could be given.

30 This assimilation (of the knowledge of a rational subject and a mere worldly process) is also suggested in a later comment on the limits of human typing: [W]e humans cannot generally type as fast or efficiently when we are paying attention to the keys but that’s just us. An automated typing machine that scanned print that was set out in a clear and undamaged font, transformed it into editable text, and then typed it out again could work as fast as any human typist, and faster if desired. [...] The constraints on the methods available for efficient typing by humans (by contrast e.g., with machines) are somatic limits; they have everything to do with us and nothing to do with the task as a task—nothing to do with knowledge as knowledge. [TEK, 104]

31 The last line highlights the assimilation of process or task and knowledge. Further, it shows why it is relevant when discussing the tacit knowledge of human typists to discuss the possible mechanization of that task. Without the assimilation of task and knowledge (including knowledge of how to do the task), such a comparison would be a non-sequitur because even if the process can be made explicit (for example, in the sense of put into words that Polanyi’s slogan suggests) that is not the source of skilled human typists’ abilities. They can type because they have tacit knowledge or know-how whether or not the same task or process could be mechanised by others. Since for Collins, mechanization makes things explicit, this is again a case where explanation and mechanization elsewhere seem to have a surprising “action at a distance” on the nature of the typists’ knowledge: it is less tacit whether or not they know that.

Philosophia Scientiæ, 17-3 | 2013 81

32 Collins warns that there is a danger in “mixing up the analysis of the way humans do things with the nature of knowledge itself” [TEK, 110]. But the problem is that by not paying attention to the way humans think of tasks, and for example the contrast between explicitly following rules about key location and having tacit knowledge of where the keys are, Collins focuses on a process which itself is not knowledge at all (though it could be the object of knowledge). Of course, a human typist can have knowledge of the process, of how to do it, and different typists may have different kinds of knowledge of this, but process and knowledge are different categories. It seems that in this discussion of somatic tacit knowledge, but also in his more general account of the long history of the working out of mechanical sequences, Collins has attempted to hold onto what is tacit but only at a cost of losing his grip on what is knowledge. He says, for example: That which is not explicit knowledge is mostly just the way the world unfolds. [TEK, 80]

33 Although he goes on to suggest that “mechanism” is a “more appropriate” label than “tacit knowledge” for the working out of mechanical sequences of greater or lesser complexity, he does not object that to call such worldly processes “knowledge” at all is an unusual anthropomorphism.

3 Signs and strings

34 Why does Collins assimilate the knowledge that subjects can possess of processes with the processes themselves? The reason seems to be one key choice of antonym for “tacit”. As I remarked earlier, Tacit and Explicit Knowledge aims to shed light on tacit knowledge by contrasting it with a suitable account of what is explicit. Collins’s initial characterisation of the explicit (which culminates in the four senses of elaboration, transformation, mechanization and explanation) is not, however, with what can be expressed linguistically—as Polanyi assumes—but rather with what he calls “strings”. “Explicit” is something to do with something being conveyed as a result of strings impacting with things. [TEK, 57]

35 Strings are: bits of stuff inscribed with patterns: they might be bits of air with patterns of sound waves, or bits of paper with writing, or bits of the seashore with marks made by waves, or patterns of mould, or almost anything [...]. [TEK, 9]

36 The motivation for this seems to be to avoid the “freight of inherent meaning that makes the notions of signs, symbols and icons so complicated” [TEK, 9]. One worry repeated in the book is that strings (and hence signs as kinds of strings) do not have an essential meaning: strings are without meaning [...] a book is a physical thing, not a meaningful thing. [TEK, 34]

37 So rather than starting with the meaning that can be expressed in words or signs he stresses instead the physical nature of strings: a string is just a physical object and it is immediately clear that whether it has any effect and what kind of effect this might be is entirely a matter of what happens to it. [TEK, 9]

38 The claim that neither strings nor signs have meanings essentially is surely correct. What kind of physical object or sign could compel a particular interpretation of it,

Philosophia Scientiæ, 17-3 | 2013 82

independently, that is, of a contingent background practice of sign use? What is less clear is why this might be thought to be such problem that it is better to invent a new ontological entity rather than contrast tacit knowledge with what can be put into words given a linguistic community. But before speculating on the reason for that, I will note four immediate consequences of discussing strings rather than linguistic signs not for Collins’s account of tacit knowledge but for his discussion of meaning and linguistic communication in Tacit and Explicit Knowledge.

39 First, without being able to rely on the idea that meaning can be expressed in the use of signs (for example, expressed in words in natural language), Collins talks of meanings as though they are independent of their means of expression: he reifies them. A language is a set of meanings located in a society, whereas, to repeat, strings are just physical objects. [TEK, 10]

40 The reification in turn leads to very frankly bizarre platonic sounding comments such as: Though strings are sometimes used to represent meanings, their relationship to meanings cannot be stabilised [...] because meaning is continually changing as it lives its life in society. [TEK, 44, italics added]

41 Second, as in that quote, it gives rise to the need to talk of using strings to represent meanings. This is a substantial philosophical commitment but one which receives little attention. How can physical items, bits of stuff inscribed with patterns, represent meanings unless they are used as signs in a linguistic community? What other account of representation of meaning is there?

42 Third, missing the possibility that, as part of a custom, a sign-post, for example, can simply mean turn left, he is forced to empty signs or strings of meaning: “There is no meaning in the book or the photograph” [TEK, 36]. But now with all (or as it turns out nearly all) the work done by human interpreters and none (or little) done by dead or empty signs there is the problem of explaining the fact that a sign-post does indicate a direction for a community. The stop-sign on roads leaves no room for doubt amongst regular drivers, for example. Collins admits to some embarrassment as he invokes the notion of affordance here: [I]nstead of saying “capable of being interpreted”, I will adopt the term “affords the interpretation”, which carries the implication that there is something in the string that makes it easier to interpret one way rather than another. [...] What “afford” does not mean is “determine”. [...] The terms “afford” and “affordance” are lazy terms [...] [which] merely paper over deep cracks in our understanding—or, or at least, my understanding—of why, given the extraordinary interpretative capabilities of humans, anything affords any one interpretation better than any other. How are meanings ever fixed, or even favoured? [TEK, 35–36]

43 Fourth, it gives rise to a distorted picture of linguistic communication (which I have abbreviated significantly to emphasise the key elements): Language translation or just plain conversation within one natural language consists of three stages. [...] Stage 1: inscription. In “telling” the attempt is made to represent lived meaning with the inscribed string. For example, in the case of conversation an attempt is made to represent the meaning as a string comprising vibrations in the air. [...] Stage 2: transmission and transformation. [...] Stage 3: interpretation. This is the attempt to recreate meaning from the string—to interpret it. [TEK, 27–28]

44 In this picture, Collins seems to subscribe to a seventeenth century view of communication in which meanings are encoded in mental items such as Lockean or

Philosophia Scientiæ, 17-3 | 2013 83

Humean ideas and inter-personal understanding involves an attempt to synchronise the same ideas in different minds through signs or sounds which are themselves lacking in meaning. Wittgenstein critically summarises this picture thus: It seems that there are certain definite mental processes bound up with the working of language, processes through which alone language can function. I mean the processes of understanding and meaning. The signs of our language seem dead without these mental processes; and it might seem that the only function of the signs is to induce such processes, and that these are the things we ought really to be interested in [...]. We are tempted to think that the action of language consists of two parts; an inorganic part, the handling of signs, and an organic part, which we may call understanding these signs, meaning them, interpreting them, thinking. [Wittgenstein 1958, 3]

45 Wittgenstein’s challenge (both here and in the Philosophical Investigations) is to ask the nature of the inner processes which are able to animate otherwise dead signs or to “recreate meaning from the [dead] string” and he finds no plausible solution. Collins does not address this worry.

46 Why does Collins reject the use of words or signs to express meaning in favour of the transformation of strings and their affordances? I can only speculate, but I suggest that the reason is his interpretation of what he calls Wittgenstein’s “rules regress” [TEK, 2, 46, 76]. Rules can never contain all the rules for their own application. [TEK, 46]

47 His interest in this goes back at least as far as Changing Order where he connects tacit knowledge with Wittgenstein’s discussion in §185 of the Philosophical Investigations of what understanding a mathematical series comprises. Collins considers the example of being asked to continue the “2, 4, 6, 8” sequence in the same way. “The immediate answer that springs to mind is ‘10, 12, 14, 16’ and, to all intents and purposes, this is indeed the ‘correct answer’” but he presses the question of how we know this [Collins 1985, 13]. It cannot, he argues, be a matter of following the rule “go on in the same way” because “this rule allows for a number of possibilities” [Collins 1985, 13]. Nor, assuming that that rule is merely insufficiently specific, does further codification that one sequentially adds 2 help because that might result in the continuation “82, 822, 8222 [...]” or other typographic variants each of which amounts to adding 2 in some sense.

48 He concludes both that the notion of “sameness” is ambiguous and that it is not possible fully to specify a rule (unless a limited range of responses is defined in advance). But “since in spite of this we all know the correct way to go on, there must be something more to a rule than its specifiability” [Collins 1985, 14]. The extra element is described in the introduction as “social entrenchment” or a “shared form of life”. Later, as in the quotation above, it is called “tacit knowledge”. Thus it is tacit knowledge that underpins the “mysterious abilities that enable us to know when to continue ‘2, 4, 6, 8’ with ‘10, 12, 14, 16’ and when with ‘who do we appreciate?’” [Collins 1985, 22, italics added].

49 This view of Collins suggests that whatever is put into words in explanations of meaning cannot go far enough. The gap between what is meant and what is actually explained has to be filled by something. It ignores what Wittgenstein himself goes on to stress in comments such as: “But do you really explain to the other person what you yourself understand? Don’t you get him to guess the essential thing? You give him examples,—but he has to

Philosophia Scientiæ, 17-3 | 2013 84

guess their drift, to guess your intention.” — Every explanation which I can give myself I give to him too. [Wittgenstein 1953, §210] “But this initial segment of a series obviously admitted of various interpretations (e.g. by means of algebraic expressions) and so you must first have chosen one such interpretation.” — Not at all. A doubt was possible in certain circumstances. But that is not to say that I did doubt, or even could doubt [...]. [Wittgenstein 1953, §213]

50 But putting aside the issue of the correct interpretation of Wittgenstein, Collins’s reading implies that signs can never simply express meaning (and hence the idea that books are not meaningful). I suspect that that is why, in setting out possible antonyms of “tacit”, he rejects signs since, on his view, they themselves cannot make anything explicit. Hence, I speculate, his turn to something more basic to make explicit knowledge explicit: strings. Explicit knowledge has substance—it is knowledge that can, to some extent, be transferred by the use of strings in the right circumstances. [TEK, 80]

51 There is a cost, however. Because strings are simply “bits of stuff inscribed with patterns” they are ubiquitous. The result is that there is no distinction between string transformations and mechanical causes and effects (“string transformation merges seamlessly into good old cause and effect” [TEK, 49]), whereas, by contrast, only some mechanical causes and effects have the right context to count as sign usage. But, since strings are used to underpin explicit knowledge, which is used as the antonym of tacit, this puts stress on the realm of tacit knowledge. For one thing, if mechanical cause and effect counts as string transformation and if string transformation is a form of explication and hence the explicit [TEK, 81], it is hard to see how anything is left for tacit knowledge.

52 There is a partial acknowledgement of this point in the form that if transformations of one pattern into another can be explained, for example, then that can no longer be a matter of, or for, fully tacit knowledge. [S]tring transformations and mechanical causes and effects are, to speak metaphysically, just two aspects of the same thing. This is why we have a strong sense that when we explain some process scientifically we have made it explicit; this is the “explicable” part of the antonym of tacit with its “scientifically explained” connotation. [TEK, 50]

53 This claim is the source of the “action at a distance” of explanation on the tacit status of a subject’s knowledge. But the deeper problem is that approaching what is explicit through the very broad notion of matter in patterns confuses what might be the object of knowledge—what a subject knows about—with the subject’s epistemic attitude to that.

Conclusion

54 This is how knowledge goes missing from Collins’s account of tacit knowledge. First, following Polanyi, tacit knowledge is contrasted with its standard antonym: explicit knowledge. But then the standard contrast is approached through the broader idea of strings rather than what can be expressed linguistically in words or signs. But since strings are just “bits of stuff inscribed with patterns” their elaboration, transformation, mechanization and explanation all count as instances of explication and hence what is explicit. Because he uses strings to explain the antonyms of “tacit” and strings are

Philosophia Scientiæ, 17-3 | 2013 85

merely “bits of stuff”, the focus of the discussion is not the nature of the knowledge that subjects possess but the processes or tasks or worldly patterns themselves.

55 Such a focus does not address the different ways in which such processes might be known by skilful or less skilful human agents (and possibly by animals) or replicated by non human artefacts or organisms. Collins’s broader account is of the nature of patterns or processes that might be known, not of the different ways in which they are known. It is pitched at the level of worldly patterns or ontology rather than the way they are known by epistemic subjects. Now that might be a perfectly fine area of inquiry. But it is not a discussion of tacit knowledge.[thorn1]

Acknowledgement

56 I am very grateful to an anonymous referee and to the special issue editors for very thorough feedback on two earlier versions of this paper.

BIBLIOGRAPHY

COLLINS, HARRY 1985 Changing Order: Replication and Induction in Scientific Practice, London: Sage, 1st ed. 2010a Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK]. 2010b Tacit knowledge: you don’t know how much you know, New Scientist, 2762, 30–31.

POLANYI, MICHAEL 1967 The Tacit Dimension, Chicago: University of Chicago Press, 1st ed.

WITTGENSTEIN, LUDWIG 1953 Philosophical Investigations, Oxford: Blackwell, translated by G.E.M. Anscombe. 1958 The Blue and Brown Books, Oxford: Blackwell.

ABSTRACTS

Harry Collins’s Tacit and Explicit Knowledge characterises tacit knowledge through a number of antonyms: explicit, explicable, and then explicable via elaboration, transformation, mechanization and explanation and, most fundamentally, what can be communicated via “strings”. But his account blurs the distinction between knowledge and what knowledge can be of and has a number of counter-intuitive consequences. This is the result of his adoption of strings themselves rather than the use of words or signs as the mark of what is explicit and, I suggest, it may stem from his earlier response to Wittgenstein’s rules regress.

L’ouvrage de Harry Collins Tacit and Explicit Knowledge caractérise la connaissance tacite à partir d’un certain nombre d’antonymes : explicite, explicitable qui est ensuite spécifié en explicitable par élaboration, par transformation, par mécanisation et par explication, et enfin, au niveau le plus fondamental, ce qui peut être communiqué via des « strings ». Mais son exposé brouille la

Philosophia Scientiæ, 17-3 | 2013 86

distinction entre connaissance et ce sur quoi peut porter la connaissance, et il aboutit à un certain nombre de conséquences contre-intuitives. Ceci est la conséquence de son adoption des strings, à la place de l’usage des mots ou des signes, en tant que marqueur de ce qui est explicite, et je suggère que ceci peut provenir de sa réponse antérieure à la régression des règles de Wittgenstein.

AUTHOR

TIM THORNTON School of Health, University of Central Lancashire, Preston (UK)

Philosophia Scientiæ, 17-3 | 2013 87

Collins’s Taxonomy of Tacit Knowledge: Critical Analyses and Possible Extensions

Léna Soler and Sjoerd D. Zwart

1 Aim and contents

1 The aim of this paper is to discuss from within the taxonomy of tacit knowledge proposed by Harry Collins in his 2010 book, Tacit and Explicit Knowledge (hereafter TEK) [Collins 2010b].

2 In TEK, Collins divides the “territory of tacit knowledge” in three areas: Relational (or weak) Tacit Knowledge (RTK); Somatic (or medium) Tacit Knowledge (CTK); and Collective (or strong) Tacit Knowledge (CTK). To give a first rough idea of the principle that underlies Collins’s taxonomy, as well as of the content of each of its three constitutive categories, we quote Collins: Collective Tacit Knowledge turns on the nature of the social, Somatic Tacit Knowledge turns on the nature of the body, but Relational Tacit Knowledge is just a matter of how particular people relate to each other. [TEK, 86]

3 In the book, RTK, STK and CTK are presented as three distinct domains (although most of the time mixed empirically and hence difficult to isolate one from the other); moreover, the three domains are supposed to exhaust the whole territory of tacit knowledge. The weak, medium and strong terminology refers to the degree of difficulty encountered in attempting to make the tacit knowledge K explicit.

4 Our discussion of Collins’s taxonomy starts with Relational (or weak) Tacit Knowledge; we aim to clarify its nature (section 2). In previous works, Collins distinguished different kinds of TK [Collins 2001]. He now considers these to be five sub-cases of RTK. We analyze these five sub-cases (section 2.1), and conclude that some of them are not adequately characterized as “relational”. Next (section 2.2), we examine further what the five sub-cases really have in common: what links them together and confers to Collins’s so-called “relational” TK its unity and identity. We propose the following

Philosophia Scientiæ, 17-3 | 2013 88

answer: for all sub-cases of the RTK category, the reason why the knowledge K under scrutiny remains tacit is contingent with respect to the nature of K. This answer, we claim, is the most adequate and fundamental feature of RTK. We conclude that it is desirable to rename Relational Tacit Knowledge. “Contingent” (or “accidental” or “circumstantial”) tacit knowledge would be better terminologies, but in order to maintain Collins already widely used initials, we suggest RTK should be read as: Relative-to-factors-contingent-regarding-the-nature-of-K Tacit Knowledge. To complete the clarification of RTK, we finally briefly discuss the relation between RTK and the social (section 2.3).

5 Having clarified the nature of RTK and what really defines the identity of the RTK category, we zoom out and consider Collins’s three TK categories altogether, aiming to investigate further the differences and possible relations between them and, ultimately, to deepen our understanding of the principle of Collins’s three-fold taxonomy (section 3). We begin to show (section 3.1) that, if we take Collins’s text literally, nothing prevents the generation of strange, if not contradictory mixed kinds of categories of TK, such as relational-somatic tacit knowledge (i.e., weak-and-medium TK) or relational-collective tacit knowledge (i.e., weak-and-strong TK). Analyzing further what has enabled the generation of such conundrums in the first place, we come to reveal an ambiguity in Collins’s classification and symbolic notations “RTK”, “STK” and “CTK” (section 3.2.1). There are actually two possible ways to interpret the first letter of RTK, STK and CTK: as reasons for tacitness, or as location of the (tacit) knowledge. To avoid the ambiguity, and to prevent the generation of the mixed, apparently self-contradictory categories, we must block certain, at first sight possible, interpretations. To indicate explicitly which of the two interpretations is involved, we introduce three conventions (section 3.2.2). By means of these conventions, we re-write and specify Collins’s categories and symbolic notations in a way that is less ambiguous and more informative than his own conventions (section 3.3.1). This leads to a refined and extended mapping of the territory of tacit knowledge at a given time t (see Table 1, section 3.3.3). But since our mapping introduces categories that are not considered in TEK (section 3.3.2), this raises open questions (sections 3.3.3. and 3.3.4), not directly addressed by Collins in TEK, with respect to which Collins’s position remains uncertain. The main one is the issue of the possibility of a partial explication of Collective Tacit Knowledge. Finally, in section 3.4, we consider the territory of tacit knowledge in a dynamic perspective. We examine how each category might evolve through time, and we discuss the possibility of a transformation of one category of TK into the other (see Table 2, section 3.4.3 for an overview).

2 Investigating Relational Tacit Knowledge

2.1 Questioning the fitness of the “relational” terminology

6 At the more general level, Collins gives the following characterization of RTK: Weak, or Relational, Tacit Knowledge (RTK) [...] is knowledge that could be made explicit in the second sense of explicable [explicable by transformation] but is not made explicit. [TEK, 86] A characteristic of weak tacit knowledge is that, in principle, with enough effort, any piece of it could be rendered explicit, [...even if...] not all of it can be rendered explicit at any one time. [TEK, 11]

Philosophia Scientiæ, 17-3 | 2013 89

7 This still says nothing about the reason why Collins calls this TK “Relational”. The “R” qualifies the kind of reason why a K that could have been made explicit is actually not made explicit but remains tacit. The main motivation of Collins for the choice of “relational” is expressed at [TEK, 86], as well as in several other passages:1 “Relational Tacit Knowledge is just a matter of how particular people relate to each other” (our italics). This quote seems to give the crucial reason for the choice of the term Relational Tacit Knowledge.

8 As particular sub-cases of RTK, Collins comes back to kinds of TK he has distinguished in his previous work [Collins 2001]: concealed knowledge, mismatched saliences, unrecognized knowledge, ostensive knowledge, and logistically demanding knowledge. We are going to argue that not all of these cases are adequately called “relational” in the previous sense of “how particular people relate to each other”. This is, of course, not just a “matter of words”, not just a “verbal issue”. What is at stake, here, is the very identity of the category: what unifies the different sub-cases constitutive of the category; what is in and what is out; what principle enables to individuate the RTK category as one category distinct from other categories such as STK and CTK.

2.1.1 Unproblematic cases: Concealed knowledge and Mismatched saliences

9 Two of the five sub-cases of Collins’s RTK are appropriately categorized as “relational” in the sense of “how particular people relate to each other”.

10 The first is concealed knowledge, such as for example secrets. This is clearly relational, because the fact that a K remains tacit or is explicated depends on the relation between two subjects (say A and B). Here the subjects might be identified with individuals or with groups of individuals: for example, individual A wants to hide K from B, because B is a scientific rival.

11 The second sub-case corresponds to mismatched saliences: K could be expressed but is not expressed because A does not realize that K needs to be revealed to B. Again this is clearly relational, because the fact that K is tacit or expressed depends on the relation between A and B. It depends on the way A conceives B and B’s knowledge, on what A presupposes about B, etc. For instance A does not express K because A believes that B already knows K.

2.1.2 Problematic case: unrecognized knowledge

12 One of Collins’s five RTK sub-cases is problematic with respect to the “relational” terminology. It is the case of unrecognized knowledge.

13 Unrecognized knowledge corresponds to the situation in which K is expressible but is not expressed by A because K is not recognized by A. For example, an experimenter A does not recognize K as an actual or relevant parameter of his experiment (see Collins’s introduction, section “Relational Tacit Knowledge”, p., for an example). A is not conscious of the role played by K, or even of the very presence of K. Here, the reason why K is not expressed by A is that A is not aware of the presence or importance of K. This has barely anything to do with the relation of A to anybody else. It is just a consequence of the fact that K is not expressed at all by A, even to himself, and cannot be expressed until K remains not conscious or not recognized for A.

Philosophia Scientiæ, 17-3 | 2013 90

14 One might try to object that if there had been no B at all, to whom A wanted to communicate his knowledge K, then A would not even have been thinking about trying to make his knowledge K explicit: so that the existence of some B to whom A wants to transmit something, and hence the existence of a certain kind of relation between A and some B, is involved here. This is true, but it is not an objection to our argument, since the existence of a relation between A and B called for by the objection, is only the motivation for A to try to make his knowledge explicit, but not the reason why part of the knowledge of A, namely K, is actually not expressed to B. Our conclusion reads therefore that the reason why an unrecognized K remains tacit has nothing to do with the way A and B relate to each other. Thus, in this case, the term “relational” is inadequate.

2.1.3 Two intermediate because mixed cases: ostensive and logistically demanding knowledge

15 The two remaining sub-cases of RTK, ostensive and logistically demanding knowledge, are intermediate cases, in the sense that the reasons why the K under scrutiny remains tacit are not reducible to relational reasons: relational reasons might play a role, but there are additional reasons of a different kind.

16 Take ostensive knowledge. A could express K to B (for example, where to find the spectrometer in the laboratory), but given what A assumes about B’s knowledge and capacities (B has just arrived in the laboratory), and given the contingent fact that B and A are in close proximity (B is, here and now, in the same room as A), A judges that it is simpler and more efficient to show K to B (i.e., to point the right direction or simply go with him) rather than to tell K to B. The same holds for logistically demanding knowledge: K could be made explicit, but it is not, because given his relations to B, and given the nature of the task involved and the available means, A estimates that it is more efficient to execute the task by tacit means.

17 The term “relational” only partially catches what is at stake here, because the reason why the expressible K is not expressed involves a mix of relational and non-relational factors: “the choice of which way to go—stick with the tacit or go to explicit” [TEK, 95], and hence the reason why K remains tacit, depends on some relational aspects, but also depends on other types of factors—such as the contingent fact that A and B are in the same room, the available technological means, the economic situation and so on— which, clearly, are not relational factors.

2.2 Suggestion of an alternative definition and denomination for RTK

18 Since the word “relational” and the associated definition in terms of “how particular people relate to each other” do not appropriately characterize all previous RTK sub- cases, we suggest an alternative name and an alternative definition. The five sub-cases are much better referred to as contingent, circumstantial or accidental tacit knowledge. Contingent (circumstantial or accidental) tacit knowledge is defined as follows: some K, which is perfectly explicable, is not explicated and remains tacit in a particular historical situation, because of reasons that could have been different historically and that are contingent with respect to the nature of K.

Philosophia Scientiæ, 17-3 | 2013 91

19 In this framework, Relational Tacit Knowledge becomes one particular case of contingent tacit knowledge. The way particular people relate to each other is one kind of circumstance or contingent factor with respect to the nature of K that is susceptible to prevent the expression of K in this or that historical situation. Since contingent- circumstantial tacit knowledge has the same acronym as CTK, we could favor “Accidental tacit knowledge” (ATK), but it is preferable to maintain Collins’s initials RTK. Thus, we propose to read the “R” of RTK as: “Relative-to-contingent-factors- regarding-the-nature-of-K” tacit knowledge.

20 Our alternative definition has several advantages. First and most importantly, it underlines what really constitutes the homogeneity of Collins’s category and justifies individuating it as one coherent category. If RTK is defined by means of the relation between people, not all K that could have been explicated but remains tacit would be included in RTK. In particular, we would have to exclude unrecognized knowledge. Moreover, our definition of RTK explicates what delineates the RTK category from Collins’s other kinds of TK categories. Contrary to Collins’s definition in terms of relational factors, our definition of RTK as “Relative-to-contingent-factors-regarding- the-nature-of-K” points at the central difference between RTK on the one hand, and Somatic/Collective TK on the other. RTK is tacit for reasons unrelated to the nature of K, whereas STK/CTK are tacit because of reasons that have to do with the nature of the K involved. This difference, which is acknowledged by Collins, will be developed in more details in sections 3.2. and 3.3.1.

21 Second, our definition fits better Collins’s own characterization of RTK in TEK. Consider just one example. Collins writes: Relational Tacit Knowledge is tacit because of the contingencies of human relationships, history, tradition and logistics. [TEK, 98]

22 Clearly, the “history, tradition and logistics” here explicitly mentioned by Collins just after “human relationships” exceed the “way particular people relate one to the other”. 2

23 Third, to read RTK as Relative-to-contingent-factors-regarding-the-nature-of-K Tacit Knowledge, clearly stresses a special feature of RTK compared to STK and CTK,3 which will prove to be important below, section 3.4. Namely, that RTK, contrary to STK and CTK, has an essentially contextual character: the fact that a given piece of knowledge K remains tacit here and now, whereas nothing in the nature of K prevents K to be made explicit, obviously depends on particular circumstances. So “any piece of RTK” [TEK, 11] should be associated with determined spatio-temporal coordinates (x, t), and the tacitness of K involved in this “piece of RTK” depends on the details of the particular situation corresponding to (x, t). We can nevertheless also consider RTK as a category independent of any determined context in a given stage of the human history, as Collins does when he talks about “all of RTK” (cf. for example when he writes: “not all of it [RTK] can be rendered explicit at any one time”). In the latter quote, “all of RTK” refers to the totality of knowledge that has been articulated in a given stage of the human history but that could, in this or that particular situation, remain tacit for contingent reasons. To the extent that “all of RTK” refers to the totality of knowledge, “all of RTK” is a global category. To the extent that it refers to knowledge that has been articulated in a given stage of the human history, it is a universal category, in the sense of being explicable by any competent person at the period under scrutiny.4 So RTK can be considered according to two perspectives: either in a contextual perspective, and then,

Philosophia Scientiæ, 17-3 | 2013 92

RTK refers to a piece of RTK associated to a given “here and now” (including some particular agents); or in a global-universal perspective, and then, RTK refers to “all possible pieces of RTK”, that is, to the totality of the available explicable knowledge at a given period, imagined as not explicated due to contingent contextual factors. In other words, the extension, in terms of knowledge, of global-universal RTK at a given period, coincides with the extension of “all explicit knowledge”, if the adjective explicit is to be understood—as it is often the case in Collins’s writings—in the universal sense of “explicable by some competent person” at the period under scrutiny. Fourth, our alternative avoids the risk of confusion between RTK and CTK. Since “relational” implies at least two persons and possibly more, “relational” might suggest a collective. To that extent, RTK and CTK might be confused, or at least, the borderline between the two might be blurred. This is relevant since Collins sometimes explains RTK by means of the vocabulary of the “social”. To help to avoid confusions and to understand better the difference between RTK and CTK, we are now going to investigate the relation between RTK and the social.

2.3 Relations between RTK and the social

24 When Collins explains RTK by means of the “social”, he contrasts RTK with CTK by stating that RTK has to do with “the way society organizes itself”, whereas CTK has to do with “the intrinsic nature of the social”. The way human societies work plays a role in our understanding of the irreducibility of the tacit [of RTK]. In society as we know it there will always be secrets, mismatched saliences, and things that are unknown but may be about to become known. This fact has to do with the way society organizes itself rather than having anything to do with the intrinsic nature of the social. [TEK, 98]

25 CTK is tacit in virtue of “the ‘ontology’ of the knowledge” [TEK, 96] involved, namely in virtue of the collective nature of this knowledge. It has to do with the “nature of human society” [TEK, 86, our italics], and “with the way society is constituted” [TEK, 85]. In brief: CTK is tacit because K has an essentially social or collective character (in TEK, Collins uses “social” and “collective” interchangeably).5

26 In contrast, RTK is not tacit in virtue of the nature of the knowledge involved. It is tacit for reasons that are contingent with respect to this nature. This is what we mentioned above, section 2.1, and it is a strong reason to re-define the “R” in RTK as “Relative-to- contingent-factors-regarding-the-nature-of-K”. Thus, the reference to the “social” in relation to RTK must be understood as an extrinsic feature with respect to the knowledge involved: it is not because the knowledge is social in nature that it is tacit. It is just because of the contingent way this or that society organizes itself.

27 Consequently, the reason for explicable knowledge to remain tacit in a particular situation may have to do with the social, but it has only to do with “extrinsic characteristics” of the social, which Collins refers to as “the way society organizes itself”: RTK remains tacit because of “the contingencies of social life” [TEK, x]. This characterization obviously raises the question: What is the “intrinsic nature” of the social?

28 We are tempted to question the adequacy of Collins’s reference to the “social organization”, as an explanation of the reason why knowledge which could be explicated remains tacit. The main reason reads as follows. The social organization is not

Philosophia Scientiæ, 17-3 | 2013 93

crucial for RTK. In particular, the specific nature of the relations between different human beings—which is a pivotal ingredient of Collins’s characterization of RTK and Collins’s primary motivation for the terminology of the relational—is certainly not reducible to social organization. The singular personality and the singular relations of the individuals certainly matter, even for a “collectivist” as Collins. When Collins defines the “R” of RTK as “how particular people relate to each other” [TEK, 86, our italics], the adjective “particular” seems to point to this singularity. This reflection shows the need for a clarification of the role of the individuals in Collins’s collectivist picture. Clearly, the social plays the primary role in Collins’s intuitions and readings of the reality under study. But the relations between the social level and the individual level remain unclear.

3 Clarifying the principle of Collins’s three-fold TK classification

29 After having come to an alternative definition of RTK as “Relative-to-contingent- factors-regarding-the-nature-of-K”, we are now going to investigate the relations between Relational Tacit Knowledge on the one hand, and Somatic and Collective Tacit Knowledge on the other. We think the principle of Collins’s three-fold classification needs clarification. Collins’s text leaves room for two different readings, thus creating an important ambiguity. We experienced this ambiguity in the course of our reflection on Collins’s book, since we successively endorsed two significantly different understandings of his classification. We finally retained what we think is the most charitable interpretation, which we assume to be Collins’s position.

3.1 Revealing an ambiguity in Collins’s classification

30 First, remember the interpretation of RTK we have adopted above: RTK is independent of the nature of the knowledge involved; the R of Relational TK means: not explained because of reasons that are contingent with respect to the nature of K.

31 Second, let us turn to the Somatic and Collective Tacit Knowledge, STK and CTK, respectively. Relying on Collins’s text, STK and CTK have the following properties: • a) STK is knowledge located in the body (bodily skills, etc., hence the “S” of Somatic) and CTK is knowledge located in the collectivity (social skills, etc., hence the “C” of Collective) • b) both STK and CTK are not explicated (hence the “T”).

32 Let us start with Somatic Tacit Knowledge. One might reason as follows. Part of STK as just understood is indeed, according to Collins, explicable nowadays, in the actual stage of scientific development. This is for example the case for parts of the somatic knowledge associated with bike riding: Collins explicates parts of this knowledge in TEK (see for example [TEK, 101, box 11], where some STK associated with bike riding is explicated in sense 4: a scientific explanation of bike balancing is given). This part of STK is explicable, hence it could be explicated, but it happens that in some particular situation, it is not explicated and remains tacit. Consequently, this part of STK has not been explicated for contingent, circumstantial reasons (reasons accidental with respect to the nature of the K involved), or, in Collins’s terminology, for “relational” reasons. Hence it should be referred to as “Relational (Somatic Tacit Knowledge)”, R (STK).

Philosophia Scientiæ, 17-3 | 2013 94

33 The same way of reasoning applies to Collective Tacit Knowledge, which would lead to R (CTK), provided we accept the following assumption: parts of the collective knowledge acquired in the actual stage of scientific development is explicable, exactly as this is the case for parts of the somatic knowledge. It is unclear whether Collins endorses this assumption. In the sequel (section 3.3.3) we will return to this issue and will discuss compelling reasons for accepting it. Meanwhile let us, for the sake of the argument, suppose that Collins endorses the assumption and that he treats STK and CTK symmetrically regarding their partial explicability.

34 At first sight, the idea of Relational (or contingent) Somatic/Collective Tacit Knowledge —which is not considered in TEK—appears problematic, if not paradoxical. It is problematic since Collins’s categories are supposed to be mutually exclusive and collectively exhaustive. The problem turns to a paradox if we substitute the “R”, “S” and the “C” with Collins’s alternative denominations “weak”, “medium” and “strong”.R(STK) would result in a weak-and-medium tacit knowledge, and R(CTK) in a weak-and-strongtacit knowledge.

3.2 Dissolving the ambiguity

3.2.1 Reasons for tacitness and locations of TK

35 To get rid of the apparent paradox just sketched and to dissolve the ambiguity which allowed for its generation, we must carefully distinguish between two possible interpretations of the first letter of the three categories in Collins’s taxonomy. These two meanings are not entirely independent to one another, but they are not reducible to one another either.

36 (1) The “R”, “S” and “C” may refer to kinds of “causes” or “reasons” that stop the knowledge being explicated (i.e., to kinds of reasons why K is tacit). This understanding is in accordance with Collins’s explicit general statements about the principles of his categorization at the most fundamental level: In this book, tacit knowledge will be analyzed and classified by reference to what stops it being explicated; there are three major reasons why tacit knowledge is not explicated; therefore, there are three major types of tacit knowledge. [TEK, 1]

37 (2) The “R”, “S” and “C” may refer to kinds of K by specifying the location of K. • (2a) As far as the somatic and the collective are concerned, this second understanding is, like the first one, also in accordance with Collins’s explicit general statements about the principles of his taxonomy at the most fundamental level. In this vein, STK is characterized as “knowledge stored in the muscles, nerve pathways, and synaptic connections” [Collins 2010a], “knowledge embodied in the human body and brain” [TEK, 2], or knowledge that has to do with “the nature of the human body and brain” [TEK, x, our italics]. In a similar vein, CTK is described as “the domain of knowledge that is located in society” [TEK, 85], “knowledge “embodied” in society” [TEK, 2], knowledge that is “a property of society rather than the individual” [TEK, 11], or knowledge that “turns on the nature of the social” [TEK, 86]. In other words, “‘somatic’ and ‘collective’ qualify the ‘ontology’ of the knowledge” [TEK, 96]. They define the fundamental nature of the knowledge involved, and for Collins, to define the nature is to specify the “location” of this knowledge, such as the body or the society. • (2b) But what about Relational Tacit Knowledge as a location? What would the second meaning imply for the “R” in RTK? Can we interpret the “R” of RTK as a location, and if yes, what would be this location?

Philosophia Scientiæ, 17-3 | 2013 95

Relying on the definition of RTK as “a matter of how particular people relate to each other”, we could be tempted, for reasons of symmetry, to locate the knowledge involved in RTK in the relation between people or groups of people, which would leave us with the task of explaining how knowledge can be located in a relation between people. But according to the understanding of RTK as we have proposed section 2,6 this cannot be the correct interpretation: RTK cannot be interpreted as a location at all. As explained before, RTK is independent of the nature of the knowledge K involved. Consequently, the location of the K in “RTK” is completely open: it might be anything, and the “R” of RTK does not indicate any determined location. This is an additional reason to prefer the name “circumstantial tacit knowledge”—or still better in this respect, “contingent tacit knowledge”— to the expression “Relational Tacit Knowledge”, since “contingent” does not suggest any location, whereas “relational” might do so. We conclude that for the “R” of Collins’s Relational Tacit Knowledge, only one of the two meanings (reason and location) applies: The R should be understood as a reason.

38 To sum up: the somatic “S” and the collective “C” can refer either to reasons or to locations, but the relational “R” can only refer to reasons. This shows an asymmetry between STK and CTK on the one hand, and RTK on the other.

3.2.2 Dissolution of the paradox and notations to avoid the ambiguities

A. A convention for specifying the reasons of tacitness

39 The previous discussion enables us to dissolve the apparent paradox related to R(STK) (or weak- and-medium TK) and R(CTK) (or weak-and-strong TK).

40 As soon as we distinguish between the two interpretations of “S” and “C” as reason and as location, and as soon as we realize that these two meanings are not equivalent, it becomes evident that: • (a) the “S” and the “C” in R(STK) and R(CTK) should be read only as two locations and not as two reasons, and • (b) the “R” should be interpreted according to its only possible reading, viz. as a reason.

41 The expressions R(STK) and R(CTK) only appear paradoxical if the “S” and the “C” are interpreted both as locations and as reasons. Interpreting “S” only as a somatic location and “C” only as a collective location prevents any substitution of “S” by “medium” and “C” by “strong”. This is because “medium” and “strong” unambiguously refer to the force of the resistance with respect to attempts of explication. Hence “medium” and “strong” cannot be understood as two locations. Consequently, the generation of paradoxical expressions such as weak-and-medium TK or a weak-and-strong TK is avoided.

42 To preclude the generation of the expressions R(STK) and R(CTK), we must block one of the two meanings of the “S” and the “C” that appear in “STK” and “CTK”. More exactly, we must prevent the “S” and the “C” to be read as location only; the “S” of STK and the “C” of “CTK” must be read as reasons. To specify explicitly the right kind of reading, and to prevent confusions between reasons and locations, let us introduce a first convention. To indicate explicitly that we understand the letter which points to a kind of TK as a reason, we will associate the subscript “r” to this letter.

43 Convention 1: If R, S or C are used as reasons for the tacitness, we write: Rr, Sr and Cr

Philosophia Scientiæ, 17-3 | 2013 96

44 Using convention 1, we translate our claim, according to which S and C should not be read as location only, as follows:

STK = SrTK, which means that the “S” in STK must be understood as the reason why K is tacit and that this reason has to do with the somatic nature (and hence with the location in the body) of K.

45 Similarly,

CTK = CrTK, which means that the “C” in CTK must be understood as the reason why K is tacit and that this reason has to do with the collective nature (and hence with the location in the society) of K.

46 Finally,

RTK = RrTK, which means that the “R” in RTK must be understood as the reason why K is tacit, and that this reason has to do with factors that are contingent with respect to the nature of K, such as, for example, some particularities of the relation between people.

47 In short: “Sr”/“Cr” indicate that something in the somatic/collective nature of K creates

a specific difficulty with respect to the task of explicating K; “Rr” indicates that something in the contingent circumstances of the historical situation—something which has no essential relation to the nature of the K involved—makes K to remain tacit.

B. A convention for specifying the location of the knowledge

48 As mentioned at the beginning of section 3.2.2, the two meanings of S/C—i.e., S/C as reasons and S/C as locations—, although not being reducible to one another, are not

independent. Obviously, S/C as reason (i.e., according to convention 1, Sr/Cr) implies S/C as location. If the reason why K is tacit “has to do with” the fact that K is some bodily/ social K (some bodily/social skill, knowing-that, commitment, etc.), then necessarily, the location of this K is determined and corresponds to S/C. But of course, the reverse does not hold: S/C as location does not necessarily imply S/C as reason: the reason why a K is not explicated might have nothing to do with the location of this K.

49 To have a means to specify explicitly, and to represent schematically, the cases in which the reason for K not being explicated and the location of K are different, we need two different notations, one to indicate the location, and another one to indicate the reason. The reason for tacitness is indicated by means of our convention 1. To indicate the location, we introduce a second convention, which associates to K an index that specifies the location of K (i.e., the kind of K involved).

50 Convention 2: If K is of the kind N (or equivalently: if K is located in N), we write: KN

51 Using convention 2, KS indicates that K is located in the body, that K has a somatic

nature. Similarly, KC indicates that K is located in the society, that K has a collective character.

Philosophia Scientiæ, 17-3 | 2013 97

C. A convention for distinguishing the tacit as not explicated and the tacit as not explicable

52 We still need a third last convention to fully clarify Collins’s RTK, STK and CTK categories. The usefulness of this convention will become clearer in sections 3.3 and 3.4. This convention provides a concise explicit means to refer to an important distinction which is already present and elaborated in TEK: the difference between tacit-as- not-explicable and tacit-as-not-explicated.

53 In TEK, this distinction enters in the very definition of the tacit: The tacit is that which has not or cannot be made explicit. [TEK, 85]

54 Collins points out that, in the literature, the neglect of this difference has often been a source of confusion. Moreover, he introduces eight different types of “cannot” [TEK, 90] and four different meanings of “explicable” [TEK, 81], to which he relates each of his three kinds of TK. But despite the welcome introduction of all these clarifying fine- grained distinctions in TEK, Collins’s three main notations, “RTK”, “STK” and “CTK”, remain unspecified with respect to the important difference between “is not” and “cannot”. To make such a specification explicit is the aim of our convention 3.

55 Convention 3: Ted means tacit as not-explicated, and Table means tacit as not-explicable

56 Following convention 3, TedK means that K is not explicated, and TableK that K is not explicable.7

57 Of course, Table logically implies Ted: if K is not explicable, a fortiori it is not explicated.

3.3 Specification of Collins’s classification by means of our distinctions and notations

3.3.1 Specification of Collins’s three tacit knowledge categories

58 Combining our distinctions and conventions, and applying them to Collins’s three categories of tacit knowledge, we come to the following interpretation and symbolic expressions, which contain more information and are less ambiguous than Collins’s ones:

STK should be specified as: SrTableKS. 59 This means: K is inexplicable, and hence not explicated, in the present stage of our knowledge, for reasons that have to do with the somatic nature of K. The fact that K is located in the body creates special difficulties with respect to the task of explicating K.

CTK should be specified as: CrTableKC. 60 This must be read as: K is inexplicable, and hence not explicated, in the present stage of our knowledge, for reasons that have to do with the social nature of K. The fact that K is located in the society creates special difficulties with respect to the task of explicating K.

61 One might object that the two subscripts “s” and “c” in SrTableKS and CrTableKC are

redundant, since SrTK implies KS and CrTK implies KC. This is true, of course. We have two reasons, however, to keep the K-indices. First, the redundancy is obvious only at the end of our analysis, after having distinguished S/C as reason and S/C as location, and after having examined the logical relations of these two meanings. Second and much more important, the redundancy does not apply to Relational TK: as we have seen,

Philosophia Scientiæ, 17-3 | 2013 98

the “R” must be interpreted as a reason (Rr), but this does not imply any particular location, and therefore any definite index for K. So taking into account the complete taxonomy of tacit knowledge, the two indexes—the index added to the first letter of each acronym and the index added to the K involved in the same acronym—are not redundant. Hence for each category, we need to specify in two different ways the reason for tacitness and the location of the tacit K. The reason/location distinction must be explicit in all the symbolic expressions used, despite some redundancies in certain categories of TK.

62 The previous remarks related to the difference between STK/CTK on the one hand, and RTK on the other hand, become evident when we re-express RTK using our three conventions:

RTK should be specified as: RrTedK 63 The latter means: a piece of knowledge K that is explicable in the present stage of our knowledge, is not explicated in one or another particular situation, because of reasons contingent with respect to the location of K (i.e., reasons that have nothing to do with the nature of K).

64 Our conventions make explicit and clearer the following similarities and differences between Collins’s three TK categories: • The “R”, “S” and “C” of Collins’s RTK, STK and CTK must always be read as reasons as why K is not explicated in some particular situation. • Whereas the tacit in RTK only points at local contingencies explaining why K is not- explicated in some particular situation, the tacit in STK/CTK points at stronger reasons, viz., reasons as to why K is not explicable in the present stage of our knowledge. For STK/CTK, in contrast to RTK, the fact that K is not explicated is a consequence of the fact that K is not explicable. • The reason why the “K” in STK/CTK is not explicable is its location, namely the body and the society, respectively: The “S” in STK and the “C” in CTK refer to two kinds of reasons as to why K is inexplicable at the present time, and these reasons are the location in the somatic and the location in the collectivity. By contrast, the “R” in RTK does not refer to any determined location.

3.3.2 Three conceivable sub-cases of RTK

65 As the “K” involved in RTK is not associated with any specific location, the corresponding K might have different locations. In particular, nothing seems to prevent this “K” to be located in the body or in the society. Consequently, at first sight, we have three conceivable sub-cases of RTK, not considered explicitly by Collins:

• RrTedKS, which abbreviates: some explicable knowledge K located in the body (an explicable

somatic resource KS) is not explicated (Ted) for reasons that have nothing to do with the somatic nature of K (but for reasons Relative-to-contingent-factors-regarding-the-nature-of-

K: Rr). This corresponds to R(STK) introduced in section 3.1.

• RrTedKC, which abbreviates: some explicable knowledge K located in the society (an explicable

social resource KC) is not explicated (Ted) for reasons that have nothing to do with the collective nature of K (but for reasons Relative-to-contingent-factors-regarding-the-nature-

of-K: Rr). This corresponds to R(CTK) introduced in section 3.1.

• RrTedKN, where N refers to any kind of K which is neither somatic, nor collective—if such exists.

Philosophia Scientiæ, 17-3 | 2013 99

3.3.3 The Refined map of the territory of tacit knowledge at a given time t

66 Table 1 maps the territory of TK at a given stage of knowledge (say at t), using the three conventions introduced above.

Table 1: The territory of TK in a given stage of knowledge at t

Explicable Not explicable

Explicated Not explicated Location of knowledge: RTK(t) STK(t) Somatic Explicated K (t) = R T K (t) S r ed S = SrTableKs(t)

Not explicated CTK(t) Location of knowledge: Explicated Explicated K (t) RTK(t) = Collective C = RrTedKC(t) CrTableKC(t)

Location of knowledge: RTK(t) Explicated Explicated KN(t) N (neither somatic, nor collective) = RrTedKN(t)

67 Table 1 gives a static representation of explicit and tacit knowledge at time t. In section 4, the situation will be considered in a dynamical perspective. But before that, let us comment shortly on what might be Collins’s position with respect to some constituents of the static table.

68 A first question arises in relation to KN. Are there other locations than the somatic and collective locations involved in Collins’s characterization? In common ways of thinking, if something X is neither somatic nor collective/social, this strongly suggests that X is intellectual and individual. What about the idea of intellectual individual tacit knowledge? It makes sense intuitively, and it seems to be missing in Collins mapping of TK. Take for instance the elaboration of a new mathematical method by an individual mathematician, and suppose that the corresponding know-how involved has not been explicated. Would this know-how not be a case of an individual, intellectual tacit-as- not-explicated knowledge? More generally (leaving the intellectual dimension aside), what about individual tacit knowledge in Collins’s framework? Should we not categorize as such the (intellectual and/or somatic) tacit knowledge involved in the pioneering contributions of individual scientists? Is there a place for an individual “contributory expertise”?

69 Let us turn to RrTedKS. As already stressed in section 3.1, Collins should be ready to allow

for RrTedKS, since some of his examples are of this kind. “Something called ‘the rules of bike [...] balancing’ can be told in sense 4 of explanation” (scientific explanation), writes Collins in [TEK, 100]: Furthermore, these rules, or some set of rules, could, in principle, be programmed into a mechanical bike balancer with elaborate feedback circuits—so they are also explicable in sense 3 (explicable as mechanization, i.e., we can build “mechanical bike balancers”). According to these quotations, some STK is clearly explicable. Hence according to the

Philosophia Scientiæ, 17-3 | 2013 100

particular circumstances, this STK could be explicated or not. Collins provides a (fictional) situation which may serve as an illustration. If we rode our bikes on the surface of a small asteroid with almost zero gravity so everything happened much slower, we ourselves could probably use Polanyi’s rules [i.e., the scientific explication of bike balancing] to balance. [TEK, 100]

70 In such a situation, a bike teacher could choose either to tell the scientific rules to the novice, or to use other ways in which the somatic knowledge required would remain tacit (for example, just show him).

71 The more difficult question, that we are going to address in the next section, is whether

Collins is ready to allow for RrTedKC in his framework.

3.3.4 Is Collective Tacit Knowledge partly explicable?

72 One reason to doubt that Collins would allow for RrTedKC is his insistence that the collective is special and fundamentally different from other kinds of realities. Applied to the reasons as to why knowledge is tacit, the collective character of K may introduce an extremely strong resistance of a unique kind which would hamper any explication. Nevertheless, Collins dedicates many pages of TEK to explicate what it is for a westerner to drive in Beijing in contrast to drive in other countries more familiar to her. He expresses many characteristics about the knowledge that must be mastered to enable one to drive in Beijing, and his explanations are perfectly understandable. Thus, Collins explicates parts of the ordinarily-tacit resources used by Chinese or non Chinese people who are able to feel at home and drive competently in Beijing’s traffic. Since these resources seem to be both collective/social in an essential sense, and since they are de facto explicated by Collins, it apparently follows that Collins makes room for 8 RrTedKC, and even provides instances of this category.

73 It should be stressed, however, that according to Collins, such explications are, and will remain in any foreseeable future, partial and insufficient. Partial because a substantial part of the social skills needed to drive in Beijing is still inexplicable, today and in any foreseeable future; and insufficient because it is this inexplicable part which is responsible for “social fluency”, i.e., for the specifically human ability to act flexibly

and relevantly in ever-changing social contexts. Hence, the explicated KC will never be enough to build machines that act as humans in this sense. Still, if our interpretation of

Collins’s Beijing driving writings is correct, part of the Collective Tacit Knowledge KC located in the Beijing society has been explicated, and therefore part of this collective

knowledge KC is explicable. 74 Does our interpretation correspond to Collins’s position about CTK? We wait for a clarification of this point, but it seems difficult to deny the conclusion that part of the

collective knowledge KC of a society is explicable. Refusing this conclusion would entail, for example, rejecting claims such as “our scientific paradigms are in part explicable”. Yet, part of a scientific paradigm—the propositional part—is explicit in an uncontroversial sense. For example we find, in handbooks, the expressions of the laws of the paradigm under scrutiny. Such scientific expressions are usually taken as the very prototype of explicit knowledge. If this is not explicit knowledge, then what is? Providing moreover that scientific paradigms are collective—something we assume Collins would not deny—, it seems that we have here a prototypical example of a partial explication of collective knowledge. Obviously, the previous discussion depends on the conceptions we endorse

Philosophia Scientiæ, 17-3 | 2013 101

of the collective/social, and on an analysis of what is collective/social and what is not. In TEK, Collins does not elaborate these matters. He repeatedly stresses the mysteriousness and the shallowness of our understanding of the socialization process; he makes the CTK acquired through a socialization process responsible for the human ability to act flexibly and relevantly in varying social contexts; but his analysis mainly relies on an intuitive, tacit understanding of what is the collective/social.

75 Whatever Collins thinks about RrTedKC, the distinctions and conventions introduced above enable us to express the difference between two conceivable kinds of Collective Tacit Knowledge: on the one hand, a Collective Tacit Knowledge which is not explicable because of the collective nature of the knowledge involved (Collins’s CTK, specified as

our CrTableKC); on the other hand, a Collective Tacit Knowledge which is not explicated for accidental reasons—for reasons that have nothing to do with the nature of the

knowledge involved (our RrTedKC, not considered in TEK).

3.4 The dynamics of RTK, STK and CTK

76 In this final section, we explore the temporal evolution of Relational, Somatic and Collective Tacit Knowledge. To do so, we need to be able to refer to the difference between today and the future. To that end, we let t refer to today in the sense of “nowadays”, and t′ refer to some later time in the future. A systematic representation of our conclusions about the dynamics of RTK, STK and CTK is given in Table 2.

3.4.1 The temporal evolution of STK and its transformation into RTK

77 Let us first consider STK. According to Collins, at least some knowledge K located in the body and not explicable at a given time t for reasons that have to do with the somatic

nature of K (i.e., SrTableKS(t)), can become explicable at t′ due to scientific progress. In

other words, some SrTableKS(t) can become somatic explicit knowledge, KS(t′), in the

future. This in-the-future-explicable KS(t′) might be explicable in different senses (at least in the two senses of scientific explanation and mechanization, see section 3.3.2).

78 Suppose now that some at-t-inexplicable somatic knowledge, SrTableKS(t), becomes at-t′- explicable but remains, at t′ in some particular situation, not explicated for reasons that are contingent with respect to the somatic location of K. According to our conventions, this

configuration corresponds to RrTedKS(t′). Hence, when time progresses from t to t′, part

of SrTableKS(t) can be transformed into RrTedKS(t′). In Table 2, this transformation is represented by the dark arrow. This transformation implies that, “all other things being equal”, in the course of time, STK diminishes and RTK increases in the same proportion.9 In the last sentence, the ceteris paribus clause means that: (a) from t to t′, no new somatic knowledge, either tacit or explicit, has been produced; (b) all somatic knowledge possessed at t, either tacit or explicit, is conserved from t to t′—i.e., no somatic knowledge is lost through time. Of course these two conditions are unrealistic, but we use them for the sake of analytical clarification. Hereafter, we will refer to them as the “no-production and no-losses” clause.

79 Let us now consider what might happen at the “ideal end of scientific research” (say at

tend). Suppose that at tend, all somatic previously-tacit knowledge has been explicated and is therefore explicable (at least in the sense of scientific explanation or

mechanization). In our notations: SrTableKS(tend) = ∅. This is not impossible according to

Philosophia Scientiæ, 17-3 | 2013 102

10 Collins. In this hypothesis, at tend, all the tacit knowledge located in the body would be tacit only for contingent reasons, and not for reasons essentially related to the fact that

this knowledge is located in the body; all the previous SrTableKS would be transformed

into RrTedKS.

3.4.2 The temporal evolution of CTK and its relation to RTK

80 Let us turn now to CTK and its evolution through time. The difference, between the STK case discussed above and the CTK case, is that in the CTK case, Collins’s position about the possibility of a partial explication of CTK is unclear (see section 3.3.3). Two interpretations of Collins’s position are possible: either (i) CTK is not at all explicable, even partially, and it is so for any foreseeable future; or (ii) parts of CTK are already explicable nowadays, and others parts could be explicated further in the future. Let us examine the dynamical behavior of CTK with regard to each of these two possible interpretations.

• (i) If CTK is fully inexplicable, today and in any foreseeable future, then no part of CrTableKC(t)

can be transformed into RrTedKC(t′). This means that in Table 2, the light arrow with the

question marks, which represents the transformation of CrTableKC(t) into RrTedKC(t′) from t to t′, does not correspond to any real process and must be suppressed. Under the “no- production and no-losses” clause, CTK would remain the same through time: CTK(t) =

CTK(t′) = CTK(t″) = … CTK(tend). • (ii) If at least parts of CTK are possibly explicable, then similar reasoning to that we have

applied to the evolution of STK in section 3.4.1 applies to CTK. Parts of CrTableKC(t) can become

explicable at t′, and if a piece of at-t′-explicable-KC remains, at t′, tacit for accidental reasons —reasons that are contingent with respect to the collective location of K—, this corresponds

to a piece of RrTedKC(t′). Hence, just as with STK, in the course of time CrTableKC(t) is

transformed into RrTedKC(t′). Under the clause of no CTK production or losses, CTK diminishes over time and is progressively transformed in RTK as long as more and more parts of the inexplicable content of CTK are made explicit. In Table 2, the light arrow represents an existing process and must be maintained.

3.4.3 Comparing the dynamical behaviors of STK and CTK

81 Besides the similarity just sketched, the dynamics of STK and of CTK differ in an important other respect. The difference concerns the asymptotic behavior of the temporal evolution of the two kinds of knowledge.

Philosophia Scientiæ, 17-3 | 2013 103

Table 2: Dynamical behaviors of the different kinds of TK

82 As mentioned in section 3.4.1, Collins’s writings suggest that at tend, all STK might be fully explicated at least in the sense of “scientific explanations”. If this were the case,

STK would vanish in the end: SrTableKS(tend) = ∅. In contrast, Collins clearly denies that CTK will ever be fully reduced in this manner: Collins’s writings stress that CTK will not be explicable in any foreseeable future. This suggests that even if Collins was ready to accept the possibility of a partial explication of CTK at t and an increase of this explicated part in time, he would nevertheless maintain that an essential part of our collective knowledge would still remain tacit at the hypothetical “ideal end of

research” tend: CTK(tend) = CrTableKC(tend) ≠ ∅. We would never reach a stage in which CTK

would be empty. Whereas SrTableKS(t) could be integrally transformed into RrTedKS,

CrTableKC(t) would never be integrally transformed in RrTedKC at tend. This is the main difference between the evolution of CTK and the evolution of STK, due to the special nature of the social compared to the somatic—or more exactly to the special kind of

resistance that a collective KC offers to explication attempts regarding any of Collins’s four meanings of “explicable”.

Conclusion

83 Let us summarize the main conclusions of our analysis of Collins’s tacit knowledge framework. These conclusions concern four points: (a) The definition and the name of one of Collins’s three TK categories, namely Relational Tacit Knowledge; (b) Some profound differences between Relational Tacit Knowledge on the one hand, and Somatic/Collective Tacit-Knowledge on the other; (c) Extensions of Collins’s categorization, through the consideration of additional categories and the temporal

Philosophia Scientiæ, 17-3 | 2013 104

evolution of the different categories; (d) Collins’s suggestion that RTK, STK and CTK are mutually exclusive categories which, taken altogether, exhaust the territory of TK. • (a) Our investigations showed that one of Collins’s sub-cases of Relational TK, viz., unrecognized knowledge, is not relational at all. The reason for it not being explicated has nothing to do with the relations between people. Along the same line of reasoning, ostensive knowledge and logistically demanding knowledge are only intermediate cases of Relational Tacit Knowledge in Collins’s sense, since beyond reasons related to the particular relations between people, they involve additional reasons of a different kind. The fundamental principle that defines RTK and individuates the RTK category as one category distinct from the two other categories of STK and CTK is the following: RTK remains tacit for reasons unrelated to the nature of the knowledge K involved. Once this is recognized, Collins’s RTK category is much more exactly characterized as contingent tacit knowledge or as “Relative-to-contingent- factors-regarding-the-nature-of-K” Tacit Knowledge. • (b) Our analyses recognized a possible ambiguity in the reading of the first letter of Collins’s acronyms RTK, STK and CTK, and specified of how to avoid the related possible confusions. Relying on Collins’s writings, we may be tempted to read the R, S and C both as reasons for the tacitness of K, and as locations (or nature) of K. We showed that on penalty of becoming entangled in paradoxes, the R should be interpreted only as a reason, whereas the S and the C should not only be read as reasons but also as locations. These findings strengthened a point implicitly suggested in Collins’s text, namely the serious asymmetry between RTK on the one hand, and STK/CTK on the other. A first source of asymmetry is that the tacit in S/C TK means “not explicable at t (and a fortiori not explicated at t)”, whereas the tacit in RTK means “explicable but not explicated at t for contingent reasons”. This difference is present in Collins’s text, but we made it more salient thanks to the introduction of our third convention. A second asymmetry lies in the fact that contrary to the S/C of STK and CTK, the R of RTK must be read only as a reason and not as a location: the R says nothing about the location, or in other words about the nature, of the K involved. This difference is not stressed as such by Collins. Finally, a third aspect of the asymmetry is the fact that RTK is primarily a contextualized or local concept, whereas S/CTK are global or “universal” ones. STK and CTK are tacit for any human being (in a given stage of knowledge). Whereas RTK is tacit relatively to a given “here and now” and because of the contingencies of this “here and now”. True, we can also consider RTK globally and talk about “all of RTK”. But in that case, what we actually do is to extrapolate to an indefinite multiplicity of possible “here and now”, and to imagine that in each, bits of the totality of our explicable knowledge (in a given stage of development) could remain unexplicated. Taken altogether, our analyses have strengthened and explicated more fully the asymmetry between RTK and S/CTK, perhaps even to such an extent that we should conclude that the RTK category is at right angles to the S/C TK categories. • (c) A third set of results pertains to our endeavors to elaborate and complete Collins’s framework. First, we elaborated the category of RTK, by introducing the possibility of different sub-types of RTK. After having recognized and stressed that the tacit in RTK had nothing to do with the nature (or location) of the knowledge K involved, it was almost inevitable to consider the possibility of several distinct locations for RTK. In this respect, we first took into account the two locations put forward by Collins in his book, namely the body and the collectivity. This led to the idea of some knowledge K localized either in the body or in the collectivity, but tacit for reasons that have nothing to do with these locations. In this way, we generated the idea of two possible sub-types of RTK not considered by Collins,

namely RrTedKS and RrTedKC. Next we raised the question of other additional sub-types, noted

Philosophia Scientiæ, 17-3 | 2013 105

RrTedKN, which would be located elsewhere than in the body or in the collectivity (for example “in the mind” of an individual scientist). Besides the elaboration of the statics of Collins’s framework, we completed his work by looking at the possible dynamics of his categorization. In TEK, Collins only considers the internal dynamics of each category. For instance, he discusses the question under what circumstances we may rightly claim that some STK has become explicated in time. Nevertheless, he does not consider the possibility of dynamic transformations between different categories. Let us recap our intermediate conclusions about dynamic transformations—much more should be done to come to more definite conclusions. For STK, relying on Collins’s writings, we should accept, first that over time some parts of STK can become RTK, second that at the “ideal end of scientific research”, all STK may become extinct (providing we accept, for the sake of the argument, the condition of “no-production and no-losses” of somatic knowledge). For CTK, things are much less clear. The text of TEK leaves us with the question whether some parts of CTK are explicable and whether some new parts might become explicable through time. If the answer is positive, some parts of CTK could be transformed into RTK in the course of time. However, Collins’s text is clear with respect to the claim that essential parts of CTK will remain inexplicable in any foreseeable future. Thus CTK, contrary to STK, will not become integrally explicated. In one word, the asymptotic behavior of STK differs from that of CTK. • (d) Collins’s text suggests that the three TK categories form a taxonomy, in the sense that RTK, STK and CTK are mutually exclusive and collectively exhaustive sets, and therefore partition the entire field of tacit knowledge. Due to (a), (b) and (c) above, our analysis challenges both claims. First, Collins’s categories are only mutually exclusive if considered at one specific time. Through time, one can be transformed into another. This is at least the case for Somatic Tacit Knowledge. From t to t′, when science advances, some inexplicable somatic TK at t may become explicable at t′. If this at-t′-explicable somatic TK is, in some particular situation at t′, not explicated for contingent reasons, then, STK(t) has been transformed into RTK(t′). Second, Collins’s categories are not obviously collectively exhaustive. What about the not-somatic and not-collective tacit knowledge of collectively acknowledged experts? What about the contributory intellectual expertise of an individual scientist who would be, according to the practitioners of the field, the only one to be able to accomplish a certain task? Nothing in Collins’s framework seems to prevent such a situation occurring. This becomes the more urgent when recalling Collins’s emphasis on the restricted size of the “core sets” of competent scientists with regard to some important scientific issue [Collins 1981]. In TEK, Collins does not consider the issue of the individual, intellectual tacit knowledge of collectively acknowledged experts. It is unclear where we should put this kind of tacit knowledge in his framework.

84 Finally, let us make a last comment associated with two suggestions. We cannot develop them here, but they are important enough to be at least mentioned, in the hope that they will prompt further investigation. In section 2.2. as well as in point (b) of this conclusion, we have presented the local character of RTK “as a special feature of RTK compared to STK and CTK”. This is true as far as Collins’s current uses of STK and CTK are involved. As we understand TEK, somatic and collective knowledge are said to be tacit/ explicit in reference to something like a “universal subject of knowledge at a given time”, or put differently, in reference to “a given stage of human development”. For example, the human somatic knowledge that must be mastered in order to be able to balance on a bike is described as explicable and explicit, as soon as “we”, human beings, are able to build a mechanical bike balancer (in which case the somatic human

Philosophia Scientiæ, 17-3 | 2013 106

knowledge is explicable and explicit in sense 3 of “mechanization”, see section 3.3.3 for quotations), or as soon as “we” have a scientific theory of bike balancing (in which case the somatic human knowledge is explicable and explicit in sense 4 of “scientific explanation”, see section 3.3.3 for quotations).

85 However, beyond Collins’s uses in TEK, we think it would be preferable—this is our first suggestion—to distinguish carefully between the global-universal sense and the contextual sense of explicable/explicit somatic knowledge and (provided that such things exist) explicable/explicit collective knowledge, and to specify systematically which sense is used. More precisely, analysts interested in issues involving tacit knowledge should care about the answer to the question: explicable/explicit for whom and when? This is required, since obviously—using the bike balancing example again— not all human beings are, nowadays, concretely able to build mechanical substitutes of the somatic human skills involved in bike balancing, or concretely able to master the scientific theories about bike balancing that could work as theoretical substitutes to the somatic human skills involved in bike balancing if we lived on “a small asteroid with almost zero gravity” (see section 3.3.3). So in practice, somatic knowledge about bike balancing is explicable/explicit in senses 3 and 4 only for certain individuals (or subgroups of the whole humanity) and not for others. The universal “we” to which the words “explicable” seems to refer most of the time in Collins’s developments related to somatic or collective knowledge, is universal in an in-principle, generic sense, but is not universal in-practice in the sense of “for all individuals”. Moreover, in case some somatic/collective knowledge K is explicable by a given person, it might be, in a given particular situation, either explicated or not by this person. In the later case, although K is explicable in a universal (generic) sense and moreover explicable in the contextual sense of explicable for this S, K remains tacit in the particular situation under scrutiny: K is contextually tacit in the sense of “tacit-now”. Thus the universal “we” to which the word “explicit” often refers in TEK, in relation to somatic/collective knowledge, is not universal in-practice in the sense of “for all situations”.

86 That is not all. Not only should the explicable/explicit/tacit character of somatic and collective knowledge K be specified in terms of “for whom and when”; but moreover— this is our second suggestion—we should also specify systematically in which of Collins’s four sense(s) K is explicable/explicit or not. Otherwise, strange effects occur. Let us continue to illustrate these effects with bike balancing. Take for granted that bike balancing is, nowadays, explicable and explicit in senses 3 and 4 and in a universal sense (i.e., “we”, human beings today, are able to explicate, in sense 3 and 4, the human somatic knowledge required to balance on a bike). It remains, however, that until further notice, nobody knows how to describe/express in words the somatic human skills actually involved in the human realization of bike balancing. As Collins insists (especially through the comparison of the way humans and computers play chess successfully), mechanical and scientific devices are just functional substitutes which are able to produce the same end-product called “knowledge” (say, for chess, to win, and for bike balancing, to go from a point to another without falling). They cannot be equated with descriptions of the bodily skills actually involved in the human realization of the task. So we can have situations in which a knowledge K is globally- universally explicable/explicit in senses 3 and 4, but is nevertheless globally-universally not explicable/explicit in senses 1 and 2. If we sum up the situation just by saying that the STK involved in the activity of bike balancing is explicable/explicit tout court, this is

Philosophia Scientiæ, 17-3 | 2013 107

puzzling. Indeed, it remains that nobody is able to tell in words what he knows (his somatic skills) when he succeeds to go from point A to point B by bike without falling, and this is precisely what is usually meant by the traditional claim that Somatic Tacit Knowledge is irreducibly tacit (i.e., will never been fully explicable in sense 1 or 2). If we claim that all STK could be explicated tout court, as Collins does, we suggest that the we have surmounted the later impossibility whereas we have not; or worse, we suggest that the invention of some mechanical device or new theory might change something to the subjectively experienced impossibility to express (even to ourselves) things that we are convinced to know (for example how to balance on a bike).

87 To conclude, we would like to thank Harry for having offered us, through his original and insightful mapping of the territory of tacit knowledge, a renewed occasion to deepen our understanding of tacit knowledge. His proposal has worked as an important source of inspiration and a powerful tool with respect to our own attempts to analyze the nature of tacit aspects of scientific practices [Soler 2011a,b]. We hope that our endeavours to explore his framework from the inside, and our critical reflections and suggestions of extensions, will prompt further refinements and developments.

BIBLIOGRAPHY

COLLINS, HARRY 1981 The place of the core-set in modern science: Social contingency with methodological property in science, History of Science, 19(1), 6–19. 2001 What is tacit knowledge?, in The Practice Turn in Contemporary Theory, edited by SCHATZCHI, T. R., KNORR CETINA, K. & VON SAVIGNY, E., London; New York: Routledge, chap. 7, 107–119. 2010a Like riding a bike, New Scientist, 2762, 29–30. 2010b Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

MARKS, PAUL 2012 Autonomous cars ready to hit our roads, New Scientist, 2858, 19–20.

SOLER, LÉNA 2011a Les expérimentateurs sont-ils substituables les uns aux autres? (opacité des pratiques expérimentales, fragmentation de l’expertise, désaccords irrésolus entre praticiens: quelles conséquences épistémologiques ?), Le Philosophoire, 35 (La science), 65–113. 2011b Tacit aspects of experimental practices: analytical tools and epistemological consequences, European Journal of Philosophy of Science, 1, 393–433.

NOTES

1. For example: RTK is “knowledge that is tacit for reasons that are not philosophically profound but have to do with the relations between people that arise out the nature of social life. The reasons range from deliberate secrecy to failure to appreciate someone

Philosophia Scientiæ, 17-3 | 2013 108

else’s need to know” [TEK, 11]; the K of RTK is explicated or not, “depending on the relationship of the particular humans who are in interaction” [TEK, 159]. 2. Actually, after we wrote the first version of this paper, Collins told us that he first chose the expression “contingent tacit knowledge” instead of “relation tacit knowledge”, but finally opted for “relational”. In TEK, he sometimes characterizes RTK in terms of contingency (as in the quotation just given). 3. See, however, the qualifications made about this point at the end of the conclusion of this paper. 4. For a second sense of “universal”, see the end of our conclusion. Above, “universal” means what we call in the conclusion the “in-principle, generic sense” of “universal” (knowledge explicable by any competent person), as opposed to the in-practice sense of universal (knowledge explicable by any human being). 5. This might be criticized—see the contribution of S. Turner in this special issue. 6. And in conformity to some explicit statements of Collins, for example : “Weak, or Relational, Tacit Knowledge (RTK) [...] is knowledge that [...] is not made explicit for reasons that touch on no deep principles that have to do with either the nature and location of knowledge or the way humans are made” [TEK, 86].

7. To be even more precise, we could add a number to Table which would specify which of Collins’s four meanings of explicable is involved. Here we will not do so. 8. It is worthwhile to note in this context that in various places in the world (Nevada, Florida, California and Berlin) autonomous or driverless cars have been finding their ways in normal traffic situations. The cars have done so autonomously on highways and in cities using radar, GPS, computer vision and optical remote sensing technology. (http://en.wikipedia.org/wiki/Autonomous_car; http://en.wikipedia.org/wiki/ Google_driverless_car) On Tuesday September 25th 2012, Jerry Brown, the governor of California, changed the law by signing a bill that legalizes driverless cars in the street of California (the New York Times, September 25 2012). If Jerry Brown did not change the law in vain and Paul Marks is right in saying that driverless cars are there “ready to hit our roads”, to find their ways within major metropolises making less accidents than human drivers [Marks 2012], then either Collins should delete his Beijing driving example or should accept that some CTK is explicable. 9. This holds for RTK considered as a global-universal category: for “all of RTK” (see section 2.2). 10. Collins does not talk in terms of the “ideal end of research”, but he claims about STK that “this kind of tacit knowledge is continuous with that possessed by animals and other living things. In principle it is possible for it to be explicated, not by the animals and trees themselves (or the particular humans who embody it), but as the outcome of research done by human scientists (the third meaning of explicable [i.e., mechanization]). It is possible to foresee, therefore, that we will one day be able to mimic animal behavior with machines [sense 4 of explicable]. The same goes for humans-as- animals” [TEK, 84, our italics].

Philosophia Scientiæ, 17-3 | 2013 109

ABSTRACTS

In this paper, we discuss and extend the taxonomy of tacit knowledge proposed by Collins in his 2010 book, Tacit and Explicit Knowledge. First, we question the definition and the name of one of Collins’s three categories of TK, namely Relational Tacit Knowledge (RTK). After having explained the true fundamental principle that individuates RTK as one category distinct from the two others (Somatic Tacit Knowledge STK and Collective Tacit Knowledge CTK), we suggest an alternative name for RTK, which fits this principle better. Second, our analyses identify a possible ambiguity in the interpretation of RTK, STK and CTK, and indicate how to avoid the related possible confusions. For this purpose, we introduce several notations that specify Collins’s ones. The corresponding developments strengthen and specify a point implicitly suggested in Collins’s book, namely the existence of a serious asymmetry between RTK on the one hand, and STK/CTK on the other. Third, the paper attempts to elaborate and complete Collins’s framework. In this vein, we introduce the possibility of different sub-types of RTK. One of these sub-cases raises the issue of individual, intellectual tacit knowledge—a case that does not seem to have any place in Collins’s picture. We also look at Collins’s framework in a dynamical perspective, and discuss the possibility—not considered by Collins—of dynamic transformations between some of the three categories. Finally, we make two suggestions in order to avoid confusions or misunderstandings when using Collins’s distinctions. When the qualities “explicit”, “explicable” or “tacit” are attributed to some knowledge, these qualities should always be accompanied by the specification for whom and when. Moreover, the attributions “explicit” and “explicable” should always indicate which of Collins’s four senses is meant—elaboration, transformation, mechanization or scientific explanation.

Dans cet article, nous discutons et développons la taxonomie de la connaissance tacite proposée par Collins dans son livre de 2010, Tacit and Explicit Knowledge. Dans un premier temps, nous critiquons la définition et le nom d’une des trois catégories de connaissance tacite introduites par Collins, à savoir la connaissance tacite relationnelle (RTK). Après avoir expliqué quel principe fondamental individualise en fait RTK comme une catégorie distincte des deux autres catégories que sont la connaissance tacite somatique (STK) et la connaissance tacite collective (CTK), nous suggérons pour RTK un autre nom, plus en harmonie avec ce principe. Dans un second temps, nous mettons en évidence une possible ambiguïté dans l’interprétation de RTK, STK et CTK, et nous indiquons comment éviter les possibles confusions associées, notamment en introduisant plusieurs notations qui précisent celles de Collins. Les développements correspondants renforcent et spécifient un point implicitement suggéré par Collins dans son ouvrage, à savoir la forte asymétrie qui existe entre RTK d’un côté, et STK/CTK de l’autre. Dans un troisième temps, l’article s’emploie à prolonger et interroger dans de nouvelles directions la classification de Collins. Dans cet esprit, la possibilité de différents sous-types de RTK est introduite. L’un de ces sous-types pose la question d’une connaissance tacite individuelle et intellectuelle – cas qui ne semble pas avoir de place dans le cadre de Collins. La classification de Collins est également envisagée dans une perspective dynamique. Nous discutons en particulier la possibilité – non considérée par Collins – de transformations dynamiques entre certaines des trois catégories RTK, STK et CTK. Pour finir, nous faisons deux suggestions en vue d’éviter les confusions et malentendus lors de l’utilisation des distinctions de Collins : s’agissant des affirmations selon lesquelles une connaissance K est explicitable, explicite ou tacite, toujours préciser pour qui et quand ; s’agissant des affirmations selon lesquelles une connaissance K est explicite ou explicitable, ne pas s’en tenir à affirmer cela tout court, mais toujours préciser dans le(s)quel(s) des

Philosophia Scientiæ, 17-3 | 2013 110

quatre sens différenciés par Collins (élaboration, transformation, mécanisation, explication scientifique).

AUTHORS

LÉNA SOLER Laboratoire d’Histoire des Sciences et de Philosophie, Archives H. Poincaré (UMR 7117), CNRS, Université de Lorraine, Nancy (France)

SJOERD D. ZWART Delft & Eindhoven Universities of Technology (The Netherlands)

Philosophia Scientiæ, 17-3 | 2013 111

Ships that Pass in the Night: Tacit Knowledge in Psychology and Sociology

Harry Collins and Arthur Reber

Introduction

1 This piece is different from others in the volume. Arthur Reber (AR) and Harry Collins (HC) have each been working on tacit knowledge for most of their careers taking leading roles in establishing the topic in psychology, sociology and philosophy. Reber published his first paper on the topic in 1967 [Reber 1967] and Collins in 1974 [Collins 1974]. But neither of them has ever cited the other or, until very recently, read the other’s work despite the fact that the words “tacit” and “knowledge” are prominent in the titles of papers and even books that each has written [Collins 2010], [Reber 1993]. When Reber was asked to referee a contribution to this volume, the two of them fell into an email exchange. They discovered that poor scholarship was only part of what was keeping them apart. As the exchange stretched across five months and some 300 contributions, they found the real problem is that they speak different academic languages. In fact they speak different languages even in respect of the central topic of this volume and of much of their academic lives—tacit knowledge. They began to feel that they had stumbled into what seems to be a living instance of paradigm incommensurability [Kuhn 1962], or something close to it. Thus, the following exchange (25 November, 2012): AR: I think that part of the difficulty I’ve been having reading your writings (especially Tacit and Explicit Knowledge) is that you seem to struggle to say things that hit me with a “well, duh, of course...”. In one email you went on about saluting and in various papers you talk about parsing various kinds of tacit knowledge and different instantiations of the act of riding a bicycle and I keep trying to figure out what you’re trying to convince me of that I don’t already know. HC: The origins [of the problem] could be that you don’t quite get that I am dealing with knowledge-stuff rather than individual learning.

Philosophia Scientiæ, 17-3 | 2013 112

2 But later (31 December, 2012) we still find: AR: Figuring out how humans do things is what I do. And it is still a strain on my brain to comprehend that it is not what you do. HC: Whereas, as I keep saying, I am interested in knowledge-stuff, not how humans learn. AR: I know this when you say this. I understand the words. Then I say to Rhiannon [AR’s professional psychologist wife], “do you know what Harry wrote today?” [...] and I find that I cannot form the words to express this thought of yours because it doesn’t fit in my framework. In my world there is no such thing as “knowledge stuff” independent of the humans holding the knowledge. It feels like saying you’re interested in how pawns move without looking at the game of chess.

3 These differences and the way they play themselves out in each discipline seemed worth exploring—so the conversation continued.

4 The authors have never met. Perhaps things would have been easier if we had. There were occasions when the dialogue came close to ending because of frustrations and personal misunderstandings that arose with brittle and inflammatory email interchanges. With face-to-face discussion it is possible to transmit more of the tacit! But we believe there is enough here to shed some light on the problems of interdisciplinarity. The purpose of this paper is, then, to explore and, to some extent, explain how such a situation can arise and continue. The focus will, of course, be centred on the example of the analysis of tacit knowledge but inter alia, the analysis may shed some light on academic misunderstandings in general—perhaps helping others who find themselves in a similar position—and it may even shed a little on the relationship between psychology and sociology. Toward the end of the paper there is a section describing the “Seven causes of misunderstanding” (p. 150) that we were able to pull out of our experience. This list may prove useful to those who find themselves faced with similar problems.

1 Different starting points

5 Some of the divergence can be explained by the parties’ paradigmatic early experiments and observations in the field of tacit knowledge. Reber asked individuals to memorize sequences of letters that were, though they did not know it, made up using complex rules—an “artificial grammar” (Reber 1967). Over time they became sensitive to these patterns and could differentiate novel well-formed sequences from those that violated the rules even though they were unaware of what they had learned or even that they had learned—it was tacit knowledge. He called the process implicit learning and he contrasted it with other approaches that treated learning as a self- conscious process of hypothesis and test.

6 Collins noted that scientists trying to learn from others how to build a new kind of laser —the TEA-laser—failed unless time was spent in a successful laser-builder’s company [Collins 1974]. Even the most detailed written specifications would not enable them to build successfully though they could follow the circuit diagram and use components from the same manufacturers. No one knew what was being transferred in these face- to-face interactions—it was tacit knowledge.

7 Reber’s approach was quintessentially psychological—the artificial lab-based phenomenon being stripped down to its basic essentials in an effort to control all the

Philosophia Scientiæ, 17-3 | 2013 113

variables. Collins’s was quintessentially sociological—a messy, natural, social situation. Collins’s initial focus was less on the reasons for failure of transmission and more on overall outcomes and the fact that face-to-face interaction was vital—though he went into more detail in later studies. One can, perhaps, see why for Reber the very meaning of knowledge became associated with the actions of individuals and was generally explored independent of society whereas for Collins its very meaning was to be found in what unfolded in society.1 Thus, for Reber, tacit came to be a synonym for unconscious, implicit. For Collins it was semantically closer to unspoken or unexplicated.2 Making matters worse was that each used explicit as an antonym, though for Reber this meant that the knowledge was made conscious by self-reflection while for Collins it meant that it could be explicated or explained.

8 While nowadays there is little reference in the psychological literature to the acquisition of tacit knowledge—the term introduced by Reber, implicit learning, having become the default term—the attention of psychologists tends to be on the acquisition of tacit knowledge rather than the substance. In sociology, particularly Collins’s approach, the effort is to analyze the substance. So we have Reber exploring process while Collins is concerned with content.

2 Ontologies

9 The authors are informed by different ontologies. Consciousness is important to Reber and unimportant to Collins. Collins puts a strong emphasis on the social nature of knowledge which is largely ignored by Reber. Reber emphasizes continuity between humans and other species whereas Collins emphasizes difference. Reber believes that consciousness is a primitive property of living organisms. He associates knowledge with consciousness and this means that machines (at least current machines which are not made of living materials) do not have knowledge. Reber is strongly informed by evolutionary theory and sees steady development from one entity into another with consciousness always present but its power evolving. AR: I’m interested in “human acquisition” of knowledge but it is an interest that is a subset of much else. Human knowledge lies on a continuum with that of other species. I argue that the implicit mode of acquisition is fundamentally the same as various learning processes seen across the phylogenetic scale.

10 The two authors discovered that their uses of the terms consciousness and reflectiveness were somewhat confounded. For example, both agreed that cats were conscious but this did no explanatory work for Collins whereas it was important to Reber. Reber also considered that cats might well be reflective from time to time. In his view, a cat always “has consciousness” like a person or a snake or any other living entity—though the form of consciousness differs dramatically from species to species. A cat that is awake is “being conscious”; one that is asleep is “unconscious”. A cat that is acting in automatic, pure bottom-up mode—what it does virtually all the time except perhaps some occasional flickering of reflection—is acting implicitly, without conscious modulation of its behaviour. Thus for Reber, self-conscious attention to activity was the same as being reflective. In implicit learning there was no reflection going on whereas self- conscious rule-based learning was reflective.

11 Collins thinks that, in the main, only humans are reflective. Reflectiveness is tied up with the use of language. He agrees that there is a fuzzy borderline between humans

Philosophia Scientiæ, 17-3 | 2013 114

and animals occupied by creatures such as chimps and dolphins which may share the rudiments of a language but Collins thinks that creatures on the borderline should be ignored if the argument about the nature of reflection is to be clear. Reber thinks the borderline is critical because he’s far from convinced that it is language per se that is key to reflection. Human language, in his view, emerged along with high-level cognitive functions. It is likely that these cognitive abilities were instrumental in language and in the capacity for self-reflection rather than one causing the other. The jury is out on this but Reber is not comfortable assigning such a vital role to language. He notes that communication is a fundamental feature of many species—this fits with his preference for continuity.

12 Collins, who is interested in differences, insists on using cats as the animal example so as to get away from the borderline—no one thinks cats are language users. But, as noted, Reber even allows cats to be part of his continuum: they don’t reflect much on their knowledge but they might do so every now and again and they certainly communicate. Collins is drawn in exactly the opposite direction and, to make the point, he now proposes the ‘tacit-to-explicit’ test. A species can only be said to be capable of reflection, according to Collins’s newly invented criterion, if we can imagine some of its members trying to make their tacit knowledge explicit so that they can store it (as they see it), in cave-paintings, in hieroglyphs, and in books, and so that they can broadcast it to others not in the vicinity, and so on.3 As can be seen, Collins’s thought style (see “Seven causes of misunderstanding” p. 150) leads him to draw things apart; Reber’s thought style leads him to draw things together. But in both worlds, it is only humans who endeavour to make their knowledge explicit.

13 Collins’s central concept is socialness which is based on human language. Language is not the same as information exchange, so bees and ants are excluded (one can see also that they would fail the tacit-to-explicit test). He believes there is a clear division between entities that are social in this sense and those that are not. Compare Reber: AR: Ever watch a cat? It moves, turns left, stops, licks its front leg, goes over to a toy, picks it up, tosses it, chases it [...] and so on and so on. [...] Now go back and think how you spent the morning... you got up, scratched, rubbed your tummy, brushed your teeth, made coffee/tea (whatever), walked down the hallway [all unreflectively]. [...] It seems to me that you and the cat (and me) are rather similar...

14 For Reber, of course, humans do additional things that involve reflection like decide to send an email, argue a philosophical point and go shopping. In the view of both Reber and Collins, however, virtually every interesting thing that humans do involves a blend of the implicit and the explicit, the unconsciousness and the conscious, the automatic and the reflective. Though, as explained, for Reber, but not for Collins, even the cat might reflect occasionally. For both, the self-conscious reflective things that humans do are different from the things that the cat does most of the time. So Reber sees two kinds of things happening in this scenario: the cat and the human doing similar unreflective things and the human doing reflective things (but in another scenario the cat could be doing reflective things too).

15 Collins, however, sees the cat and the human as very different even when they are both doing things unreflectively. In particular, the cat cannot brush its teeth, make coffee/ tea or even, in his view, walk down the ‘hallway’—a hallway connotes a great deal to a human while a cat is just walking along an elongated space. Humans can only do things like brush teeth, make caffeinated beverages and walk in halls, however unreflectively

Philosophia Scientiæ, 17-3 | 2013 115

they do them, because of the existence of a range of corresponding institutions linked together by language. Nothing the cat does is like this. Cats’ activities are circumscribed by their evolutionary history; different groups of humans are, however, enormously different, the differences emerging from the reflective activities of other humans who are distant in time and space from what is going on now. From Collins’s perspective, all these humans are linked together by a network of common social activity and language. So while both cat and human may be doing things in an unreflective way, most things that humans do depend on a history of reflection by other humans of a kind that the cat has not shared and cannot share since it has no language nor social life in the strong sense of social. Reber has no deep problems with this form of analysis which he views as a viable, if different, way to approach learning, language, communication and social function.

2.1 The tacit

AR: [...] when you use the term “tacit” [...] to distinguish it from “that which can be explicated”, you are flying in the face of a half-century of usage in psychology. [...] When I introduced the term “implicit” (in my MA thesis), I specifically chose it as a contrast against the “explicit” hypothesis-testing process that was being championed by people like Jerry Bruner at Harvard. Bruner and colleagues were developing a theory of knowledge acquisition that was based on the assumption that people learned new stuff by testing explicit (i.e., consciously held) hypothesis about how the world about them functioned [Bruner, Goodnow, & Austin 1956]. If their guesses were confirmed, knowledge became fixed. If disconfirmed, they tried another one. It was an extension of the “hypothetico-deductive” approach to science. Bruner was trying to push it into personal epistemology—a position which you would almost certainly call “idiotic”. I did but for different reasons. My view was that this approach wasn’t wrong in any fundamental way. It was merely grotesquely limited. Yes, in rare circumstances people behave this way but they are few and far between. For example, it made no sense when talking about how an infant learned language or a child became inculcated with the mores of the surrounding social world. And, as you would agree, it made no sense at all when looking at how science was actually done—despite the fact that this is where Bruner began. None of these things were learned consciously, not language, not socialization, not even most scientific knowledge—they were learned unconsciously. [...] In cognitive psychology, making knowledge “explicit” is an act of an individual who is discovering the nature and form of knowledge previously held implicitly.

16 Both parties agree that (some) knowledge can be tacit for one person and explicit for another (in Reber’s sense of tacit/implicit), and can be tacit for one person at one time and explicit for the same person at another time. Both agree that it is not uncommon for a single person, who has had some piece of knowledge rendered explicit for him or her, to switch between using that knowledge in an explicit/self-conscious way or an implicit/tacit/unself-conscious way.

17 Consider the example of gear-changing in a car. Usually a novice driver will initially be taught to change gears in an explicit way—something like “shift into 2nd when the revs reach 2000”, or “change when the sound of the engine reaches a high pitch”. After a while the rules are forgotten and gear-changing becomes automatic, implicit. On the familiar commute to work we can be thinking about all manner of things while changing gears without being aware that we are doing it. Experienced drivers change gears unconsciously while thinking about something else—but there is nothing to stop

Philosophia Scientiæ, 17-3 | 2013 116

them executing the shifts in a more self-conscious manner. For Reber the task changes when focus of attention changes and the Reberian analysis concentrates on the different nature of the task under conscious control versus non-conscious control. For example, non-conscious execution is generally more efficient for normal road conditions whereas self-conscious attention might be better when the road is icy. Collins acknowledges all this but it is not a central feature of his analysis of knowledge.

18 For Collins, gear-changing is the same ‘mimeomorphic’ (and therefore explicable) action whether it is currently being executed by a human in a self-conscious way, by a human in an implicit manner, or by an automatic gearbox mimicking the action. What makes it the same kind of action irrespective of how it is carried out at any specific time is that the action can be reproduced without reference to social context. Collins and Kusch called this a mimeomorphic action because it can be reproduced (or mimicked) by merely reproducing the behaviour associated with the action (e.g., as with a salute) [Collins & Kusch 1998]. In contrast, a polimorphic action depends on sensitivity to social context because the same behaviour does not always reproduce the same action (e.g., a greeting which, to be authentic, has to be varied from time to time). The terms mimeomorphic and polimorphic refer to whether or not the externally visible ‘shape’ of the action is merely copied or must change from social instance to social instance. To Collins, the very fact that there can be such a thing as an automatic gearbox is a consequence of gear-changing being a mimeomorphic action rather than polimorphic. It means that, in so far as gear-changing can be imagined to be learned entirely tacitly (and one can imagine such a scenario), it would be a species of Relational or Somatic Tacit Knowledge, not Collective Tacit Knowledge. Reber appreciates this parsing of the domain of tacit knowledge but it does not play a significantrole in his thinking.

2.2 The key distinction

19 The examples of gear-changing and the cat reflect Collins’s interest in “knowledge- stuff” and Reber’s in individual learning and execution of acts. Collins and Reber cut up the world in different ways. For Reber, the essential thing is that actions like gear- changing can be done self-consciously or unselfconsciously depending on circumstances; the topic is the different ways of executing gear-changing. For Collins all gear-changing, however it is actually done, is of the same kind; it is knowledge that can be explicated and (potentially) automated with foreseeable technology as opposed to knowledge that cannot be explicated and automated with foreseeable technology.4 For Reber, the cat and the human, when they are being unreflective, are exhibiting the same kind of knowledge—or at least using similar evolutionarily ancient systems for expressing the knowledge. For Collins, in most instances the knowledge is very different. In Reber’s work, gear-changing can take two forms, conscious and unconscious; in Collins’s work there is but one form—mimeomorphic. For Reber, unreflective cat and unreflective human exhibit one kind of knowledge; for Collins they exhibit two.

20 Finally, notice that the data from Reber’s experiments have been simulated by neural net models, suggesting that associationistic models can capture aspects of the tacit dimension of human knowledge. However, in Reber’s view a neural net is just a model of a system that likely exists in brains. Nets don’t have knowledge just like chess

Philosophia Scientiæ, 17-3 | 2013 117

playing computers don’t know anything about chess. For Collins, in spite of Reber’s qualifications, the success of computers like neural nets in reproducing the effects shows the restricted range of the tasks that are modelled in the Reber experiments— they concentrate on the mimeomorphic aspects of language.

3 Knowledge and consciousness

21 Eventually they began to realize that they were using consciousness in different ways; in some of these the connotations overlapped with knowledge and broader issues of epistemology but, alas, in others they didn’t. AR: As we move along from simple and primitive organisms to the complex and sophisticated there is a shift from being dominated by the implicit to subsuming the implicit under an increasingly important explicit system. This shift occurs for learning, for memory and for encoding emotional situations. They shift from being utterly implicit to being open to introspective scanning and available for conscious recollection. There is a continuity here and, as we move along the phylogenetic scale, various specialized forms of “knowing” and “acting” and “retrieving” emerge. They do so to fit the demands of particular ecological circumstances. As brains got bigger the role of top-down, modulating functions increased. In humans it reached its pinnacle.

22 In the Reber thought-style, knowledge is on a continuum understood to emerge out of evolutionary forces. Therefore, it is Reber’s view that machines cannot know things. They cannot have knowledge because, being made of non-biological materials and, not belonging on the same evolutionary tree as humans, there is no consciousness there and the very notion of knowledge is simply not applicable.

23 Collins went back to Reber three or more times to ask the question about how he justified the claim that machines could not have knowledge but cats could, even though both operated unconsciously at least some of the time. He was never convinced by Reber’s answers even though he believed he understood what Reber was saying. Here are examples of the interchange: AR: “Consciousness” is a feature of particular kinds of organisms. It denotes a continuum of subjectivity. Its most complex (and intellectually seductive) instantiation occurs in humans. “Knowledge” is a body of facts and information that organisms have. Since these organisms have consciousness, their knowledge is linked with this phenomenal state—with the understanding that much of this knowledge is acquired and held tacitly. You can study knowledge in a disembodied way, of course, just like you can study how a taste-bud responds to sugar but this won’t get at what you experience when you eat chocolate cake. So I guess the issue here isn’t so much whether these things are ontologically distinguishable but what the goal of the exploration is. I don’t think you’ll learn much about “knowledge” if you dissociate it from the organism doing the “knowing”. HC: In your world machines cannot have knowledge because they do not have consciousness? AR: When we say a computer “knows” how to do arithmetic, the “knowing” here is, in my world, very different from when we say a person “knows” how to. If you wish to say that the computer has “machine knowledge” I guess that would be okay but it detracts from the epistemic character we typically assign to “knowledge”. HC: But a cat can have knowledge because it is conscious? AR: I think of it in a Tom Nagel’ish fashion: There is something it is like to be a cat.

Philosophia Scientiæ, 17-3 | 2013 118

There is nothing it is like to be a computer. HC: Even though a cat usually uses its knowledge unconsciously. AR: Yes, but so do humans. Much (most?) of the time we’re on automatic pilot [...] I spent this morning just like a cat.

24 We have already seen that Collins disagrees with this last claim but he also just does not understand the role of consciousness. He does not understand the confidence with which subjective understanding is readily imputed to cats and readily denied to machines. This confidence seems to rest solely on the theory that only living material is conscious. So Reber’s world view is consistent but it seems to Collins that the position is not and cannot be established by reference to evidence. Reber, of course, disagrees, feeling that there is substantial empirical support for his position. It is a characteristic of incommensurability that, where one person sees evidence, another person does not. Here, from Collins’s viewpoint there is no evidence for Reber’s position, while from Reber’s viewpoint there is more evidence for his position than there is for Collins’s. Each party believes that the other is basing its argument on something less than adequate. Collins thinks Reber does not care about evidence, whereas Reber thinks what Collins counts as evidence is no more substantial than what he bases his own argument on. The interchange, then, can be said to be characterised by “mismatched explanatory adequacy” (see “Seven causes of misunderstanding” p. 150) even though each party believes the other is making a mistake in the way it views the opponent’s position.

25 This tension reiterates the central discontinuities at the heart of both approaches. For Reber, he really was acting like a cat when he got up because what constitutes Reber’s universe is how creatures attend to what they are doing and he and the cat attend (or not) in the same way much of the time; furthermore, they are both conscious. For Collins, the knowledge of the cat and the knowledge of the human are of completely different types and while the human can act like an animal (e.g., scratching) the cat can never act like even an unreflective human engaged in acts that depend on language. Furthermore, Collins does not understand the explanatory status of claims like a cat has knowledge because “there is something it is like to be a cat” whereas a computer does not because there is nothing it is like to be a computer.5 Reber notes that cats and people howl and jump if you stick them with a pin. A computer makes no such response; it merely loses a couple of bytes. In the former we have subjectivity and phenomenal experience, in the latter we have neither. Collins understands what Reber is saying but does not see it as evidence for consciousness being a correlate of knowledge.

26 Reber feels no need to define knowledge beyond what can be found in the dictionary but Collins has to define knowledge-stuff. He inclines toward what he thinks of as a Wittgensteinian meaning for knowledge, namely, that to understand the meaning of words one must understand their use. To know the meaning of a word, then, is to be able to use it as it is used in society. This approach fits, as it happens, with the philosophy of the Turing Test where intelligence is demonstrated by a performance that is indistinguishable from that of an entity that is known to be intelligent—in the Turing Test it is use that is the criterion.

27 Collins, then, thinks of knowledge as the stuff you have when you can do certain things. If TEA-laser builders hung around with successful laser builders something passed to them that was still in them on their journey home—and which they could then use to build a successful laser when they arrived. There was no reason to think that their

Philosophia Scientiæ, 17-3 | 2013 119

knowledge was affected by what it felt like to be a laser-builder and the readily available criterion of having knowledge was being able to build a working laser. For Collins, it follows that we can imagine a machine building a laser and this makes machines candidates for the possession of knowledge and it means that work is required to show if and why they would be different from human laser-builders. There is nothing in the definition of machines or substance of machines that prevents them being such candidates. Reber, from his evolutionary stance, demurs. The “Nagelish” point isn’t that the notion that “there is something it is like to be laser-builder” necessarily affects knowledge. It is a mental state whose causal roles need to be determined. Collins thinks he understands what Reber is saying about consciousness but does not see what it has to do with knowledge.

28 Collins argues, however, that existing and foreseeable machines cannot have full human-like knowledge because they do not share human social life. This argument is testable. It is, for example, why all current and foreseeable machines fail properly conducted Turing Tests. The failure is visible from the outside without reference to internal states.

29 Naturally, we had a long interchange about the Chinese Room which may illustrate “focus blindness” (see “Seven causes of misunderstanding” p. 150) or at least the difference between our projects. Searle’s Chinese Room, it will be recalled, shows that a performance equivalent to that of a conscious human does not prove consciousness [Searle 1980]. But this makes no difference to Collins as he is only interested in performance, not consciousness. If the Chinese Room worked in a linguistically perfect way as advertised, however, it would disprove Collins’s view that no foreseeable computer can act as a socially embedded being. But Collins believes the Chinese Room would not and could not work as Searle describes it: HC: [...] because language is the property of the embedding society, is continually changing, and there is no mechanism linking the Chinese Room to the changing society. Over time it would start to perform archaically (just like a human isolated from society). If a mechanism is put in place to link it to society (e.g., the database is updated by humans) then the ‘socialness’ will be located in that mechanism, not the Chinese Room. AR: I have difficulty with this argument. It seems to me pretty straightforward that the Room is not conscious and, as with any artificial entity, there is nothing it is like to be the Chinese Room. In so far as it needs to be social, the social aspects of language are embedded from the outset. Whatever information the Room has at the beginning of the test was put there by programmers with social knowledge. So why are they excluded from adding information over time? The humans it’s being compared with have input. And if, as you say, this added socialness is then not “in” the mechanism, why is it not “in” it at the very beginning? HC: Yes, the social is in there when the Room is first set up but it is a frozen snapshot. If the social is updated by human attendants it is the attendants that are making the connection between the Room and society—that is where the social gets in—the link via the human attendants. That is the point! What we do not know how to do is to automate that process. That, in my view, is the key to making computers handle language like humans handle it.

30 We leave this exchange uncommented on as a classic instance of academic misunderstanding.

Philosophia Scientiæ, 17-3 | 2013 120

4 Conclusion on consciousness

31 There seemed to be only one place where we wanted to say to each other “you are wrong in a really serious way and I am right”. Reber thinks Collins will never understand knowledge if he does not take consciousness into account as a central feature. Of course, Collins thinks no one understands knowledge if they do not understand socialness but, more importantly, he thinks that Reber’s division of the world into conscious and non-conscious, and therefore knowledge-possessing and non- knowledge-possessing things is arbitrary and explanatorily inadequate. Plants are out, but every other thing made of living matter is in; everything not made of living matter is out. It seems to Collins that, here, constructing a consistent world view has taken priority over the desire to develop a theory with observable consequences.

32 Reber maintains that consciousness does have observable consequences. One can see the difference when humans move from implicit to self-conscious execution of a task. In fact, a variable often manipulated in experiments on implicit learning is awareness. It turns out to be a critical component in recruiting different kinds of cognitive processes and impacts on how individuals engage in remembering material, making decisions, forming preferences and making aesthetic judgements [Reber 1993], [Zizak & Reber 2004].

33 Collins finds that this reply does not bear on his question and sees this difference as being no more significant than using two different computer programs to execute the same task. The question for Collins is: Are there things that humans can do in virtue of their consciousness that machines cannot do in virtue of their lack of consciousness? Collins cannot think of anything that lack of consciousness in this sense, rules out (whereas he can list many things that machines cannot do in virtue of their lack of socialness). Reber thinks that Collins’s question is not a good one, or at least not useful, and he fails to grasp how Collins deals with things like aesthetic judgements which require consciousness and cannot be made by machines. Collins responds that if machines had socialness they would be able to make aesthetic judgements—which are quintessentially social.

34 What divides the two parties here might be explained with a philosophical analogy using the old philosophical puzzle, “If a tree falls in the forest and there is no one around, does it make a sound?” One possible answer is “yes” and “no” for it depends on your initial stance. To a physicist the tree creates waves in the air called “sound waves” whether or not anyone hears them. To a psychologist, for there to be sound, there must be someone who has the subjective experience which we call “sound”.

35 In the same way one might ask: “If there were two machines having a human-like conversation in a room and there is no one about to join in, is there knowledge being exchanged?” Collins, from his “knowledge-stuff” perspective, answers “yes” because, presumably, the machines are learning to do new things from each other. Reber, from his subjectivist point of view, says “no”. Reber acknowledges that there is this stuff called “knowledge” in the room but claims it has nothing to do with humans knowing things any more than the physicist’s description of the sound of a tree falling tells us anything about whatit sounds like.

Philosophia Scientiæ, 17-3 | 2013 121

5 Evolution

36 Reber argues that evolutionary theory is the key to understanding human knowledge. In Collins’s view, evolutionary theory is of very little help; the most remarkable characteristic of humans is how different from each other groups of humans have become since significant evolution ceased. This point of departure is, perhaps, typical of tensions between psychologists and sociologists. Interestingly, however, Collins’s position is not so far from Reber’s in one of its aspects as this extract from the interchange illustrates: HC: In TEK I say that tacit knowledge is unexceptionable because that is what animals and other living things have done since they emerged from the slime and it is explicit knowledge that is extraordinary. AR: I could have written this last sentence myself...

37 So far so good, but then things started to diverge. Reber believes that human knowledge evolved and that consciousness came with it. As Reber points out, this has some interesting consequences for our expectations of the relative role of implicit and conscious handling of knowledge: the early implicit elements of our abilities (shared with animals, of course) should be relatively robust and free from cultural variation.

38 This cuts right across what interests Collins. Almost the whole of Collins’s academic life has been devoted to showing that what seems quintessentially to be the product of conscious processes (much of what happens in science) is deeply tacit (as in the example of the TEA-laser). Furthermore, language, which comes very late on the evolutionary scene and is very much associated with consciousness, is, for Collins, largely tacit in use. Collins, then, finds Reber’s use of evolutionary theory and consciousness orthogonal to what he wants to say about the relationship between language and the tacit.

39 There is, however, something more positive to be said. In TEK Collins continually states that he does not understand the explicit; he does not understand how strings carry meaning. For example, he points out that the icon for house is nothing like a house so it is very hard to say what it is that is upside down about the upside down version. In the face of these difficulties Collins adopts the term affordance which he calls a “conceptual bandage” since he has no concepts that would allow him to explain why an icon or anything else has any affordance in the first place. TEK works by addressing an easier problem which is how some string that has insufficient affordance to take part in an act of communication can be enhanced (for example, by making it longer or more elaborate).

40 But there is a big gap in Collins’s theory. How can any string have any affordance in the first place? Were Collins ever to want to fill this gap he suspects he might well be forced to adopt something like Reber’s evolutionary approach. He would have to say that animals, as they emerged from the slime, evolved to extract certain basic shapes from the environment (such as the upright triangle representing the roof in the house icon). That would be a way, and the only way he can think of, to explain why we can see that the upside-down icon is upside down. Collins, trying for a moment to adopt Reber’s perspective, would predict that upright triangles, since they must have become recognisable as distinctive entities so early on in human evolution, should be recognised across all cultures. There must be, he would argue, a substrate of fundamental shapes, patterns and perhaps colours and textures that all humans

Philosophia Scientiæ, 17-3 | 2013 122

recognise and upon which the variations in culture are built.6 Such influence as this underlying structure has is clearly “bottom-up”.

41 The trouble, for Collins, is that much of cultural variation is “top-down”. Humans are capable of seeing almost anything as anything else. The question is going to be how the top-down variation of actual and potential human culture is related to the bottom-up structural vocabulary and how bottom-up explanations can bear on top-down explanations. All that has been accomplished, then, is to see the point of evolutionary theory and acknowledge the gaps in Collins’s theory in a more self-conscious way. But this has to be better than mutual incomprehension and talking at cross-purposes.

6 Overall conclusion

42 The question we started with is how, for nearly half a century, two academics could work on the analysis and exploration of tacit knowledge in their own disciplines and find no deep need to refer to each other’s work. The dialogue and its analysis have shown, we believe, how it can be—they start with different methods, fit into different world views, have different explanatory goals and use different language to talk about them. In the main, these are just differences and, in another universe, either party would be happy to have accomplished what the other has done.

43 Nevertheless, one can see why the ships pass in the night. In TEK, Collins claims to have constructed a map in which the many approaches to tacit knowledge listed can be related. Staggeringly, Collins finds that he did not even include Reber’s work in his inventory of approaches to tacit knowledge and now he finds that it will not fit on the map, certainly not easily. The map was based on the ways that the different approaches dealt with Collins’s three kinds of tacit knowledge but the conscious/implicit individual works with a different substance. Reber works with the forms of human attendance to tasks—that is the stuff of his world; Collins deals with the nature of the tasks irrespective of how they are attended to from moment to moment and that is what the map portrays.

44 There are other differences in the two approaches to the world that are not so marked as the consciousness business but where Collins and Reber had difficulty talking to each other. One is how language is acquired: Reber has argued that the mechanism of implicit learning allows infants to pick up the patterns of spoken signals and notes that there is a considerable literature supporting this notion [Reber 1993, 2011]. Collins believes we simply do not know how humans pick up the cues from social life that are required for fluency and, a fortiori, we have no idea how to build such a capacity into a computer. Reber thinks that the implied link between knowing a good deal about language learning and building such knowledge into a computerisn’t warranted.

45 Another difference turned on the rather unusual perspective on scientific truth of the sociology of scientific knowledge as it manifested itself in a discussion of parapsychology—which began as a sideline issue but quickly became the focus of an unusual kind of mismatch. Reber has a robust view of parapsychology—it is nonsense. Collins refuses to accept that as a useful professional attitude and spent a long time trying to convince Reber that the sociologist of scientific knowledge cannot begin by believing that one kind of scientific view is nonsense if he or she is to investigate the social forces that lead to it being widely seen as nonsense. Reber agrees that studying how scientific nonsense becomes recognized as nonsense is a useful endeavor but it is

Philosophia Scientiæ, 17-3 | 2013 123

still nonsense. Collins thinks that the sociology of knowledge way of thinking takes years of practice before it becomes natural and thinks that the difficulty of this aspect of the exchange arose out of the impossibility of conveying the approach through a few emails. Some of the main disagreement may have been flavoured by Collins’s history in sociology of scientific knowledge—anesoteric position.

Seven causes of misunderstanding

46 Toward the end of the exchange we were able to identify seven systematic causes of misunderstanding.

47 Mismatched thought styles: Reber generally sees continuities whereas Collins is drawn to bringing out sharp differences and classifications. Reber focuses on individuals, Collins on collectivities.

48 Semantic mismatch: The parties often use the same word but with different meanings without realizing it. It applies even to words at the very centre of the discussion—as we mentioned earlier, for Reber tacit is a synonym of “unconscious”, for Collins, “unspoken” or “unsaid”. Reber feels he is being loyal to Polanyi. Collins notes in TEK that he is deliberately going beyond what Polanyi intended. In some instances we were agreeing on the nature of tacit knowledge; in others disagreeing—and were often bewildered by the incoherencies that emerged. This kind of disconnect is dangerous because the terms are so familiar that it is hard to imagine they might mean something different to the other party.7

49 Mismatched explanatory adequacy: As Collins sees it, the parties justify claims in different ways. We saw this in a discussion of whether machines can be classified as potentially knowledgeable or non-knowledgeable. According to Collins, Reber works by building a consistent world view based on consciousness as inherently subjective and embedded in evolutionary theory. Collins claims there must be observable consequences and Reber does not have them. Reber thinks Collins is wrong and that he does have evidence for his position that is as strong or stronger than the evidence for Collins’s position. But the notion of mismatched explanatory adequacy remains useful even if, in this case, both parties are wrong about what the other is trying to do. There are many cases where parties disagree about what kind of grounding is needed to establish a scientific result.

50 Mismatched saliences: Negative mismatched saliences occur because to remedy an information deficit one needs an inventory of what is in the other party’s head so one can see what is missing: such inventories do not exist. Positive mismatched saliences affected the current exchange because one party continually explained at length what the other party already knew. For example, Reber explained the psychological equivalent of Dreyfus’s five-stage theory of expertise several times because Collins’s ignorance of it seemed, to Reber, the only way to make sense of aspects of Collins’s position. Positive mismatched saliences are very frustrating as they stop debates moving forward but they would probably be less marked in face-to-face conversation.8

51 Focus blindness: It is sometimes impossible to see a contribution that lies in the peripheral field of a strongly focussed gaze. On occasions one of us thought he had asked a certain question but the other did not see the question because it was outside

Philosophia Scientiæ, 17-3 | 2013 124

his view of the scope of their project. The exchange would continue on the assumption that the other had seen and appreciated the contribution. Confusion followed.

52 Reversion: Often one of us would explain an effect X to the other whose response made it evident that he understood. But the understanding was temporary. The problem was that X was held together in the longer term by a semantic net which included W, Y, Z, etc., and the whole structure only maintained its integrity through continual use. These W’s, X’s, Y’s and Z’s are like the spinning plates in a juggler’s act—if they are not kept spinning they fall. The dialogue, which was continued by one of the parties as though X is still in play, reverts to the earlier state of mutual incomprehension.9

53 Misplaced engagement: Often, to explore and explain two cross-cutting views of the world, one needs to be distanced from them. But because we were engaged in the worlds we were trying to explore, it was almost impossible for us not to slip, every now and again, into trying to convince the other that they were wrong—the traces are still in the text. Where possible the argument should come only after the mutual exploration.

Coda

54 The thing about incommensurables, at least in Kuhn’s classic view, is that they are, well, incommensurable. Perhaps not surprisingly, at the outset we disagreed on this issue as well. Collins, having explored it at length [Collins & Pinch 1982], [Collins, Evans, & Gorman 2007], felt that situations like this occurred often. Reber, armed only with a snifter of cognac, felt that dedicated scientists should be able to bridge whatever conceptual gaps divided them through careful reading, dispassionate reflection and intellectual empathy—accompanied by tolerance and trust. He’s not so sure now. We found tolerance and trust and we have tried our hardest to be clear, but there are still aspects of the way the other thinks that seem strange. We’ve ended up in a unique place: we have agreed to disagree even though, in more than a few cases, we aren’t completely sure what we’re disagreeing about or why. But we have learned something that we think is important. We have both been focused on the same end point throughout our intellectual lives, understanding the human mind, the cultures we have constructed around us and the manifold ways in which we function as individuals and as collectives. We understand there is no one best way to get there. In fact, we both suspect that it is from the conflicts, the misunderstandings, the incommensurables that we have a decent chance at real progress—so long as the parties are willing to respect and acknowledge the legitimacy of the other. We are now even more disdainful of gangs of academics whose identity and self-esteem is tied up with cleaving to one intellectual position and scorning those of others. We have come to see that all intellectual positions are likely to have something valuable to offer even if you cannot fully understand them and that the way forward is often a mixture. We also realize that we both understand a lot more than we did before we began this exchange.

Philosophia Scientiæ, 17-3 | 2013 125

BIBLIOGRAPHY

BIEDERMAN, IRVING 1987 Recognition-by-components: A theory of human image understanding, Psychological Review, 94, 115–147.

BRUNER, JEROME S., GOODNOW, JACQUELINE & AUSTIN, GEORGE 1956 A Study of Thinking, New York: Wiley.

COLLINS, HARRY 1974 The TEA set: Tacit knowledge and scientific networks, Science Studies, 4, 165–186. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK].

COLLINS, HARRY, EVANS, ROBERT & GORMAN, MIKE 2007 Trading zones and interactional expertise, Studies in History and Philosophy of Science, 4, 657– 666, special issue, edited by Collins, H.

COLLINS, HARRY & KUSCH, MARTIN 1998 The Shape of Actions: What Humans and Machines Can Do, Cambridge, MA: MIT Press.

COLLINS, HARRY & PINCH, TREVOR 1982 Frames of Meaning: The Social Construction of Extraordinary Science, Henley-on-Thames: Routledge & Kegan Paul.

KUHN, THOMAS S. 1962 The Structure of Scientific Revolutions, Chicago: University of Chicago Press.

NAGEL, THOMAS 1974 What is it like to be a bat?, The Philosophical Review, 83, 435–450.

REBER, ARTHUR S. 1967 Implicit learning of artificial grammars, Journal of Verbal Learning and Verbal Behavior, 6, 855– 863. 1993 Implicit Learning and Tacit Knowledge: An Essay on the Cognitive Unconscious, New York: Oxford University Press. 2011 An epitaph for grammar, in Implicit and Explicit Language Learning, edited by SANZ, C. & LOEW, R. P., Washington: Georgetown University Press, 23–34.

SEARLE, JOHN 1980 Minds, brains and programs, Behavioural and Brain Sciences, 3, 417–424.

ZIZAK, DIANE M. & REBER, ARTHUR S. 2004 The structural mere exposure effect: The dual role of familiarity, Consciousness and Cognition, 3, 336–362.

NOTES

1. The difference between the controlled experimental set up and the untidy and relatively uncontrolled social observation often gives rise to friction between psychologists and sociologists. In this case, however, both contributors were open-minded enough to admire the work of the other and the experimental/natural divide did not play a significant part in the debate.

Philosophia Scientiæ, 17-3 | 2013 126

2. AR: As derived from taciturn. 3. In TEK this criterion is what marks out “knowers” from non-knowers (knowers are not the same as entities that have knowledge—they must also be able to reflect on it). Initially, in this paper, Collins proposed the “vegetarian test”—to qualify to be a reflective species, sub-groups within it must be capable of self-consciously choosing to change their diet. The tacit-to-explicit test serves the same purpose but fits better with the topic of tacit knowledge and the notion of “knower” discussed in TEK. 4. Another influence on Collins’s approach to the debate is his later critique of artificial intelligence [Collins 2010], [Collins & Kusch 1998]. 5. Collins found Nagel’s article on being a bat disappointing [Nagel 1974]. Reber liked it. 6. In one exchange, Reber noted that there is research in perceptual and cognitive psychology that coordinates with Collins’s speculations. Irv Biederman developed a sophisticated model of perception based on 2D and 3D primitive forms like circles, cones and ellipses called “geons” that form the components of complex objects [Biederman 1987]. Real and stylized houses, such as in Collins’s figures, are easily abstracted as composed of just such forms. Biederman’s theory uses a surprisingly small number of geons (less than 40) to account for virtually all relatively fixed objects. 7. Collins recalls another instance when philosopher Martin Kusch and he spent months talking across each other while writing The Shape of Actions. The reason, as they eventually realised, was that they were using the central term “action” in different ways. To Kusch an action could be something accidental so long as it had consequences—the sort of things that law courts are interested because it is necessary to explore them in order to assign culpability. To Collins an action had to represent a society—it had to be something like taking out a mortgage or divining a witch—an accidental happening could not be an action however consequential it turned out to be. 8. The term “mismatched saliences” (used for the negative version) is drawn from Collins’s discussion of tacit knowledge (e.g., [TEK, 96]). Regarding positive mismatched saliences, in face- to-face conversation a quick technical comment can reveal that one of the parties already understands a set of issues so it is appropriate to move on. Collins uses the technique in interviews with scientists. 9. Locke said, you do not own something unless you mix your labour with it and this applies to concepts too. That is why good educational systems teach concepts via essays and seminars and why socialization is such an important part of education and essential to genuine interdisciplinary work. Collins once took a leading part on a seminar for natural science faculty that went smoothly for two years, the only problem being that the participants tended to ask the same questions at the end of the two years that they had asked at the beginning. Day-to-day everyone understood everything that was being argued, but it slipped away because they were not mixing their labour with it.

ABSTRACTS

Reber and Collins are each major researchers in psychology and sociology respectively. Both focus on the analysis and investigation of tacit knowledge. Yet neither had read or cited the other’s work. Here we explore how this similarity of interest can coexist in the midst of ignorance. Over many months we explored the differences in our world views, our approaches to

Philosophia Scientiæ, 17-3 | 2013 127

the topic and the difficulties of interdisciplinarity. This paper is a summary of that exchange presented as a kind of case-study in doing science. We conclude with a list of the general properties of the dialogue associated with this kind of “incommensurability” and state our distaste for tribalism in academic life.

Reber et Collins sont l’un et l’autre des chercheurs reconnus, respectivement en psychologie et en sociologie. Tous deux ont pour objet central d’intérêt l’analyse et l’investigation de la connaissance tacite. Pourtant, aucun d’eux n’a lu ou cité le travail de l’autre. Nous nous demandons ici comment cette proximité d’intérêt peut coexister avec cette ignorance. Pendant plusieurs mois, nous avons exploré les différences entre nos visions du monde, nos approches du sujet et les difficultés de l’interdisciplinarité. L’article est un résumé de cet échange, présenté comme une sorte d’étude de cas sur la manière de pratiquer la science. Nous concluons par une liste des propriétés générales que possède le dialogue associé à ce genre d’ « incommensurabilité » et faisons état de notre aversion pour le tribalisme dans la vie académique.

AUTHORS

HARRY COLLINS

School of Social Sciences, Cardiﬀ University, Wales (UK)

ARTHUR REBER Brooklyn College and the Graduate Centre of the City University of New York (USA) University of British Columbia, Vancouver (Canada)

Philosophia Scientiæ, 17-3 | 2013 128

Refining the Tacit

Harry Collins

General

1 For one’s work to be made the topic of a special issue of a journal is an enormous honour. That it is a philosophy journal makes the honour still greater since I am not a professional philosopher. Though I have no technical and scholarly training in philosophy, I have, however, learned hugely from a certain style of philosophical work, and from the start of my career in sociology, the later philosophy of Wittgenstein has been a dominant role model. Thus I thank the editors for providing this context and I thank my critics for being willing to allow the light-ray of their continuing work to be at least slightly diverted by the gravitational field of my book. Not that all the light rays have come especially close. Two of the “critiques”, those of Lynch and Pinch, are more in the way of saying “I’m over here and you should be over here too”. At the other extreme Turner has gone straight for the core of the book and tried to explode one of its most basic principles, while Soler & Zwart have tried, and to some extent succeeded, in moving its orbit. Tim Thornton is somewhere in between. I’ll start with the more distant beams and work inwards to Thornton, I’ll then discuss Turner and finish with Soler & Zwart.

1 Pinch and Lynch (and Doing) and the big shift

2 There seems almost to be a “Cornell School” of tacit knowledge because both Pinch and Lynch want me to switch all or some of my attention from what tacit knowledge is to how it is socially constructed or performed. I say a Cornell School because this was exactly the burden of the review of the book, published in 2011, in Social Studies of Science, by Cornell graduate student and researcher, Park Doing [Doing 2011]. Lynch was the editor of the journal at the time and I am surprised that he does not cite Doing’s article because his main point more or less exactly repeats what Doing says. Pinch’s article also goes in the same general direction and he too could well have cited the review. Both do cite earlier papers by Doing. So we can say that the Cornell School

Philosophia Scientiæ, 17-3 | 2013 129

favours a neo-Marxist approach, or interest theory, or something close, in the analysis of tacit knowledge. I will start, then, with the Cornell School, including Doing’s review which, as can be seen in the bibliography, is available online, and then go back to particular points in the Pinch and Lynch papers.

3 In his paper Trevor Pinch says that I am engaged in employing one of my favourite tactics, “compartmentalization”, in order to resolve the paradox of the relativism of the experimenter’s regress and the realism of the notion of there being some substance to tacit knowledge (and as he might have said, knowledge in general). It is true that compartmentalization is one of my favourite strategies but I do not think I am deploying it here. I have mainly employed compartmentalization to try to weaken the obsession of many in the field of science studies with reflexivity. Reflexivity, it seemed to me, threatened to paralyse everyone because they felt they could never put forward any sure finding about the natural sciences from their position within the social sciences without betraying their relativist credentials—given what they were saying about natural sciences they felt that they were not entitled to make positive claims. Malcolm Ashmore was the most obsessed and agonised but, at one time, it was a rapidly spreading affliction [Ashmore 1989]. So, just as Pinch points out, I said one should compartmentalize one’s activities into realism when one was being a social scientist but relativism in respect of the product—the analysis of science. I argued that without compartmentalization none of us could live even our ordinary lives.

4 But I do not think I try to resolve the apparent contradiction that Pinch points to via compartmentalization. In the early days of my work I was always deeply worried about that contradiction but nowadays I think I have a satisfactory solution: it is a matter of the direction of attention; if I pay attention to what is in front of me I see one thing, if I turn round and look in the other direction, I see another thing, but these things are not in tension, they are just different things.

5 To be clear, let us go back to the early days and see what was going on. In 1974, I published my first paper on the TEA-laser and in 1975, I published my first paper on gravitational waves: the second paper introduced the concept of the experimenter’s regress though I did not think up the name for it until this early work was drawn together into Changing Order (Collins). Shortly after the publication of those first two papers I realised that there was a contradiction at the heart of them: the realist 1974 paper was the basis of the relativist 1975 paper. For a long time this scared me and I kept quiet about it, waiting for the devastating criticism to come. But it never did come (at least, not until now). The worst of the crisis was over by about 1981 when I decided I was not a philosophical relativist but a methodological relativist and this took the logical force out of any contradiction.

6 Pinch’s critique has now forced me to think again. In 2002, Evans and I introduced the idea of three waves of science studies and, in particular, the new Third Wave. At the heart of the Third Wave was a move from the analysis of truth to the analysis of expertise and experience. The rest is history—a very unpleasant history to start off with and a quite pleasant and productive history now (e.g., see www.cf.ac.uk/socsi/ expertise ). I am now going to characterise the Third Wave move in a slightly different way: it was a shift in attention from the interpretative flexibility of words and things to the affordance (or fixedness) of words and things. It was a shift from looking at how procedures and languages didn’t work to how procedures and languages do work. There is no clash here. When I say to you, “pass me a knife”, it may be that you will interpret this as a

Philosophia Scientiæ, 17-3 | 2013 130

request for a dagger so I can stab my enemy in the course of a fight or it may be that you will pass me a knife to go with my fork so I can eat my dinner—such is the flexibility of language—but it is vastly less likely (though not completely impossible), that you will interpret the string of words as meaning “the top lead of a TEA-laser must be less than 8 inches long”—such is the affordance of language. The Second Wave concentrates on the flexibility of the interpretation while the Third Wave concentrates on the fixedness of the interpretation and these are no more in logical conflict than the view from in front of me and the view to the rear. Of course, there is some logistical tension because one cannot look in both directions at once: you have to choose what to look at. It was a failure to appreciate the lack of logical tension that led to the early unhappy history of the Third Wave (at least that is the charitable interpretation); we continually said that the Third Wave was compatible with the Second Wave but our critics were determined not to hear. The Third Wave must be compatible with the Second Wave because otherwise it would simply be the First Wave revisited—which is one of the things our critics say. The hard bit about the Third Wave is to make it possible to say something positive about the role of science without sacrificing all the wonderful insights that came to light under Wave 2. The first move in accomplishing this is turning attention to skill and expertise, the second is thinking about affordance instead of flexibility, and the third, if I have my way, is going to be looking at science as a moral enterprise (“elective modernism”).

7 I now want to recruit the 1974 paper, and my subsequent work on tacit knowledge and artificial intelligence into the Third Wave. I have, as it were, “been speaking prose all my life, and didn’t even know it!”, because the term, “Third Wave”, did not come along until 2002 and the idea of it being a potential shift of attention across all the social sciences and humanities did not come along until right now. So there is no need for compartmentalization because there is no logical contradiction to be resolved.

8 We are now in a position to make a general response to Pinch and Lynch and Doing. There are, of course, ways of analysing expertise as a matter of reinforcing professional interests, or “performing the notion of skill”, but to cleave to them would simply shift attention back to flexibility and away from affordance. Consider Trevor Pinch’s daughter: tell us, Trevor, was there sufficient interpretative flexibility associated with her performance for it to be the case that throughout her entire piano-playing career to date she had really been totally incompetent and only awarded those superior marks and gold cups because the teachers were scared of you as a powerful Cornell professor? There is no need for an answer. It is logically possible but at this point I am sure you, Trevor, are thinking of your daughter’s skill and the narrow affordance it offered to the assessor (playing the piano that beautifully she must have been at least “excellent” or “superior”), not the flexibility of interpretation so it is just a matter of what you choose to look at.

1.1 Pinch in detail

9 First I would like to thank Pinch for his generous introduction and description of the book. One could not ask more from one whose duties are criticism. Then I have to bite the hand that feeds me with a nasty technicality: At one point Pinch asks: does Collins want to buy into the “tacit knowledge regress”? The answer is that I have no choice since the “tacit knowledge regress” is the experimenter’s regress. I argued from the outset that the reason scientists could not be sure who had properly repeated another’s

Philosophia Scientiæ, 17-3 | 2013 131

experiment was because it required tacit knowledge to make an experiment work and tacit knowledge was invisible in its transmission and possession.

10 To continue gnawing at the fingers, I did not really understand what Pinch was arguing in his analysis of the two scrambled passages of text. “Yes”, there may be many other cues that we use when we decide whether or not to read on but this seems to me just pointing out some imperfections in the “demonstration” rather than anything deep. More important is the claim, and I am sure it is true, that some human editors would try to correct the text. This is a common response to critics of artificial intelligence but was answered in 1974 by Joseph Weizenbaum, the inventor of ELIZA. He pointed out that an AI device that was said to be capable of mimicking a human could not be one that would mimic an incompetent human because an electric typewriter just sitting and humming could mimic the capabilities of some humans. That some human editors would understand what was needed but no machines is all that is required for the example.

11 Pinch points out something that I had not noticed about the book: there are far fewer examples of Collective Tacit Knowledge than of the other kinds. Pinch wants to make this out as telling us something but the reason is mundane. The chapter on CTK was much harder to write than the others but only because I was so familiar with the material that it was hard to work up the energy; in contrast, the other chapters were exploring new things. Without using the term, I have been writing about Collective Tacit Knowledge for most of my academic life. Anything that uses language as the example is about CTK and anything that is a polimorphic action—and The Shape of Actions is full of examples of polimorphic actions—is an example of CTK. There are then, lots of examples of CTK in my work if there are few in TEK [Collins 2010 hereafter TEK]. Here, then, I think Trevor is “barking up the wrong tree”.

12 On cycling in the Netherlands and driving in traffic, Pinch, once more, invokes the difficulty of knowing whether there is real tacit knowledge at play or if it is just a way of belittling “outsiders”? Again, sometimes it will be the latter—there is enough interpretative flexibility—but unless Trevor thinks he could drive in China without alarms, and I certainly do not, there are also occasions when the example affords an analysis in terms of affordance. Whether you choose flexibility or affordance is up to you: in any case, I do not think social science allows you to tell an honest salesperson from a crook so I see no reason why the Third Wave should be able to do it. To use a concept does not require that one can infallibly identify all instances of its occurrence.

13 It is true that there is something more systematic to be said. When knowledge is under serious dispute, as in the circumstances of the experimenter’s regress, then interpretation is more easily afforded than affordance. And that is why it is so hard to work out how to make policy decisions in respect of disputed science. In Cardiff, our approach to policy has to be essentially normative: we think using the advice of experienced experts is better than tossing a coin but we cannot really explain why. The argument is based on a kind of self-evident concept of what is best where “best” is a moral term. In Cardiff we are now moving even further, toward “elective modernism” where scientific expertise is valued over other kinds of expertise because of the values and aspirations of the (ideal) scientific community.1

14 When the knowledge is not disputed an affordance approach is more readily afforded than a flexibility approach. Thus there is no real problem in the case of Trevor’s daughter. Trevor is a good enough appreciator of music to engage in the “downward

Philosophia Scientiæ, 17-3 | 2013 132

discrimination” (see Periodic Table of Expertises in [Collins & Evans 2007]), that enables him to understand that she is not bottom of the class and he is good enough at understanding our society and the role of experts in it to know that her teacher ought to be able to execute still more reliable downward discrimination. But, like any good “Wave 2er”, he also knows how to doubt it: he has two sets of resources. The first set is his training in sociology of scientific knowledge, which has taught us how to doubt anything we take a fancy to doubting. The second resource comes simply from the fact that, as explained in Rethinking Expertise [Collins & Evans 2007], those who are on the receiving end of downward discrimination tend not to recognise its force. During the UK’s public debate over genetically modified organisms (GMOs) a member of the public claimed that because gene manipulation was aided by radioactive markers, GMO crops would be radioactive. Even a person with as little knowledge of biology as me could tell that this was not true but I doubt the objector could be convinced: “they would say that wouldn’t they!” More seriously, it takes little expertise to understand that Andrew Wakefield’s claims about the relationship between MMR vaccine and autism were based on nothing and it was very easy to explain but, tragically, those who believed in the link remained unconvinced.

1.2 Lynch in detail

15 I start with an error made by Lynch that is also made by Turner (below). Lynch says, in a rather strange and convoluted sentence, that the core category of socially founded knowledge defies the very possibility of explication in a foreseeable future in his [Collins’s] version of tacit knowledge. [This volume, 62]

16 What I say is simply that we cannot foresee how CTK could be explicated—it is currently unforeseeable—but that does not make it impossible. Why does Lynch say that I deny “the very possibility of explication in a foreseeable future” when he could have said: “Collins says he can’t see how to do it”? It seems to be to prepare for the next paragraph which begins: Collins faces a formidable challenge in his attempt to specify types of knowledge that are beyond explication in any foreseeable future. [This volume, 62]

17 “Any foreseeable future”: what’s that? All I talk about is not currently being able to foresee. Furthermore, I neither promise nor make any attempt to specify types of CTK, I just say it is all CTK. Here Lynch simply fails to engage with the book which puts a lot of effort into exactly defining the meaning of “cannot”.

18 To get back to the main thrust of his piece, Lynch advertises that his “questioning extends to the very coherence and reality that Collins attributes to the overall ‘terrain’ of tacit knowledge”. That’s pretty exciting stuff so it is disappointing that he pulls back only a few lines later: […] its marginal relations to conceptions of explicit knowledge are crucial for understanding the meaning (the uses) of tacit knowledge—such relations constitute the relevance of tacit knowledge. Among the various uses of the concept of tacit knowledge are polemical uses that serve to enhance and defend the autonomy of academic and non-academic professions. Consequently, my aim in the remainder of this paper is to examine what is done with the notion of “tacit knowledge” rather than to define and classify what it is as a substantive domain. I will leave it to readers to decide if my treatment simply differs from Collins’s or whether it reveals a basic misconception of the phenomenon on his (or my) part. [This volume, 58]

Philosophia Scientiæ, 17-3 | 2013 133

19 It turns out to be left to the reader to decide how radical Lynch’s approach is. Well, I am a reader and it will be obvious by now that I do not think anything radical is going on here; what is going on is a Park Doing-like preference for looking at the way people use the notion of tacit knowledge to strengthen their claims to professional excellence. Of course they do! But people also use the concept of explicit knowledge to strengthen their claims to professional excellence. Go into your doctor’s office and you will see certificates on the wall attesting to the formal training they have received. Does that mean we need a radical reinterpretation of the notion of explicit knowledge? If we do it is not for want of trying. Is it any kind of revelation that professionals use their displays of qualifications indicating their explicit knowledge to place barriers to entry around their professional domains? I do not think this is new and I do not think it disqualifies the study of explicit knowledge and its kinds. Why should such a “revelation” work that way for tacit knowledge?

20 Lynch says that this interest theory interpretation of tacit knowledge would be: a shift away from philosophical debate about what can or cannot be made explicit, in principle or in practice, and toward studies of what counts—and, equally important, what does not count—as tacit knowledge in specific historical and contemporary circumstances. [This volume, 58]

21 Just to underscore what is going on, try the same claim on explicit knowledge: “an interest theory of explicit knowledge would be”: a shift away from philosophical debate about what can or cannot be tacit, in principle or in practice, and toward studies of what counts—and, equally important, what does not count—as explicit knowledge in specific historical and contemporary circumstances.

22 Sure, either of those projects are viable but they aren’t exactly radical. The “specific historical and contemporary circumstances” we have experienced since the middle of the last century are such that it was radical to do such an analysis half-a-century ago but now it is just going on the same old way. Such analyses are nowadays “normal science” in STS. It seems to me that in the particular historical location we find ourselves in now the radical thing for a social scientist of science to do would be precisely a shift toward a “philosophical debate about what can or cannot be made explicit, in principle or in practice” while trying to retain everything we know from the last half-century (Wave 2), about context-specificity. This would be a hard thing since Wave 2 has been looking in the other direction for half-a-century. The political and institutional position of Wave 2 in the universities is nowadays far more powerful and conservative than Mertonianism ever was and it is time for Wave 2ers to stop claiming they are radicals.

23 Lynch’s “radical” approach also takes us all the way back to Harold Garfinkel. Here is how he sets out the message of Garfinkel for tacit knowledge: […] the situational, interactional, and temporal relations that arise in the course of the practices of instructed action constitute the very relevance of tacit knowledge [TEK]. As Garfinkel’s students discovered when performing an exercise to follow directions for getting from a starting point to a destination in a city, the difficulties and contingencies they encountered were particular and mundane: they did not enter a “dimension” of tacit knowledge, instead they ran into a stream of difficulties during their ongoing attempts to reconcile the sketchy details of their directions with the unfamiliar and overwhelmingly dense “lived” course of the journey. [This volume, 60–61]

Philosophia Scientiæ, 17-3 | 2013 134

24 Followers of Garfinkel often seem to want to have their cake and eat it too. Look at the description of the students’ experience. It is not presented as something that happened to a particular group of students at a particular time. It does not say that “at 10.30 on 16 May, 1965, John and Jane Doe were instructed by Harold Garfinkel to follow an instruction that would take them from his office in UCLA to the seventh bus stop along the side of La Brea tar-pits and they ran into some difficulties following the instructions”. Instead it says having difficulties following instructions always happens! As Bruno Latour and Steve Woolgar explain in Laboratory Life [Latour & Woolgar 1979], the indicator that science—generalization—has been achieved is the removal of modalities (descriptions of the particular contingencies of persons, places and times) from the descriptions of experimental activity. Here we have experimental activity and the paragraph clearly points to the fact that the difficulty of following instructions described here is a general feature of all instruction-following, not a contingency that happened to John and Jane at 10.30 on 16 May 1965. Is Lynch founding the new ethnomethodological science of pointing or is he (Ashmore-like), simply using ethnomethology to cut away the basis for the very possibility of science without trying to found a new one? It seems to me that when Lynch turns the ethnomethodological critique on his fellow social scientists (as opposed to natural scientists), he should turn it on himself too and this would behove him to cease from social scientific generalization. But oddly, not only does he generalize about pointing, he goes on to follow Doing’s 2011 paper in proposing a neo-Marxist, or interest theory, sociology of tacit knowledge in which he relaxes his strictures on generalization and treats a study of UK Victorian doctors and a study of Australian doctors in the 1970s as showing the same thing which is, incidentally, applicable to the notion of tacit knowledge in general! We are a long way from John and Jane Doe’s actions on 16 May 1965. What is going on is that Lynch is “doing specific historical and contemporary circumstances”. He is using the accusation of lack of context specificity, in the flexible way all such accusations can be used, to rule his work in and other work out. This is a place where a bit more reflexivity would help rather than hinder.

2 Thornton

25 I like Thornton’s characterisation of my second two meanings of “explication”— building a machine to do the job and creating a scientific explanation—as “action at a distance”. I’ll use the phrase in my discussion of the Soler-Zwart critique. In the main, however, I suspect the right way to handle the disagreement between Thornton and I would be to engage in an exercise something like that found in the paper by Collins & Reber [this volume, 135–154]. We should spend a long time exploring what we each mean by knowledge, etc. Let us jump straight to Thornton’s concluding sentence: Collins’s broader account is of the nature of patterns or processes that might be known, not of the different ways in which they are known. It is pitched at the level of worldly patterns or ontology rather than the way they are known by epistemic subjects. Now that might be a perfectly fine area of inquiry. But it is not a discussion of tacit knowledge. [This volume, 106]

26 Those who read the Collins-Reber piece will see, what I think is, something similar: Reber, the psychologist is concerned with what is in peoples’ heads, Collins is

Philosophia Scientiæ, 17-3 | 2013 135

concerned with the nature of knowledge stuff and how it is transmitted and implemented.

27 Now I have to agree that what I am doing does not fit well with how most other analysts think about tacit knowledge and it happens that a good proportion of these other analysts are drawn from areas of psychology and philosophy which are closely allied. This close alliance is reinforced by a common concern with consciousness, qualia, and the like, all of which take the human experience as the central feature of the domain to be analysed. Hence, Thornton, if I understand him, not only has trouble with my notion of tacit knowledge but also any notion of tacit knowledge since when we have knowledge we are usually conscious of knowing something and this does not fit with the knowledge being hidden from us.

28 But what happened to me was that I just started to write about tacit knowledge, expecting, and promising others, to write a quick summary in four weeks, and found to my surprise, that I could not write about the tacit without understanding the explicit and, to my still greater surprise, that the explicit was the real mystery. This came about not from any determination to do things differently but because with my existing resources the words of the unfolding manuscript refused to lie quietly on the page.

29 I keep wanting to say that my treatment does not jibe with many of the psychologists and psychologically inclined philosophers because one of my central problems (after writing two books and many papers about artificial intelligence), is what kinds of knowledge-based performance can be reproduced by machines. But many of the group we are talking about also worry about artificial intelligence so a lot more disentangling would be needed to get to the heart of the matter.

30 I could have resolved my problems in a Thornton-like way. I could have started with language instead of strings. If I had started with language and signs, then there would be no particular difficulty about explicit knowledge—not the kind of difficulty I struggle with—because languages, signs, icons, and so forth are already invested with meaning. But Polanyi has it right—the way that they are invested with meaning is tacit. So when one is dealing with signs etc., one is not dealing with explicit knowledge, one is already dealing with the explicit mixed with the tacit. In that case, the relationship between tacit and explicit knowledge would turn out to be simple—explicit knowledge is already invested with tacit knowledge and there is nothing to resolve. It would be, as I say in the book, circular. And one can see how circular it is by asking whether it helps with the problem of what a machine can do and immediately see that it helps in no way whatsoever.

31 So, where Thornton says that what I call knowledge is not really knowledge at all, I say, in that case it is time to change the way we use the word “knowledge”. To classify tacit knowledge, to know the extent to which machines can mimic it, and to understand explicit knowledge, one must start from a different place. The way humans per se experience knowledge is one topic but the nature of tacit and explicit knowledge is another and to confuse them is to be confused about tacit knowledge too. I thought the example of bicycling in very low gravity had established this point.

32 I apologise for the fact that I also want to maintain the cross-cutting, and therefore confusing, idea that only humans should be called “knowers”. Humans are, in another sense, continuous with animals and sieves when it comes to using knowledge without talking about it but I still do not want to call animals and sieves knowers. This is a bit

Philosophia Scientiæ, 17-3 | 2013 136

messy. It is not a deep point, I just think that the term “knower” should be restricted to entities that are, in principle, capable of reflecting on what they are doing and are potentially capable of describing it. These are knowers in Thornton’s and Reber’s sense. But knowledge is a more general idea. You can have knowledge without knowing it and, therefore, without being a “knower”.

3 Turner

33 While responding to Turner’s elegant critique I will take the opportunity to mention his earlier review of TEK [Turner 2011] (Hereafter, T2011). T2011 starts with a generous assessment of this author’s qualities and flaws. It goes on to a comprehensive and nearly completely accurate description of what is in TEK. But the punchline, and never has the term punchline been more appropriate, is: […] both the starting point of the argument and its conclusion are groundless. So is most of what comes between. [Turner 2011, 475]

34 I could not help grinning. One of the most attractive things about our strange academic life is that it is possible to enjoy the work of one’s critics even as in the film, Jurassic Park, the white hunter could not but appreciate the stalking skills of the velociraptors that were about to rip him apart. Such things concentrate the mind. Note that the analogy stops at the concentration and before the ripping apart.

35 I now want to correct a couple of small errors in the exposition part of T2011. Firstly, there is meant to be much less difference between digital and analogue strings than can be read from Turner’s treatment. We experience them as different but I spend a lot of time saying that at the deepest level they are the same. Secondly, T2011 makes the same mistake as Lynch and others in thinking that I believe RTK and STK are of little interest because I call them “weak” and “medium” whereas I call CTK “strong”. I also say that CTK is the irreducible heartland of the problem. Thus in T2011 we find: Once the pretenders to the title “tacit knowledge” have been sent packing, the real thing we are looking for will be revealed. [Turner 2011, 474]

36 But, though CTK may be the most interesting category from a philosophical point of view, RTK and STK are always with us and are just as important from a practical point of view and mostly indistinguishable from CTK for practical purposes. Yes, they can be explicated in principle but I stress that only a bit of RTK can be explicated at any one time and that the explicated bit is a moving target. When we come to STK, it is just as necessary for balancing on your bike as it ever was, even though there is a sense in which it can be explicated. Indeed, in some ways RTK and STK are more interesting because, as ideas, they are more novel than CTK which was what was being explored a lot of the time under the old undifferentiated category of tacit knowledge.

37 T2011 advances two main objections to TEK’s position. The second objection is repeated in his contribution to this volume and we can deal with it under that heading. The first objection is as follows: The basic structure of the argument [of TEK] as a whole is this: Much of what passes for tacit knowledge, namely the individual part, can be done by or imagined to be done by machines; the rest cannot be done by machines, or at least machines can’t be fluent at it, so this remainder must be collective. The reasoning is transparently faulty. This is why: […] The fact that machines can simulate something that humans do tells us nothing about how humans do it. This is a point Collins himself makes. So the entire

Philosophia Scientiæ, 17-3 | 2013 137

discussion of string transformations is irrelevant: he gives us no reason to believe that what people do when they communicate has anything to do with strings, string transformations, or anything like it. This is just an analogy. [Turner 2011, 475]

38 As far as I can see, and very uncharacteristically, Turner has gone completely wrong here. It is true that I frequently help the argument along by imagining some action being carried out by a machine. I might say something like: so a salute is a mimeomorphic action whereas a greeting is a polimorphic action and one can see that it would not be difficult to build a saluting machine if you were crazy enough but there is no way of building a greeting machine.

39 Don’t worry about what salutes and greetings are, just notice the logic of the argument. It is not that the fact that a certain action can be mimicked by a machine proves it is mimeomorphic, it is that the fact it can be mimicked helps one to see that it is mimeomorphic while the reason it is mimeomorphic will have been explained by other means—e.g., we always try to carry out a salute in the same way. It could not be the other way round or a tractor would always be proving that pulling something was always a mimeomorphic action but it isn’t because when you have a tug-of-war with your lover you do it in a different way than with your deadly enemy. On the other hand, one is continually assailed by claims that the latest AI-based machines are executing polimorphic actions so the theory must be wrong. But then one has to examine these claims with special attention to discover if the machine is really doing the same thing as the human, which, to date, it never is.2 Turner seems to think I am using executability by machine in the way electricians use one of those little screwdrivers with a light in to find out if the current is live—it is either on or off. It can never be used this way but it is a powerful aid to thinking about these things—like, in Wales, looking out of the window to see if it is likely to rain in the near future. So that is why the first point in T2011 goes off the rails.

40 Then T2011 offers that this renders the entire discussion of string transformation irrelevant. I have no idea why Turner would say that. And he says that I provide no reason to believe that what people do when they communicate has anything to do with strings. But this seems completely bizarre. That communication is in terms of strings is almost definitional: it is just a physical description of what is going on causally when one person, say, speaks to another. How could I provide a reason for saying that when people communicate they use strings—I am simply picking one description of what they do as I might say, “when I look at a painting light is involved”. Turner is saying something like Collins gives us no reason to think light is involved. Of course, I know how to “deconstruct” the idea that light is involved but I do not think that is what is at issue here.

41 Now we can turn to Turner’s essay in this volume (75–92) and, by default, the second criticism in T2011. What Turner really does not like is the idea of the social collectivity as a fundamental unit of explanation. I say things like, “knowledge is the property of collectivities”, while Turner says collectivities do not exist, there are only social relations between individuals. Turner does valuable work in distinguishing between the two, noting that the word “social” is often used to cover these two very different things, something I have probably been guilty of myself. So let me make clear, as Turner also makes clear, that when we come to Collective Tacit Knowledge I am indeed talking of social collectivities, not just the social. Furthermore, I refer in TEK and elsewhere to “socialness”:

Philosophia Scientiæ, 17-3 | 2013 138

It is only humans who have the ability to acquire cultural fluency. It is only humans who possess what we can call “socialness”—the ability to absorb ways of going on from the surrounding society without being able to articulate the rules in detail. [TEK, 125]3

42 Therefore there is a very clear division between us and when I say, in the introductory remarks to this piece, “Turner has gone straight for the core of the book and tried to explode its most basic principle”, this is what I am referring to. It is the uncompromising nature of this attack that is so valuable because it forces us all to state our positions clearly, my contribution being to champion a very old fashioned position, something like that of Durkheim.4

43 The larger part of Turner’s critique in this volume is a very long and detailed analysis of how social and collectivity-based explanations work, along with a scorecard comparing each. I’ll adopt a Turner tactic and step back, trying instead to say what is going on in a simple way that I think Turner would agree with. Here we go! The collectivity approach provides parsimonious explanations of what we see going on in the world, with a non-parsimonious ontology—it introduces a “mysterious” new thing called the collectivity. It also finds great difficulty in explaining how individuals communicate with the new thing. The social approach has a parsimonious ontology, the only things it needs being individuals and their interactions. There is no special difficulty in explaining how individuals communicate with each other in the way there is difficulty in explaining how individuals communicate with the collectivity. And, crucially, it is possible to use the social approach to explain all that the collectivity approach explains even though it might be a bit more complicated.

44 What this reveals, I believe, is a deep difference in my approach and Turner’s approach to scientific explanation. I think this might reflect something different about our starting points (we have both ended by doing philosophy, or quasi-philosophy). My starting point is in fieldwork and practical problems whereas his starting point, as his critique exemplifies, is the philosophical canon: he deals on even terms with Kant and Quine whereas I am just thrilled to be mentioned in the same paragraph.

45 I think this difference might be closely related to what Reber and I call “Mismatched explanatory adequacy” [this volume, 135–154]. Turner and I just want different things from our explanations. What I value most is something that makes it easy for me to understand what I see going on around me while what Turner values most is something that does not introduce any unnecessary explanatory apparatus where it can be avoided. Indeed, Turner’s desire in respect of this is so strong that he not only eschews collectivities, he eschews the whole idea of tacit knowledge, wanting to replace it with habits (see, e.g., [Turner 1994]). If that is indeed the case there is not much else to do except say why I don’t share his desire and why I think my approach is better even though it does introduce a new, or at least quasi-Durkheimian, ontology. I do not think any of this is going to come out as a “proof” in the mathematical sense, just a discussion about the best way of going on. Therefore, what follows is more of a list of attempts to persuade, rather than an “argument” in the logical sense.

46 First, let us see if there are any successful scientific theories that might be said to have introduced suspect ontologies. Newton’s theory of gravity springs to mind. It invented something you could not see, smell or touch that implied the mysterious action at a distance. But it wasn’t a bad theory. We probably prefer Einstein’s theory because it does not involve action at a distance or a mysterious thing that you cannot see, smell or

Philosophia Scientiæ, 17-3 | 2013 139

touch but I don’t think we want to say that in the intervening years between the invention of the idea of Newtonian gravity and Einstein’s better conception it was only bad science that was going on even if the ontology was not parsimonious enough to meet certain philosophical standards. Though I would hesitate to compare myself with these famous men, I too am saying only that this equivalent to action at a distance (the notion of social collectivity) works (more on this below), and that there is nothing in the theory that says it could not one day be replaced by something better. I am very careful to claim only that at the moment we don’t know what it is. To repeat, I cannot prove that Turner is wrong in any strong sense, it might be an immensely complex set of individual interactions, but to insist that this must be so because of an ontological preference is to give too much authority to philosophy over science. It seems to me to condemn us to going along in the same old way that has been failing us for decades without any impetus to see to the heart of the problem and try to think in new ways.

47 By the way, early in our email exchange, Reber (this volume) wrote to me: [Y]ou realize, of course, that living in America gives a very different tinge to this broad idea of collectivism in all and any instantiation. This crazy land is still wrapped up in an individualistic ideology and makes ideas like yours suspect – and not just as a particular sociological model. [Arthur Reber, private communication, 25 Nov 2012]

48 We oughtn’t to forget the sociology of scientific knowledge entirely.5

49 Of course, we are paid by our universities to act as though “external” forces have no effect on our reasoning, so let us get back to the substance. The remaining points can be briefly set out. “Yes”, the collectivist approach has enormous difficulty with explaining communication between individuals and the collectivity. But to me that is its strength. All the problems that we encounter with, for example, interdisciplinarity, “incommensurability”, and the creation of intelligent machines that mimic polimorphic human actions, emerge precisely because we do not know how the communication works. I argue this over and over again. The reason that even a giant like IBM and Google cannot make a speech transcriber, or even a spell-checker, that works like a human being, is because we do not know how to plug intelligent computers into societies in such a way that they can suck their metaphorical blood. The developers of “intelligent machines” understand this in that they are now plugging their machines into the internet and, as a result, the machines are getting better at a restricted range of activities (such as IBM’s “Watson” computer winning at “Jeopardy”). But they are still subject to the most colossal errors when it comes to what, for humans, are the most simple tasks, such as keeping up with slang. Here is something from 7 January this year (2013), which is about as recent as I can make it: Two years ago, Brown attempted to teach Watson the Urban Dictionary. The popular website contains definitions for terms ranging from Internet abbreviations like OMG, short for “Oh, my God”, to slang such as “hot mess”. But Watson couldn’t distinguish between polite language and profanity—which the Urban Dictionary is full of. Watson picked up some bad habits from reading Wikipedia as well. In tests it even used the word “bullshit” in an answer to a researcher’s query. Ultimately, Brown’s 35-person team developed a filter to keep Watson from swearing and scraped the Urban Dictionary from its memory. But the trial proves just how thorny it will be to get artificial intelligence to communicate naturally. [http://tech.fortune.cnn.com/2013/01/07/ibm-watson-slang/]

Philosophia Scientiæ, 17-3 | 2013 140

50 Notice that it is IBM trying to make this work. We can assume they have pretty-well unlimited resources and, as the article says, they have a 35-person team working on it. We know that Google is trying to achieve similar things. We know that such corporations and research teams in universities have been trying to achieve what they are trying to achieve here ever since the 1960s, when it was first claimed that it had been done with ELIZA. Over and over again during half-a-century it has been claimed that it has been done. And yet, to date, there is nothing that can survive the Turing Test for more than a couple of conversational turns—or succeed in even one turn if the judge knows how to ask the question.

51 We have a choice. We can say that what is needed is understood, it is just a matter of going on the same way until we get there, or we can say it is time to rethink the problem in a more thoroughgoing way. Turner may believe that the social and non- collective approach provides all the answers but I say that this half-century of experience suggests to me that it is time to start again with sociology, and that is not the milk-and-water sociology of individuals in interaction but the true sociology of the notion of the collectivity as the basic unit of explanation. Try it! It explains everything that you see going on around you that touches on the frontiers of artificial intelligence by pointing out that we don’t know how to plug machines into society. It says that the problem of artificial intelligence, and I mean the “simple” problems, like building a human-like spell-checker, will not be solved by continuing in the current fashion. The social approach of Turner just does not give a reason why the huge effort over many decades has not worked; it just says do more of the same and we will get there. The collectivist approach may turn out to be wrong in the long term but even if it is, it concentrates the mind. And it concentrates the mind in the right place. It tells us what to look for when the next falsely-promised generation of fully “human-like” machines comes along. For those who are willing to learn the lesson, it tells how to build much more successful intelligent machines: give up trying to make humans and make “social prosthesis” [Collins 1990, 14], with clearly defined areas for humans to do the “repairs”. And it tells us that if we really want to make automated humans then it is their interface with society that is going to be the key. That is the reason I prefer the collectivist approach to Turner’s, and the extra object we have to purchase, like gravity for Newton, is well worth the ontological price we have to pay for it.

52 As I said, it gives me a thrill to be mentioned in the same paragraph as Kant and Quine even if it is the case that this “major figure” in the naturalization of science (thanks Steve) now finds himself inadvertently rooting for his one-time antagonist, Kant, instead of his proper champion, Quine. As explained in the introduction, I do not think there is any irony here, I think it is simply a matter of choosing to look one way rather than the other. But I have not been trained as a philosopher.

4 Soler & Zwart

53 Soler & Zwart have taken a very different approach to that of the other critics. They have started with what the book is trying to do and set out to discover technical errors, correct them, and do it better; their critique, as they so aptly put it, is written “from the inside”. There is a lot of enormously carefully worked out critical detail in the Solar-Zwart (SZ) paper and theirs is the only contribution that has caused me to think about doing some things differently. If there is ever a new edition of TEK, some of what

Philosophia Scientiæ, 17-3 | 2013 141

they argue will find its place therein, probably in the new material at front or back. What I particularly like about SZ is that they have not only found problems but tried to solve them in a positive way and in a way that leaves the original terminology as little altered as possible.

54 But SZ have adopted a very technical approach to the work, with complicated notations which make it difficult to read. Because what they say is important I will start by trying to pull out the main points of their argument in plainer language. There may be some losses but I think the subsequent discussion will be more accessible.6 The ordering in the list below does not correspond exactly to the ordering in the SZ paper nor to my subsequent discussion but I think it works.

4.1 Collins’s summary of main Soler-Zwart points

1. The “unrecognized knowledge” element of RTK has nothing to do with the relations between the parties trying to transfer the knowledge (in the way that secrets do have to do with relations between the parties). Therefore the term “relational” would be better replaced by another term such as “contingent” or “accidental”. 2. The reason that there is confusion is that the tacitness of STK and CTK is related to their location, in brain and muscles and society respectively, whereas RTK can cover any topic and its tacitness is related to various contingencies. 3. One implication of the idea of RTK is that the contingency is local. What is RTK in one place is explicit knowledge in another. 4. Elements of Somatic Tacit Knowledge can also be rendered explicit and whether they are is also an accidental or contingent matter. Therefore, in so far as Relational Tacit Knowledge exists there must also be relational Somatic Tacit Knowledge. 5. This opens up the possibility that there could also be relational Collective Tacit Knowledge. When we describe some of the differences between national driving habits we appear to be rendering some CTK explicit and this would exemplify the possibility. 6. The relations that can prevent a piece of RTK becoming explicit have sometimes to do with individual motivations and this is closer to the idea of social rather than collectivist (I am borrowing Turner’s vocabulary). 7. It is possible for there to be individual tacit knowledge—e.g., what might be in the head of a pioneering scientists who has just made a breakthrough but cannot explain it. Collins deals only with group tacit knowledge. 8. Given the reformulation that follows from the above clauses, it is possible to construct a new dynamics of the evolution of tacit knowledge into explicit knowledge.

55 I will start by eliminating points that I think are wrong or pull in unproductive directions and narrow down to the important points and what I think should be done about them.

4.2 Points that I do not think are helpful

Point 1: The “Relational” in RTK: should it be something like “Contingent”?

56 RTK began life as “Contingent Tacit Knowledge” just as SZ would prefer. It changed to RTK quite late in the process simply because I wanted it to have a distinct acronym and CTK was already in use for Collective Tacit Knowledge. Therefore I cast around for another term that would fit the requirement and came up with “relational”. I could

Philosophia Scientiæ, 17-3 | 2013 142

well have picked one of their suggestions: “accidental”. It might have been better if I had. But I do not think it is a matter of as much importance as they say it is. I think they are bit harsh on “relational” because they seem to have interpreted it more narrowly than I intended. They interpret it as to do with the relationship between the transmitter, A, and the receiver, or potential receiver, B, of a piece of knowledge but I always meant it has to do with relations between wider sets of human beings. Thus, if relations between scientists building lasers had been different what was the RTK of the length of the top lead might have been explicit while something else might have been tacit. In other words, where they say “history”, I say, “yesterday’s set of human relations”, so “relational” still applies. I think once it is seen that the relations are intended to be relations in the wider community rather than between “A and B” many of their objections to the term disappear. But it does not matter too much so long as we all still know what we are talking about.

Point 5: Can there be relational Collective Tacit Knowledge?

57 I am going to argue that their idea of relational somatic TK is important but that relational CTK is best forgotten about. The very essence of CTK is that it is the property of the collectivity and therefore cannot be explicated for the consumption of the individual. As SZ point out, however, it is possible to say something expository about CTK. Thus I can summarise the difference in driving patterns in different countries. But I believe this is best not looked at as the explication of CTK but as providing “hints”, or “coaching rules”, which are an aid to the acquisition of CTK. One reason is that this keeps the scheme neat and consistent. Another reason is that if we take this kind of thing as the explication of CTK we will go wrong. Searle’s Chinese Room is an example of how we could go wrong. Searle thinks he has invented a scheme that involves the explication of the TK embedded in natural languages. But he has not. What he has done is taken a frozen moment of language and represented that in look-up tables. The Chinese Room would degrade as fast as the language changes.7 The Chinese Room is best thought of as comprising a huge number of not very efficient coaching rules on how to speak Chinese. The same goes for traffic rules: at best, that is without even thinking about the rules regress, they represent a frozen moment in time and if one tried to apply them without also being a parasite on traffic society, one will find oneself at risk of going badly wrong.

Points 7 and 6: What should we do with the individual?

58 Removing the possibility of relational CTK makes everything simpler and so does removing individual TK—ITK as we might label it. ITK, as SZ put it forward, seems to have something in common with what Polanyi wanted to say about the special insight of scientists who could jump to a conclusion before they knew why. I am convinced that this ability exists. Indeed, I often cite my “instinct” that this will work out or that this won’t work out long before I know how to show it and I am mostly right. Nevertheless, I do not think that ITK should be part of the scheme even though it might be a good topic for psychologists or some kinds of philosophers. The reason is that the scheme is based in a collectivist (Turner’s “social” works just as well here)notion of knowledge.

59 This preference goes all the way back to the sociology of scientific knowledge where we discovered that the way to handle knowledge was to treat it as the product of society

Philosophia Scientiæ, 17-3 | 2013 143

and thus avoid any discussion of what it would be like to have a private science (or private language). Only by removing scientific knowledge from the discourse of “individual genius” could we find a way to analyse it. We could understand paradigms, which were social entities, but we did not have to deal with geniuses, cranks or eccentrics. This attitude carried forward very productively into the construction of the Periodic Table of Expertises [Collins & Evans 2007], where the meaning of expertise was equated with the acquisition of the tacit knowledge associated with an expert domain, not the expertise of individuals. Of course, some individuals are more expert than others within expert domains and I have tried to represent this with the Z-axis on what I call the “Expertise Space Diagram” [Collins 2013], but even this does not deal with the pioneering, or crazy, individual. I think individuals are best set aside for the sake of holding the world of tacit knowledge together.

60 To go back to point 6, though it may be the case that sometimes RTK is TK because of the personal interrelationships of individuals, I do not think this adds anything much to the general points being made.

4.3 Points that seem correct to me

Points 2 and 3: RTK differs in formation from STK and CTK and RTK varies with locality as well as with time.

61 SZ show that the reason RTK is tacit is contingent in respect of the topic it deals with whereas the reason that STK and CTK are tacit had to do with the substance of the knowledge. I had never “seen” this. In retrospect it is obvious because RTK is a residual category. The history of this is that I struggled and struggled to make some sense of the categories until I came up with the residual RTK that gathers together all the knowledge thathas no special reason for being tacit but is tacit, to use SZ’s term, for “accidental” reasons.

62 SZ have also done a great job in clarifying that RTK is local, not only temporally but also spatially. I think this was always implicit in the book but, as a result of their analysis, I have a much clearer image in my mind. I now see RTK as a wide river of tacit knowledge with islands of explicit knowledge popping up here and there and slowly developing into more extensive territories. I had the idea of RTK as a narrow stream, continually building land but without diminishing the flow of water. I now see that there is another dimension of variation which is possible because it is a wide river and that the extent and position of the new lands can be different across the flow.

Point 4: Some STK is also relational

63 That, for me, STK is also relational is SZ’s most telling point; I had completely missed it. It being the case that STK is explicated by doing science or engineering it is contingent on whether the science and engineering get done and in my acronym-preserving language it is, therefore, a matter of social relations between scientists and engineers.

4.4 A solution

64 I want to avoid anything as complicated as SZ’s notation because I want the concepts to have a wide audience. In the last two weeks I have spoken about these things to

Philosophia Scientiæ, 17-3 | 2013 144

audiences as diverse as software testing engineers and musicologists along with theatre directors and it would be a shame if the work were made difficult to access.

65 What we are trying to do here is characterise different types of tacit knowledge so that people can see what follows from their own involvement with the types. It seems to me that the one serious flaw that SZ have uncovered is the last one: the term “relational” covers both RTK and some of STK. But for the classification scheme to work it does not have to be based on only one principle. All we need to say is that RTK is simply relational whereas STK is both relational and somatic. In other words, the description becomes multi-stranded.

66 If we accept this, we can take the opportunity to add a third strand. Thornton calls the second two categories of explication that are found in TEK’s Table 4 [TEK, 81], “action at a distance”. That is a nice phrase because it captures the fact that in the case of the first two categories: “Explicable by transformation” and “Explicable by elaboration”, it is the communicator who mostly is in control of the process. In the case of the second two categories: “Explicable as mechanization” and “Explicable as [scientific] explanation”, it is mostly others who are doing the explication, not the communicators. I think the four categories do represent the ways we talk and think of something being explicated and, as SZ point out, there is no clash between them—we just talk of something being explicated in ways 1 and 2 or 3 and 4. Nevertheless, it is not a bad idea to bring out the different “feel” of two pairs of categories.

67 With these three strands in mind we can add some detail so as to characterise tacit knowledge as in Table 1.

Table 1: Multi-stranded description of types of tacit knowledge

RTK S-l-TK S-a-TK CTK

Relationality Relational Topic driven

Explicability Direct Distant None

68 The table retains the basic three-way division but enriches it following what we have learned. What I have given more salience to, however, is the distinction between Somatic limit and Somatic affordance Tacit Knowledge (S-l-TK and S-a-TK). This is because this division corresponds to other aspects of the multi-strand characterisation.

69 Somatic limit Tacit Knowledge, it will be remembered, has to do with the inability of humans to follow sets of rules such as describe how to balance on a bicycle because their brains are not fast enough (in other cases it might be that their bodies are unsuitable). Somatic affordance Tacit Knowledge, it will be recalled, is the opposite. It refers to the fact that certain activities can only be executed by entities constructed of the right materials—for example, the elasticity of my sinews may afford a delicate action such as picking up an egg without breaking it whereas this would be impossible to achieve in a robot made of metal alone. SZ correctly point out that STK is relational but it is not so clear that this applies to Somatic affordance TK. To build a machine that reproduces S-a-TK it would be necessary to reproduce the materials of the human body or brain and this is so much like copying, rather than reproducing in another form, that it does not seem like explication. After all, one does not say that photocopying a

Philosophia Scientiæ, 17-3 | 2013 145

printed page is explicating it! One might argue that if one could explain the qualities of the materials of body and brain that made it impossible to reproduce their performance using other materials, one might be doing explication of a scientific explanation kind, but it is a difficult point. So to keep things simple, S-a-TK will be kept with CTK in the inexplicable category, as indicated in the bottom line of Table 1.

70 S-a-TK and CTK are, as SZ point out, tacit in virtue of substance—one being located in the material of the body and one being located in society. But as the second line of the table indicates, RTK and S-l-TK both have a relational dimension.

71 The last line uses Thornton’s term to set out the way that this table relates to Table 4 in TEK. Much of the time, when RTK is being explicated, it is being done by the communicators in a “direct” way. When, however, it is being worked out that the top lead of the TEA-laser should be less than 8 inches long, then the explication is “action at a distance” or, simply, “distant”. The same applies to S-l-TK. Thus Table 1 preserves the main three-way distinction but is enriched to take into account aspects of the relationship between the three that my critics have brought out.

Point 8: The dynamics of explication

72 There are a number of ways in which I depart from SZ’s attempt to develop a dynamics of explication. As has already been discussed, I do not believe it is wise to talk of any of CTK becoming explicated; we do not want to wind up saying that the guide-book to France contains France’s culture nor that the French dictionary contains French. But there is a deeper and more revealing problem. SZ’s dynamics begins with what they call a “no-production and no-losses” clause. They say: “these two conditions are unrealistic, but we use them for the sake of analytical clarification”. I hate it when I put forward a thought experiment and someone replies that the assumptions are unacceptable; in a thought experiment one should be able to make any assumptions one likes. But the no- production and no-losses assumption actually changes the question. If the society we are imagining is completely static then it becomes identical to the frozen moment of language which is what (I claim) Searle’s Chinese Room represents. It is possible to explicate a frozen moment of a society in a Chinese Room type of way. So a dynamics that works on the basis of frozen moments is not telling us anything about CTK. The same argument applies to STK: I cannot imagine what it would be like for humans not to be continually developing new bodily skills.

73 My dynamics would have no explication of CTK and, I think, no explication of S-a-TK (for the reasons given above). Where explication was possible—RTK and S-l-TK—I think the new islands of explication in a wide river of tacitness is the right model.

Conclusion

74 What does TEK try to do? In my introductory chapter to this volume [–], I suggest that one of the things it does is to bring the extraordinary nature of explicit knowledge to the fore. But none of my critics have chosen to discuss the analysis of explicit knowledge in any detail. The second thing that TEK tries to do, which was also explained in the introductory essay, is to be an antenna that can pick up the ephemeral signal of the otherwise invisible, tacit knowledge. It has been 50 years since Polanyi introduced the term and, though it has been used, notably in management and science

Philosophia Scientiæ, 17-3 | 2013 146

studies, and investigated (see Reber’s work in Collins & Reber, [this volume, 135–154]), the concept has remained mysterious; people worry about how you can say anything about a concept that has no visible contents. In so far as philosophers have explored the concept, their aim has often been to try to do away with it; Turner, for example, argues that we do not need the concept, all we need is the idea of habits.8 TEK tries to show, for what I believe is the first time, that different things can be meant by tacit knowledge and that there are three basic kinds. It takes the concept seriously and tries to analyse it and move it forward. It tried to give us something to discuss, not simply to dismiss.

75 Unfortunately, in this volume, only Soler & Zwart really take up the challenge. They take it up so effectively that they uncover some serious problems. What I have tried to do is fit their findings into the scheme in such a way that the original three-way classification is enriched rather than weakened. In any case, I still think the three-way classification still exhausts the world and cuts it at, or very close to, the joints. This exercise, thanks principally to SZ, has revealed that one can cut things up in a slightly more refined way and cook a still more satisfying meal.

BIBLIOGRAPHY

ASHMORE, MALCOLM 1989 The Reflexive Thesis: Wrighting Sociology of Scientific Knowledge, Chicago: Chicago University Press.

COLLINS, HARRY 1985/1992 Changing Order: Replication and Induction in Scientific Practice, Chicago: University of Chicago Press, 2nd ed., 1st edition Sage publications, 1985. 1990 Artificial Experts: Social Knowledge and Intelligent Machines, Cambridge, MA: MIT Press. 1998 Socialness and the undersocialised conception of society, Science, Technology and Human Values, 23, 494–516. 2010 Tacit and Explicit Knowledge, Chicago: University of Chicago Press, [TEK]. 2013 Three dimensions of expertise, Phenomenology and the Cognitive Sciences, 12(2), 253–273. doi: 10.1007/s11097-011-9203-5.

COLLINS, HARRY & EVANS, ROBERT 2007 Rethinking Expertise, Chicago: University of Chicago Press.

DOING, PARK 2011 Tacit knowledge: Discovery by or topic for science studies?, Social Studies of Science, 41(2), 301–306. doi:10.1177/0306312710397690.

FORMAN, PAUL 1971 Weimar culture, causality, and quantum theory, 1918-1927: Adaptation by German physicists and mathematicians to a hostile intellectual environment, Historical Studies in the Physical Sciences, 3, 1–115.

Philosophia Scientiæ, 17-3 | 2013 147

LATOUR, BRUNO & WOOLGAR, STEVE 1979 Laboratory Life: The Social Construction of Scientific Facts, London; Beverly Hills: Sage.

NOTES

1. In Mertonianism, the efficacy of science came first and the values followed—it is only under these norms that science can be efficient. In elective modernism, the values come first irrespective of efficacy. 2. It was Arthur Reber’s suggestion that cars that could drive themselves had been developed that led to the paper by Collins & Reber [this volume, 135–154]. Soler-Zwart also mention this possibility (see their paper note 8). Examined closely, such cars do not do drive themselves in a human-like way but are one more example of the gap between artificial intelligence and what it sometimes claims for itself. 3. See also [Collins 1998]. There is a possibility of terminological confusion here since Turner talks of the social as meaning the complex relationships between many individuals whereas I am implying something more “ontological” with the term “socialness”. 4. And interestingly, an instance of yet another disagreement between myself and Latour (though it is really the old disagreement in another form), since Latour is nowadays championing Gabriel Tarde, who was Durkheim’s rival, precisely in the matter, as Turner points out, of insisting on analysis in terms of the social as opposed to the collective. 5. For a classic instance of this kind of explanation see [Forman 1971]. 6. Soler (private communication) says that aside from point 2, these points are a close but not exact fit to their argument but they are close enough to support what follows. In the case of point 2 they say their important argument has to do with the difference between reasons and topics. I find their point hard to follow and, on my reading, my response to what they say is apt. Readers will have to decide if I have missed the deeper point both here and in other places. I would not be surprised if our dialogue were to continue. 7. Continually updating it merely seems to solve the problem by passing it to human updaters who are embedded in society. 8. Thornton also wants to question the very possibility of a tacit knowledge and there are many more philosophers working along the same lines.

AUTHOR

HARRY COLLINS

School of Social Sciences, Cardiﬀ University, Wales (UK)

Philosophia Scientiæ, 17-3 | 2013 148

Varia

Philosophia Scientiæ, 17-3 | 2013 149

Unix selon l’ordre des raisons : la philosophie de la pratique informatique

Baptiste Mélès

NOTE DE L'AUTEUR

L’auteur remercie les relecteurs anonymes pour leurs précieuses remarques.

1 Introduction : la philosophie de la pratique informatique

1 La méthode est parfois féconde en philosophie des sciences, de chercher si les concepts techniques relèvent d’une nécessité conceptuelle ou des hasards de l’invention.

2 Telle était déjà la gageure de la Philosophie de la nature de Hegel [Hegel 2004], qui rapportait à leur construction logique l’ensemble des concepts par lesquels l’esprit appréhende ce qui lui paraît au premier abord le plus extérieur et le plus étranger : la nature. Ce projet n’était peut-être pas si stérile qu’on a voulu le croire. Jean Cavaillès [Cavaillès 2008a], puis le Jules Vuillemin de la Philosophie de l’algèbre [Vuillemin 1962], se sont efforcés, en mettant au jour le contenu philosophique immanent aux concepts scientifiques, de comprendre quelle était la part de nécessité interne — c’est-à-dire dans quelle mesure la structure de la connaissance elle-même justifiait le cheminement de l’invention — dans le développement d’une science en un moment charnière de l’histoire des mathématiques.

3 La méthode est aisément transposable dans l’encore jeune philosophie de l’informatique. L’idée même d’une philosophie de l’informatique est parfois encore accueillie avec surprise ; mais si l’on est prêt à concéder que les concepts mathématiques présentent quelquefois de quoi piquer l’intérêt du philosophe, que les pratiques et les

Philosophia Scientiæ, 17-3 | 2013 150

œuvres des mathématiciens ne sont pas moins passibles d’un examen philosophique que leurs concepts, et si l’on est prêt à appeler cela une philosophie des mathématiques, nous espérons montrer que les concepts, les pratiques et les œuvres des informaticiens peuvent faire l’objet d’une philosophie de l’informatique tout aussi instructive.

4 Peut-être au passage délivrera-t-on les sciences de l’ingénieur en général de certain préjugé : de ce qu’elles sont applicables, on croit pouvoir les conclure empiriques. Mais à qui prend la peine de les examiner, elles n’apparaissent guère moins belles que les mathématiques pures : bien loin que l’esprit s’y soumette à la matière, c’est la matière qui n’y existe qu’à titre de construction de l’esprit ; leur monde est celui de Berkeley. La philosophie de l’informatique en particulier n’est donc pas celle d’une technique sans âme — à supposer qu’il en existe.

5 Il n’est d’ailleurs nul besoin de se borner aux abstractions de l’informatique théorique pour montrer que l’informatique n’a pas les mains moins propres que les mathématiques pures : la philosophie de la pratique informatique, c’est-à-dire l’étude de la pratique qu’est la programmation et des œuvres que sont les programmes, n’est pas moins instructive que la philosophie de l’informatique théorique développée par les logiciens. Nous essaierons de présenter ici une application de cette méthode en étudiant si l’on peut fonder en raison les deux notions fondamentales des systèmes d’exploitation que sont les concepts de fichier et de processus.

2 Une ontologie d’actes et d’objets

6 Les programmes que nous utilisons au quotidien — traitements de texte, navigateurs, etc. — ne sont pas écrits dans les langages abstraits des théoriciens de l’informatique tels le lambda-calcul ou les machines de Turing, mais dans ce que l’on pourrait appeler des langages « naturels » de programmation comme le C et le C++. Or tous ces programmes reposent en dernière instance sur un programme fondamental, qui administre directement le matériel informatique — processeur, mémoire vive, disques durs, etc. — pour le rendre accessible à l’utilisateur humain : le système d’exploitation. Le système d’exploitation est en ceci le programme des programmes.

7 Or toute étude des systèmes d’exploitation commence par une exposition des concepts de fichier et de processus. C’est ainsi que Andrew Tanenbaum, créateur en 1987 du système d’exploitation MINIX, dérivé simplifié d’Unix, écrit dans la deuxième édition d’Operating Systems : « Les appels système MINIX se répartissent en gros en deux grandes catégories : ceux qui se rapportent à des processus et ceux qui se rapportent au système de fichiers [Tanenbaum & Woodhull 1997, 15]2 ». L’auteur entreprend ensuite de « passer en revue » ces concepts, l’essentiel étant pour lui, et pour ses étudiants informaticiens, d’en venir le plus vite possible aux problèmes et solutions concrets qui en dérivent — système de fichiers, arborescence, droits, lecture et écriture des fichiers, hiérarchie et priorité des processus, états, signaux, fils d’exécution, ordonnancement, parallélisme. Dans le cadre de l’exposé technique, il suffit que les notions semblent tomber du ciel, sans que l’auteur ait à déterminer si elles ont été dictées par Dieu, mathématiquement démontrées, sélectionnées pour leur efficacité pratique, ou simplement sédimentées par la tradition ; il suffit d’introduire une notion nouvelle par la formule « Un autre concept clé présent dans quasiment tous les systèmes d’exploitation est... » [Tanenbaum 2003, 41].

Philosophia Scientiæ, 17-3 | 2013 151

8 Mais ces formules ont des parfums d’empirisme qui ne sauraient satisfaire le philosophe. On aimerait savoir pourquoi l’exposition des systèmes d’exploitation passe par ces notions précises, si cela a toujours été le cas et le sera toujours. Pourquoi le processus et le fichier sont-ils donc les notions fondamentales de ces programmes, eux- mêmes fondamentaux, que sont les systèmes d’exploitation ?

2.1 Les processus comme actes

9 Qu’est-ce d’abord qu’un processus ? Le terme est souvent défini comme « l’instance d’exécution d’un programme » [Tanenbaum & Woodhull 1997, 15], [Bovet & Cesato 2001, 10], [Tanenbaum 2003, 37], [Lions 1996, 6–3c]. Mais la définition est philosophiquement peu éclairante en ce qu’elle présuppose la connaissance de ce qu’est un programme ; et techniquement inexacte, car même si ce cas est le plus fréquent, rien n’exige que le processus dérive d’un programme écrit à l’avance.

10 John Lions — auteur en 1976 d’un commentaire du code source de la version 6 d’Unix qui fut aussitôt reconnu comme classique, et qui a longtemps circulé sous le manteau avant d’être enfin distribué officiellement [Lions 1996, 3] — a bien vu que le problème de définition du processus dépassait la simple technique et relevait de la philosophie : Fournir une définition généralement acceptable de la notion de « processus » soulève de nombreuses et de sérieuses difficultés. Elles s’apparentent à celles qu’affronte le philosophe voulant répondre à la question « qu’est-ce que la vie ? ». Nous serons loin d’être les seuls à mettre légèrement de côté les points les plus subtils. [Lions 1996, 7–1a]

11 C’est donc précisément ici que le philosophe prend le relai de l’informaticien : il lui faut cueillir les concepts à la racine.

12 L’hygiène la plus élémentaire en philosophie de la pratique informatique requiert d’adopter le principe méthodologique suivant : l’essence des notions réside tout entière dans leur code source. La réponse à la question de savoir ce qu’est un processus est donc à chercher dans la table des descripteurs de processus, qui résume l’ensemble des propriétés essentielles que le système d’exploitation a besoin de connaître et de manipuler.

13 Le code source sur lequel nous nous appuierons principalement est celui de la version 6 d’Unix (1975). Pourquoi ce système en particulier ? D’abord, parce que son code source a été intégralement publié ; ensuite, parce qu’il comprend toutes les caractéristiques d’un système Unix mature ; enfin, parce que sa briéveté exceptionnelle — 9000 lignes — lui permet d’être appréhendé dans sa totalité par un seul être humain. Nous avons donc la chance rare de pouvoir étudier un système libre, complet et minimal3. Le premier critère exclut les systèmes propriétaires tels que Microsoft Windows et MacOS ; le deuxième, les toutes premières versions d’Unix ; le troisième, les systèmes libres nés dans les années 1990 comme GNU/Hurd (commencé en 1990), Linux (1991) et les systèmes BSD récents (FreeBSD, NetBSD et OpenBSD, 1993–1995). On aurait en revanche tout aussi bien pu étudier MINIX, dont le code source n’est que trois fois plus long que celui de son ancêtre Unix.

14 Examinons donc la structure de données que cette version d’Unix décrit en langage C dans le fichier d’en-tête proc.h [Lions 1996, 7–2] :

Philosophia Scientiæ, 17-3 | 2013 152

Langage C

15 Un processus y apparaît comme une structure (struct proc), c’est-à-dire un « faisceau » de données composé en particulier d’un identifiant numérique unique ou PID (p_pid), de l’identifiant d’un certain utilisateur du système (p_uid), d’un certain état (p_stat — ces états, définis aux lignes 382–387, déterminent notamment si le processus est en cours de création, d’exécution, de sommeil, ou de terminaison), d’un certain niveau de priorité (p_pri), de certains fanions (p_flag) qui sont des propriétés binaires (les différents fanions sont définis aux lignes 391–396), etc. La structure de données proc définit ainsi quelles sont les propriétés élémentaires d’un processus ; celles-ci se sont certes considérablement enrichies depuis Unix 6, comme on peut le constater en comparant sa taille croissante dans les codes sources de MINIX 2.0 [Tanenbaum & Woodhull 1997, 580–581], Linux 2.2 [Bovet & Cesato 2001, 65–66], et la taille qu’elle occupe actuellement dans Linux 3.2 ; mais toutes les propriétés originelles sont demeurées intactes.

16 A-t-on tout dit du processus quand on a décrit la structure de données qui lui est associée ? Certes non : en informatique comme en ontologie, le tout n’est pas de savoir ce que sont les choses ; il faut savoir si elles existent, et ce qu’elles peuvent devenir. Aussi faut-il examiner, comme le fait méthodiquement la théorie de la démonstration depuis Gentzen [Gentzen 1955], et comme le proposent dans un autre contexte Raymond Turner et Amnon Eden [Turner & Eden 2007], comment ces objets sont d’abord introduits et ensuite éliminés. Tels sont les actes canoniquement associés aux objets.

17 Un processus standard est toujours introduit, sous Unix, par un appel à la fonction newproc() (commentée dans Lions [1996, 6–4a et 4-5]), définie dans le fichier slp.c (lignes 1826–1919). Cette fonction ajoute un nouvel élément à la table des descripteurs de processus, le munit d’un nouvel identifiant (p_pid), et enregistre l’identifiant du processus qui l’a créé (p_ppid). Le processus est donc un acte engendré par un acte antérieur, selon une chaîne — ou plus précisément un arbre — de mises en branle.

18 Tout naturellement se pose la question du premier moteur. Le processus init joue pour ainsi dire le rôle de la causa sui, voire d’une création continuée. Dans Unix 6, la structure de processus est initialisée — c’est-à-dire passe de l’essence à l’existence — dans la fonction main() du fichier main.c, qui est la partie principale du programme :

Philosophia Scientiæ, 17-3 | 2013 153

Langage C

19 Cette initialisation (commentée dans Lions [1996, 6–3c–6–4a]) détermine d’abord l’adresse (p_addr) et la taille (p_size) du segment de données associé au processus 0 ; elle attribue ensuite à ce processus l’état SRUN, c’est-à-dire « prêt à être exécuté » (ready to run) ; elle indique enfin par les fanions (flags) SLOAD et SSYS que le processus initial réside et résidera toujours dans la mémoire centrale. Dans Linux 2.2, l’ancêtre de tous les processus, appelé le processus 0 [...], est un thread noyau créé à partir de rien durant la phase d’initialisation de Linux par la fonction start_kernel() [...]. Le thread noyau créé par le processus 0 exécute la fonction init() [...]. Puis init() invoque l’appel système execve() pour charger le programme init. [...] Le processus init ne se termine jamais, car il crée et surveille l’activité de tous les processus qui implémentent les couches externes du système d’exploitation. [Bovet and Cesato 2001, 93–95]

20 Les processus sont ainsi « créés », ou plutôt mis en branle, par d’autres processus, dont l’ancêtre ultime est un premier moteur créé ex nihilo, créateur et maître de lui-même et de toutes choses. Contingents comme le sont les actes ici-bas, ils sont « détruits » par un appel système ou en recevant un certain signal envoyé par un processus habilité à le faire : ainsi le Créateur nous rappelle-t-il à lui. Les processus sont donc des actes possédant un certain faisceau de propriétés, amorcés et interrompus par d’autres actes. Et, comme « les seules entités actives d’un système Unix sont les processus » [Tanenbaum 2003, 730], la réciproque est vraie : tout et seulement ce qui agit est processus.

21 La notion de processus est en informatique l’exact équivalent de ce que la notion d’acte est en ontologie. À proprement parler, les processus n’« existent » pas : ils opèrent. Ce sont des actes et non des objets. John Lions n’était donc pas si mal avisé, laissant la question aux philosophes, d’apparenter la question « qu’est-ce qu’un processus ? » à « qu’est-ce que la vie ? ».

2.2 Les fichiers comme objets

22 Qu’est-ce maintenant qu’un fichier ? Conformément au principe énoncé plus haut, observons la structure de données associée au concept de fichier dans Unix 6 afin de dégager ses propriétés essentielles.

Philosophia Scientiæ, 17-3 | 2013 154

Langage C

23 Un fichier est ici défini comme une entité identifiée par un certain numéro (i_number), appartenant à un propriétaire (i_uid) selon des droits d’accès déterminés (i_mode), située à une certaine adresse (i_addr) d’un certain périphérique (i_dev) et occupant une certaine longueur (i_size0). Cette structure de données est introduite et éliminée par les fonctions de création et d’effacement (creat() et unlink(), lignes 5781–5796 et 3510–3535), et manipulée par les fonctions d’ouverture et de fermeture (open() et close(), lignes 5765–5775 et 5846– 5855), de lecture et d’écriture (readi() et writei(), lignes 6221–6320). La définition habituelle du fichier est donc cette fois légitime : « Un fichier Unix est une suite éventuellement vide d’octets contenant une information quelconque » [Tanenbaum 2003, 773]. Le fichier est un objet que certains actes peuvent créer, qui demeure stocké sur un périphérique, et que de nouveaux actes peuvent manipuler sous certaines conditions.

24 Le fichier est un objet, soit ; mais on peut aller bien plus loin en décrétant que tout objet est un fichier. Unix joua en effet un rôle majeur dans l’histoire des systèmes d’exploitation en généralisant ce concept avec la notion de « fichiers spéciaux » : Les fichiers spéciaux constituent la fonctionnalité la plus inhabituelle du système de fichiers d’Unix. Chaque périphérique d’entrée/sortie supporté par Unix est associé à au moins un fichier de ce type. Les fichiers spéciaux sont lus et écrits exactement comme les fichiers normaux sur disque, mais les demandes en lecture et écriture entraînent une activation du périphérique associé. [Ritchie & Thompson 1974, 367]

25 Tout objet, sous Unix, peut être considéré comme un fichier. Ce principe vaut évidemment pour les fichiers texte et les fichiers binaires que tout utilisateur de systèmes d’exploitation modernes manipule au quotidien — documents, images, films — mais également pour des objets moins triviaux. Il suffit de taper « ls /dev/ », ou plus encore « ls /proc/ » ou « ls /sys/ » à l’invite de commande d’un système de la famille Unix pour constater avec surprise l’ampleur de ce qu’il considère comme fichiers. Le processeur, les disques durs, le clavier, l’écran, les informations relatives au processeur, la version du noyau ne sont que des fichiers. Même l’imprimante ! Le système d’exploitation en lui-même ignore tout de ce qu’est une imprimante : imprimer n’est pour lui rien d’autre qu’écrire dans le fichier /dev/lp0, dont il ignore tout, laissant aux pilotes de périphériques le soin de traduire le message dans l’idiome de ce que nous autres humains savons être une « imprimante ». De même, le scanner

Philosophia Scientiæ, 17-3 | 2013 155

n’est pour lui qu’un fichier à lire ; la carte son, un fichier dans lequel on écrit ; le disque dur tout entier, un fichier que l’on lit et modifie. Traiter de cette manière les périphériques d’entrée/sortie présente un triple avantage : les entrées/sorties de fichiers et de périphériques sont aussi similaires que possible ; les noms de fichier et de périphérique possèdent la même syntaxe et la même signification, de sorte qu’un programme attendant comme paramètre un nom de fichier puisse recevoir un nom de périphérique ; enfin, les fichiers spéciaux sont sujets au même mécanisme de protection que les fichiers ordinaires. [Ritchie and Thompson 1974, 367]

26 On voit à quel point la notion de fichier qu’a adoptée Unix est éloignée de l’intuition familière que nous en avons. Pour Unix, non seulement tous les fichiers sont des objets manipulés par le système d’exploitation, mais tous les objets manipulés par le système d’exploitation sont des fichiers.

27 Si l’on entend par ontologie la théorie de ce qui est, le seul et unique engagement ontologique du système d’exploitation Unix et de tous ses descendants est donc que tout ce qui agit est processus, et que tout ce qui existe est fichier. Le couple que forment fichiers et processus ne tombe pas du ciel, mais d’une ontologie strictement composée d’actes et d’objets.

3 La catégorie des fichiers et des processus

28 Cette ontologie fonde une interaction. Une analyse du shell montrera que les propriétés caractéristiques de cette interaction sont exactement celles de la théorie mathématique des catégories.

29 Dans l’invite de commande standard d’Unix qu’est le shell, les processus ont le clavier pour entrée standard (STDIN), et l’écran pour sortie standard (STDOUT) ; c’est-à-dire que tout processus, en l’absence d’indications contraires, recevra ses arguments du « fichier spécial » qu’est le clavier (et son périphérique caché qu’est l’être humain), et enverra ses résultats dans le fichier spécialqu’est l’écran.

30 Prenons pour exemple la commande Unix cat, dont la fonction est simplement de transcrire sur la sortie ce qu’elle reçoit sur l’entrée. Si l’utilisateur tape simplement « cat », le processus n’affichera rien en attendant de recevoir une entrée ; si l’utilisateur tape du texte puis appuie sur la touche d’entrée, le texte est simplement recopié : invite_de_commande $ cat Il y a de l'écho. Il y a de l'écho. invite_de_commande $

31 Par défaut, un processus comme celui qui précède est donc un acte

32 Le shell propose cependant des redirections permettant aux processus d’écrire de n’importe quel fichier vers n’importe quel fichier (Ritchie and Thompson 1974, 371). Le caractère < permet de forcer l’utilisation d’un certain fichier comme entrée, le caractère > de forcer l’utilisation d’un certain fichier comme sortie. On peut ainsi demander à la commande cat de copier le fichier entree.txt dans le fichier sortie.txt :

Philosophia Scientiæ, 17-3 | 2013 156

invite_de_commande $ cat < entree.txt > sortie.txt invite_de_commande $

33 La commande peut échouer pour de nombreuses raisons, typiquement si le fichier d’entrée n’existe pas ou que le fichier de sortie soit protégé en écriture : la sortie d’erreur STDERR, par défaut l’écran, existe pour informer l’utilisateur de l’échec du remplissement de l’acte qu’il ambitionnait. Ainsi, au moins dans son intention, un processus quelconque peut être transformé en acte utilisant n’importe quel objet pour produire n’importe quel objet :

34 Naturellement, le fichier de sortie d’un programme peut être utilisé comme entrée d’un nouveau programme, par une simple séquence d’instructions notée par l’utilisation du point-virgule. En utilisant la commande cat suivie de la commande wc -l, qui compte le nombre de lignes d’un fichier, on peut construire l’instruction suivante : invite_de_commande $ cat < A > B ; wc -l < B > C invite_de_commande $

35 Cette commande écrit le contenu du fichier A dans le fichier B, puis compte le nombre de lignes du fichier B et inscrit le résultat dans le fichier C. L’ensemble de la commande peut donc être vu comme un diagramme

36 Mais on peut préférer, par souci d’élégance et pour éviter de laisser éparpillées les traces d’étapes intermédiaires de calcul dont nous n’aurions plus besoin par la suite, ne créer les objets intermédiaires qu’à la volée au lieu de les écrire comme d’authentiques fichiers. Un tel procédé est caractéristique du style fonctionnel de programmation, celui du Lisp pur et du Haskell, par opposition au modèle impératif qui domine dans la plupart des langages de programmation — C, C++, Java, Perl, etc. Dennis Ritchie et Ken Thompson expliquent dans l’article de 1974 « The UNIX Time-Sharing System » que le concept de tube (pipe) permet d’économiser la création de fichiers temporaires [Ritchie & Thompson 1974, 371]. Les auteurs n’évoquent pas dans cet article la fonction plus générale du tube, qui est de faire communiquer des processus s’exécutant en parallèle. Mais selon les témoignages des créateurs d’Unix, telle n’était pas sa fonction originelle : absent de la première version d’Unix et du premier Unix Programmer’s Manual [Ritchie & Thompson 1971], le concept fut introduit en 1972 sous la lourde insistance de Doug McIlroy, qui considérait les commandes comme des opérateurs binaires dont les opérandes de gauche et de droite étaient respectivement l’entrée et la sortie, ce qui permettait de les composer. Dans une note du 11 octobre 1964 datant du projet MULTICS, McIlroy comparait cette technique à un « tuyau d’arrosage » (garden hose- screw), métaphore qui marqua profondément les concepteurs d’Unix, Ken Thompson et Dennis Ritchie. Les explications qu’il leur livra, un après-midi sur le tableau noir, finirent par triompher de leurs réticences initiales [Ritchie 1984].

37 La commande précédente peut donc être réécrite en se passant de toute mention explicite du fichier B, qui n’est pas essentielle :

Philosophia Scientiæ, 17-3 | 2013 157

invite_de_commande $ cat < A | wc -l > C invite_de_commande $

38 Une telle commande prend le contenu du fichier A en entrée, envoie le résultat dans un tube, c’est-à-dire un fichier temporaire, qui est immédiatement utilisé comme entrée du programme wc -l, dont le résultat est écrit dans le fichier C. L’utilisation de tubes, qui élimine les B d’une commande en branchant directement la sortie B d’une instruction sur l’entrée B de l’autre, correspond donc très précisément à la notion de coupure en théorie intuitionniste de la démonstration, qui correspond elle-même en mathématiques, comme l’observe Jean-Yves Girard4, à la composition d’applications :

39 L’utilisation de tubes est associative et la commande cat peut y jouer le rôle de fonction identité. Les trois commandes suivantes sont en effet équivalentes : invite_de_commande $ wc -l < A invite_de_commande $ cat < A | wc -l invite_de_commande $ wc -l < A | cat

40 Objets, actes, composition, identité, associativité : les interactions entre actes et objets dans l’ontologie de processus et de fichiers du système d’exploitation Unix présentent toutes les propriétés qu’un théoricien des catégories est en droit d’en attendre.

4 Thématisation et antithématisation

4.1 L’antithématisation en mathématiques

41 L’ontologie des actes et des objets est au cœur de l’analyse par Jean Cavaillès de la dynamique des mathématiques, qui permet d’éclairer certains concepts actuels de l’informatique.

42 Si les concepts d’acte et d’objet furent initialement empruntés à la phénoménologie, ils sont pourtant à considérer ici comme le produit d’une conscience anonyme ou polymorphe plutôt que comme celle d’un sujet mathématicien, fût-il idéalisé [Sinaceur 1990, 2581], [Sinaceur 1994, 99–100], [Schwartz 1998]. En concluant son ouvrage par l’évocation de « moments » d’une conscience mathématique qui se transforme et se révèle à elle-même dans l’histoire d’une « dialectique » [Cavaillès 2008b, 78] plutôt que comme une subjectivité personnelle et immuable, peut-être Cavaillès a-t-il voulu être à Husserl ce que Hegel avait été à Kant.

43 Les mathématiques appliquent certains actes à certains objets — par exemple l’addition à des nombres. Objets et actes sont pour ainsi dire « typés » : chaque acte est défini

Philosophia Scientiæ, 17-3 | 2013 158

pour un certain type d’objet. Mais en complément au bagage d’actes fourni par le cours normal de la science, Cavaillès en a défini d’autres que l’on pourrait appeler, par analogie avec le lambda-calcul du second ordre [Girard 1972], [Reynolds 1974], « polymorphes », en ceci qu’ils peuvent s’appliquer à des objets de type quelconque. Ces actes polymorphes s’appellent paradigme et thématisation [Cavaillès 2008b, 27].

44 Étant donné un type et certains objets qui lui appartiennent, le passage au paradigme est la transformation des objets en objets plus généraux, et des actes en actes plus larges. À titre d’exemple, lorsque l’on généralise l’addition de nombres en addition de vecteurs, le concept de « nombre » a été élargi et l’addition revêtue d’une signification plus vaste, mais les objets sont restés des objets et les actes des actes.

45 La thématisation est l’acte polymorphe que Cavaillès définit comme opposée au paradigme. Elle fait abstraction des objets d’un certain type, considère les actes comme des objets, et détermine quels actes de niveau supérieur sont applicables à ces nouveaux objets. Tel est le cas lorsque des applications linéaires, d’abord actes effectués sur les objets particuliers que sont les vecteurs, deviennent à leur tour des objets, sous forme de matrices passibles d’actes de niveau supérieur tels que la mutiplication correspondant à la composition d’applications.

46 La dynamique de l’histoire des mathématiques progresse selon Cavaillès dans le sens du paradigme et de la thématisation, c’est-à-dire d’une abstraction croissante.

47 On peut cependant s’interroger sur l’asymétrie de l’opposition entre les deux concepts. La combinatoire des transformations possibles permet en effet théoriquement de définir (1) un geste transformant les objets en objets et les actes en actes — le paradigme ; (2) un geste transformant les actes en objets — la thématisation ; mais également (3) un geste transformant les objets en actes, et qui, s’il existait, serait le véritable inverse de la thématisation.

48 Or lorsque Cavaillès envisage l’opération inverse de la thématisation, ce n’est que sur le mode du « principe de réductibilité » qu’il attribue à Husserl [Cavaillès 2008b, 50], selon lequel les constructions abstraites sont toujours en dernière instance convertibles en constructions concrètes, apparemment sur le seul mode du retour à l’origine. Redescendre l’arbre de l’abstraction ne pourrait être que revenir à des théories du passé, modèles particuliers — et réducteurs — des théories présentes. La « dialectique » des mathématiques, selon Cavaillès, évoluerait à sens unique vers l’abstraction.

49 Il ne serait pourtant pas monstrueux d’imaginer un processus inverse de la thématisation, disons d’« antithématisation », qui ne se réduise pas à un simple retour aux objets passés, mais qui nous apprenne du nouveau sur ce que nous croyions être des objets atomiques. On trouverait ainsi en mathématiques l’analogue du constat enthousiaste de François Jacob selon lequel « même l’atome, le vieil irréductible, est devenu un système » [Jacob 1970, 344].

50 Le procédé existe. Les théoriciens des catégories ne font pas autre chose en considérant notre vieux singleton {a} comme s’il était un acte, une transformation illustrée par le morphisme a de l’élément terminal 1 dans un ensemble A [Lawvere 1964, 1507] :

51 Et nous qui croyions naïvement que le singleton n’était un objet ! le voici devenu acte, prenant à contre-pied toute thématisation. On voit également le point, notre autre « vieil irréductible », devenir un morphisme de corps — quand ce n’est pas un foncteur,

Philosophia Scientiæ, 17-3 | 2013 159

incarnation d’un acte opérant sur des actes. Il nous faut donc introduire, en regard du paradigme et de la thématisation, le concept d’antithématisation, qui eût peut-être paru monstrueux auxyeux de Cavaillès.

52 Contrairement à la thématisation, qui traite le complexe comme s’il était simple, l’antithématisation met au jour la complexité cachée du simple. Loin d’être condamnée au retour à des objets anciens et bien connus, l’antithématisation est parfois l’occasion de découvrir du nouveau dans ce que l’on croyait bien connaître. Les mathématiciens ont leur microscope.

4.2 L’antithématisation en informatique

53 Notre théorie des systèmes d’exploitation a les fichiers pour objets et les processus pour actes. Nous savons par ailleurs qu’il existe en mathématiques une transformation d’actes en objets — la thématisation — et une transformation d’objets en acte — l’antithématisation. Trouve-t-on l’équivalent dans l’ontologie des systèmes d’exploitation ?

54 On trouvera sans peine comment des processus peuvent, par thématisation, être transformés en fichiers. On en trouvera un premier exemple dans la programmation, écriture de programmes dont on attribue parfois l’idée moderne au « First Draft of a Report on the EDVAC » [von Neumann 1945], l’ordinateur électronique EDVAC, conçu de 1945 à 1949, permettant, contrairement à son prédécesseur l’ENIAC, conçu de 1943 à 1946, de charger des programmes archivés dans la mémoire ; mais l’idée de programmation était déjà inscrite au cœur de l’article fondateur de Turing [Turing 1936] : la machine universelle de Turing est une machine capable d’exécuter des programmes inscrits sur son propre ruban. Écrire un programme ou l’enregistrer en mémoire, que ce soit sous forme binaire ou textuelle, c’est faire d’un acte un objet ; car si le programme en cours d’exécution est un acte, son code source est un objet.

55 Par « sérialisation », un processus peut également être figé en un objet, ce qui permet par exemple sa transmission sur un autre support. Il en résulte une ambiguïté dans la notion même de processus, selon qu’elle désigne l’acte en cours d’exécution ou l’objet qui l’incarne. John Lions montrait ainsi que le pseudo-parallélisme d’Unix 6 — c’est-à- dire le fait que l’absence de véritable parallélisme au niveau matériel soit compensée par une alternance dans l’exécution des processus, l’ordonnancement de leurs exécutions respectives étant laissée aux bons soins du système d’exploitation — nous contraignait à distinguer dans la notion de processus un niveau supérieur et un niveau inférieur : Au niveau supérieur, le « processus » est un concept important d’organisation pour décrire l’activité d’un système informatique pris dans son ensemble. [...] À ce niveau, les processus eux-mêmes sont considérés comme étant les entités actives du système [...] : les processus naissent, vivent et meurent ; ils existent en diverses quantités ; [...] ils peuvent interagir, coopérer, entrer en conflit, partager des ressources ; etc. Au niveau inférieur, les « processus » sont des entités inactives qui sont activées par des entités actives telles que le processeur. En permettant au processeur de passer fréquemment de l’exécution d’une image de processus à une autre, on peut créer l’impression que chacune des images de processus se développe continûment, ce qui mène à l’interprétation du niveau supérieur. [Lions 1996, 7–1b]

Philosophia Scientiæ, 17-3 | 2013 160

56 Certes, il ne s’agit là encore que d’un niveau assez rudimentaire de thématisation, dans la mesure où il n’y est question que d’interrompre et de reprendre l’exécution de processus. L’acte a beau devenir objet, cet objet n’est pas véritablement soumis à des actes de niveau supérieur. On n’obtiendrait donc guère plus que ce que Hegel appelait un « mauvais infini » : une répétition indéfinie à l’identique.

57 Mais le processus ainsi thématisé peut devenir objet pour des actes de niveau supérieur, donnant naissance au « bon infini » d’une réelle progression qualitative de la connaissance. Il est courant de voir des processus dont la fonction est d’écrire des programmes — et bien évidemment des processus écrivant des programmes à écrire des programmes. Les débogueurs sont des programmes capables de figer l’exécution d’un processus pour analyser, étape après étape, chacun de ses états instantanés, et par là son comportement général. L’utilisation d’un compilateur n’est autre qu’un processus analysant les propriétés d’un autre processus en en scrutant la forme objectifiée, à savoir le code source. La hiérarchie peut même encore être enrichie d’un niveau : lorsque Xavier Leroy a certifié en Coq le compilateur C CompCert, il a exécuté un programme dont la fonction était d’analyser à sa façon le texte d’un programme qui lui-même analysait d’une autre façon des programmes. Lorsqu’un assistant de preuve thématise un compilateur qui lui-même thématise des codes sources, la connaissance des actes de niveau inférieur progresse véritablement grâce à des niveaux supérieurs qui n’en sont pas le simple décalque. La thématisation informatique n’est pas moins dialectique que celle que Cavaillès a vue dans les mathématiques.

58 Mais comme les mathématiques, la technologie des systèmes d’exploitation comprend également un procédé d’antithématisation, c’est-à-dire de transformation d’objets en actes. Telle est l’exécution d’un programme, lorsqu’un texte enregistré en mémoire est soudain considéré par le système — via un appel de type exec(), qui prend en argument un nom de fichier suivi des arguments qui seront passés au programme — non plus comme un objet existant pour lui-même et passible de certains actes tels que la lecture et l’écriture, mais comme une suite d’instructions que le système lui-même doit exécuter : un texte qui n’est pour lui pas tant à lire qu’à assumer. On voit une fois encore comme il est réducteur de définir le processus comme l’instance d’exécution d’un programme ; non seulement ce n’en est qu’un cas particulier, mais la notion de « programme » elle-même n’a de sens que définie à partir des concepts de fichier et de processus, et surtout de la transformation de l’un en l’autre.

59 Le concept d’antithématisation ne permet donc pas seulement de décrire un procédé que l’on trouve attesté dans l’histoire récente des mathématiques : il permet également de décrire un phénomène inscrit dans l’ontologie d’actes et d’objets qui structure les systèmes d’exploitation depuis Unix.

5 Conclusion

60 Les notions de processus et de fichier en ingénierie informatique ne sont donc pas les fruits du hasard et de la contingence. Elles sont la transposition d’une ontologie de l’acte et de l’objet dans l’ontologie des systèmes d’exploitation. Aussi n’est-il pas surprenant de constater que cette ontologie est accompagnée des deux procédés transformatoires dont témoigne l’histoire des mathématiques : le procédé de thématisation, transformation d’actes en objets de niveau supérieur, correspond à

Philosophia Scientiæ, 17-3 | 2013 161

l’écriture ou à la manipulation de codes sources ; et le procédé d’antithématisation, qui transforme un objet en acte, correspond à l’exécution de programmes.

61 Peut-être cet exemple aura-t-il montré que les concepts des sciences de l’ingénieur, et par exemple de la théorie des systèmes d’exploitation, loin d’être condamnés à l’empirisme, peuvent relever d’une nécessité conceptuelle et être ancrés dans des concepts ontologiques fondamentaux que la philosophie a mission de dévoiler. Par ses concepts comme par ses procédés, la pratique informatique est pour la philosophie un objet aussi pur que les mathématiques.

BIBLIOGRAPHIE

BOVET, DANIEL P. & CESATO, MARCO 2001 Le Noyau Linux, Paris: O’Reilly.

CAVAILLÈS, JEAN 2008a Remarques sur la formation de la théorie abstraite des ensembles, dans Philosophie mathématique, édité par ARON, R. & MARTIN, R., Visions des sciences, Paris: Hermann, 23–176. 2008b Sur la Logique et la théorie de la science, Bibliothèque des textes philosophiques, Paris: Vrin.

GENTZEN, GERHARD 1955 Recherches sur la déduction logique, Philosophie de la matière, Paris: Presses Universitaires de France, 1re édition : “Untersuchungen über das logische Schliessen”, Math. Z., 39, 176–210 et 405– 431. Traductions et commentaires par Robert Feys et Jean Ladrière.

GIRARD, JEAN-YVES 1972 Interprétation fonctionnelle et élimination des coupures dans l’arithmétique d’ordre supérieur, Thèse de doctorat, Université Paris VII. 2006 Le Point aveugle. Cours de logique, I : Vers la Perfection, Visions des sciences, Paris: Hermann.

HEGEL, GEORG WILHELM FRIEDRICH 2004 Encyclopédie des sciences philosophiques, II. Philosophie de la nature, Bibliothèque des textes philosophiques, Paris: Vrin.

JACOB, FRANÇOIS 1970 La Logique du vivant. Une histoire de l’hérédité, Tel, Paris: Gallimard.

LAWVERE, F. WILLIAM 1964 An elementary theory of the category of sets, Proceedings of the National Academy of Sciences of the United States of America, 52(6), 1506–1511.

LIONS, JOHN 1996 Lions’ Commentary on UNIX 6th Edition with Source Code, Charlottesville: Peer-to-Peer Communications.

REYNOLDS, JOHN C. 1974 Towards a theory of type structure, dans Lecture Notes in Computer Science, Springer-Verlag, t. 19, 408–425.

Philosophia Scientiæ, 17-3 | 2013 162

RITCHIE, DENNIS M. 1984 The evolution of the Unix Time-sharing System, Communications of the ACM, 17, 365–375.

RITCHIE, DENNIS M. & THOMPSON, KEN 1971 Unix Programmer’s Manual. http://www.cs.bell-labs.com/who/dmr/1stEdman.html. 1974 The UNIX Time Sharing System, Communications of the ACM, 17, 365–375.

SCHWARTZ, ÉLISABETH 1998 Jean Cavaillès et la « philosophie du concept, Philosophia Scientiæ, 3(1), 79–97.

SINACEUR, HOURYA 1990 Thématisation, dans Encyclopédie philosophique universelle, Volume II, Les notions philosophiques, Paris: Presses Universitaires de France. 1994 Jean Cavaillès. Philosophie mathématique, Philosophies, Paris: Presses Universitaires de France.

TANENBAUM, ANDREW 2003 Systèmes d’exploitation, Paris: Pearson.

TANENBAUM, ANDREW & WOODHULL, ALBERT S. 1997 Operating Systems. Design and Implementation, Upper Saddle River: Prentice Hall.

TURING, ALAN M. 1936 On computable numbers, with an application to the Entscheidungsproblem, Proceedings of the London Mathematical Society, 2(42).

TURNER, RAYMOND & EDEN, AMNON H. 2007 Towards a programming language ontology, dans Computation, Information, Cognition—The Nexus and the Liminal, édité par DODIG-CRNKOVIC, G. & STUART, S., Cambridge: Cambridge Scholars Press, 147–159.

VON NEUMANN, JOHN 1945 First draft of a report on the EDVAC, Rap. tech., University of Pennsylvania, report prepared for U.S. Army Ordinance Department under Contract W-670-ORD-4926.

VUILLEMIN, JULES 1962 La Philosophie de l’algèbre, tome I : Recherches sur quelques concepts et méthodes de l’Algèbre moderne, Épiméthée, Paris: Presses Universitaires de France, réédition en 1993.

NOTES

2. L’expression « en gros » de Tanenbaum cache en réalité d’autres formes d’appels système, qui se rapportent par exemple au temps, comme les appels gettimeofday et select. 3. Le critère de minimalité a des conséquences que l’on ne saurait négliger ; il conduit par exemple à réviser les rapports entre les processus et les fils d’exécution (threads), à mettre de côté les innovations liées au matériel et notamment au parallélisme. C’est le prix à payer pour une première approche, qui devra donc être complétée par une analyse des dynamiques à l’œuvre dans l’histoire des systèmes d’exploitation — analyse qui, rompant avec le critère de minimalité, devra s’appuyer sur un corpus plus vaste et plus complexe. 4. Au niveau logique qui correspond à la logique intuitionniste, « la règle de Modus ponens, ou plutôt la transitivité de l’implication, le syllogisme [qui sont des « cas particuliers de coupure », cf. p. 48], devient la composition des morphismes » [Girard 2006, 155].

Philosophia Scientiæ, 17-3 | 2013 163

RÉSUMÉS

Il est parfois fécond, en philosophie des sciences, de chercher si les concepts techniques relèvent d’une nécessité de structure plutôt que des seuls hasards de l’invention. En essayant de fonder de la sorte les concepts fondamentaux des systèmes d’exploitation que sont les notions de processus et de fichier, on s’aperçoit qu’ils sont, depuis Unix, les pendants des notions ontologiques abstraites d’acte et d’objet, et qu’ils satisfont toutes les propriétés que la théorie des catégories peut en attendre. La programmation peut dès lors être vue comme une thématisation, c’est-à- dire la transformation de l’acte qu’est le processus en l’objet qu’est le fichier ; mais on découvre également que l’exécution de programmes joue le rôle d’une « antithématisation », transformation d’objets en actes dont l’histoire récente des mathématiques fournit également des exemples. Par ses concepts comme par ses procédés, la pratique informatique est pour la philosophie un objet aussi pur que les mathématiques.

In philosophy of science, it is sometimes fruitful to examine whether technical concepts originate from a structural necessity, rather than from the contingency of invention. Operating systems, for instance, rely on two main concepts: the notions of process and file. Trying to identify their raison d’être, we can see that they are, since Unix, the transposition in computer science of the abstract ontological notions of act and object, and that they satisfy all the properties that category theory may expect. Programming can therefore be seen as a “thematization,” i.e. the transformation of the act or process into an object or file. But we also realize that the execution of programs acts as an “antithematization:” a transformation of objects into acts, which also has instances in recent history of mathematics. Concepts and methods of computing practices can be for philosophy as pure objects as mathematics are.

AUTEUR

BAPTISTE MÉLÈS Laboratoire d’Histoire des Sciences et de Philosophie, Archives H. Poincaré (UMR 7117), CNRS, Université de Lorraine, Nancy (France)

Philosophia Scientiæ, 17-3 | 2013 164

Introduction au rapport inédit de Helmholtz sur Mosso

Alexandre Métraux

1 On ignore presque tout des circonstances qui ont amené Hermann von Helmholtz à s’engager en 1878 dans ce que l’on pourrait appeler une « campagne de promotion académique ». Seul parmi les historiens des sciences, Philipp Felsch mentionne dans son excellente monographie consacrée au physiologiste italien Angelo Mosso [Felsch 2007, 43] le fait que Helmholtz rédigea, très probablement à la demande de son collègue cadet ou à celle d’un secrétaire (permanent ou non) d’une académie des sciences européenne, un rapport qui nous est parvenu sous forme d’ébauche autographe avancée1.

2 Tentons, sinon de reconstruire en détail — faute de documents d’archives — l’épisode en question, du moins d’en indiquer quelques traits saillants.

3 Titulaire de la chaire de physique la plus renommée d’une Allemagne récemment unifiée (le décret de nomination est signé par l’empereur à Versailles le 13 février 1871), Helmholtz peut enfin réaliser à Berlin le rêve de ses rêves : se consacrer librement à des recherches de physique qu’il considère comme urgentes, intéressantes, prometteuses. Compte tenu de la liberté d’enseignement et de recherche qui lui est garantie, compte tenu aussi de l’allocation de moyens matériels considérables (entre autres choses, la construction d’un nouveau laboratoire de physique), personne ne pourra le faire dévier du parcours de recherche qu’il s’est tracé ou qu’il anticipe plus ou moins clairement. Chose assez rare — non seulement pour son époque — c’est un médecin qui a enseigné l’anatomie avant de devenir un habile physiologiste (avec un penchant prononcé pour des problématiques d’épistémologie), et qui couronne sa carrière en acceptant la nomination sur une chaire de physique, à l’endroit même de ses études universitaires.

4 Élu membre ordinaire de l’Académie des sciences le 1er avril 1871, il y présente une série de mémoires consacrés à divers aspects de la théorie électrodynamique. En 1875, il étudie les phénomènes propres aux cyclones et aux orages, deux ans plus tard, il est nommé à une deuxième chaire de physique, celle de l’Académie de médecine et de

Philosophia Scientiæ, 17-3 | 2013 165

chirurgie des armées. Il y inaugure son enseignement par une leçon devenue célèbre sur la pensée dans la médecine. Et, pour ainsi dire, en simultané, il se penche sur la thermodynamique, sur l’acoustique du téléphone, et revient, en 1878, à des problèmes d’optique et de microscopie, cf. [Königsberger 1903, 170–263].

5 Lui qui, vers la fin de son enseignement à Heidelberg, n’a pas eu d’hésitation à exprimer un sentiment de lassitude envers la physiologie (et la psychologie), n’a par conséquent pas délaissé la physiologie pour de bon à son retour à Berlin. C’est sans doute une des raisons pour lesquelles il est amené à évaluer quelques travaux qui lui parviennent très probablement de Turin, où Angelo Mosso, qui enseigne alors la pharmacologie avant de succéder en 1879 à Jakob Moleschott sur la chaire de physiologie, vient de terminer une série d’études concernant les effets de l’activité psychique sur la masse cérébrale chez l’homme.

6 Grâce à trois sujets, vivant soit à Turin, soit dans les environs de la capitale piémontaise, Mosso peut saisir littéralement sur le vif le flux du sang dans le cerveau humain. Ces sujets ont en effet subi un traumatisme crânien dont les séquelles peu ordinaires — une ouverture béante — offrent un accès immédiat à la matière nerveuse. On peut ainsi non seulement voir à l’œil nu les mouvements qui se produisent à la surface du cerveau, mais encore appliquer à la dure-mère des instruments de mesure permettant de transformer ces mouvements en tracés inscrits durablement sur papier enfumé.

7 Mosso a préalablement analysé le flux du sang2 au niveau de l’avant-bras. Pour ce faire, il a élaboré un pléthysmographe, un dispositif qui saisit des variations volumétriques. L’appareil comporte un cylindre en verre dont un côté est ouvert, l’autre se terminant par un goulot. On introduit l’avant-bras jusqu’au coude par le côté ouvert, qui est ensuite fermé par un anneau en caoutchouc. On remplit le cylindre d’eau provenant d’un récipient externe. La pression du liquide est maintenue à un niveau constant, mais de telle sorte que le liquide exerce une légère pression sur l’avant-bras. Toute variation de pression de l’avant-bras est transmise par l’ouverture-goulot soit à un tambour de Marey, soit, selon le modèle en question, à un dispositif d’une autre espèce destiné à la production de tracés. Étant donné que l’onde sanguine modifie le volume de l’avant bras, donc la pression que celui-ci exerce sur le liquide ambiant, le pléthysmographe est censé enregistrer d’infimes variations de la pression sanguine causées ou bien par un effort physique du sujet, ou bien par un acte mental intense, ou bien par tout autre processus exerçant un effet sensible sur l’activité cardiaque3.

8 Dans son rapport, Helmholtz renvoie aux études du pouls dit négatif que Mosso a récemment entreprises à l’aide du pléthysmographe de son invention, études publiées en 1878 d’une part dans l’Archivio per le scienze mediche sous le titre de « Il polso negativo e sui rapporti della respirazione addominale e torracica nell’uomo [Le pouls négatif et ses rapports à la respiration abdominale et thoracique chez l’homme] » [Mosso 1878a] et, d’autre part, comme monographie, sous le titre de « Sulle variazioni locali del polso nell’antibraccio dell’uomo [Sur les variations du pouls local dans l’avant-bras de l’homme] » [Mosso 1878c]4.

9 C’est à la suite de ces recherches, qui portent déjà sur les rapports entre actes psychiques et processus corporels, que Mosso se lance dans l’examen d’un thème apparenté, mais cette fois en observant et en mesurant le comportement du sang dans les vaisseaux cérébraux situés sous la dure-mère, à l’endroit de l’ouverture de la boîte crânienne.

Philosophia Scientiæ, 17-3 | 2013 166

10 Dans ses analyses, Mosso (qui est souvent assisté par quelques-uns de ses collaborateurs) applique un hydrosphygmographe à la dure-mère de ses sujets, Giovanni Thron (ou Tron), 11 ans, Caterina X, 37 ans, et Michele Bertino, 36 ans. L’hydrosphygmographe résulte de l’adaptation du célèbre sphygmographe de Marey, que Mosso ré-invente en partie en fonction des besoins (et des contraintes physiques) auxquels lui et ses sujets ne peuvent se soustraire. Au lieu de saisir les variations du pouls à l’aide d’une plaque solide montée au bout d’une pièce métallique et mise au contact immédiat de la peau recouvrant un vaisseau sanguin, c’est un petit récipient plat en caoutchouc empli de liquide qui fait fonction d’« organe sensible » captant les mouvements effectués par les vaisseaux sanguins situés immédiatement sous la dure- mère. Les variations ainsi saisies par le récipient-capteur sont transmises à un tambour de Marey, lequel produit, comme dans le cas du pléthysmographe évoqué plus haut, des tracés sur papier enfumé5.

11 Mosso non seulement varie les tâches de ses trois sujets, mais encore procède à des expériences sur divers sujets animaux qu’il soumet à des agents physico-chimiques. À chaque fois, il considère, pour ainsi dire, le corps entier comme une entité vivante dans laquelle se réverbèrent des processus soit physico-psychiques, soit purement physiques.

12 En outre, il effectue sur ses sujets humains des mesures parallèles par l’emploi simultané de l’hydrosphygmographe et du pléthysmographe afin de mieux saisir quelques éléments du complexe physico-physio-psychique formé par l’enchevêtrement du psychique et du physique chez l’homme.

13 On ne sait pas trop lesquels des travaux (manuscrits et/ou publications) de Mosso sont sous les yeux de Helmholtz lorsqu’il rédige son rapport. On ignore même la date à laquelle le texte est envoyé de Berlin en Italie (car c’est parmi les inédits de Mosso que le manuscrit est conservé, et non pas dans les archives helmholtziennes de l’actuelle Académie des sciences de Berlin et Brandebourg).

14 On peut affirmer avec certitude qu’il a lu, ou du moins consciencieusement feuilleté, non seulement un ou plusieurs textes de Mosso sur la pléthysmographie (voir plus haut), mais encore l’article rédigé par celui-ci en collaboration avec l’anatomiste turinois Carlo Giacomini sur des expériences des mouvements cérébraux [Giacomini & Mosso 1876–1877], ainsi que le manuscrit qui sera impirmé dans les actes de l’Accademia dei Lincei en 1880.

15 Quelques réflexions pour conclure ces remarques en guise d’introduction à l’inédit de Helmholtz.

16 Le rapport met en évidence l’aspect instrumental et technique des analyses effectuées par Mosso. Rien de surprenant à cela, car il s’agit de la partie des recherches psycho- physiologiques qui lui est très proche compte tenu de sa propre carrière de chercheur en physiologie sensorielle. Ou encore, il n’y a rien de surprenant à cela vu que la psycho-physiologie lui est plus familière que l’expérimentation sur des sujets animaux que l’auteur a, certes, pratiquée au début de sa carrière, mais qu’il n’apprécie pas trop. La ligne de démarcation entre la psycho-physiologie effectuée à sec (c’est-à-dire en l’absence du liquide sanguin) et la physiologie effectuée dans des bains de sang se fait ressentir jusque dans ce bref rapport.

17 Enfin, Helmholtz n’évoque nulle part le problème métaphysique (ou philosophique) des rapports de l’âme au corps. Comme Mosso, comme nombre d’autres physiologistes de

Philosophia Scientiæ, 17-3 | 2013 167

l’époque, il se contente de signaler les effets que peuvent causer des attitudes mentales, des actes psychiques ou l’intensité avec laquelle s’effectue un processus attribué à l’intelligence humaine. Mais contrairement à certains théoriciens (et neurophilosophes) actuels, il n’est nulle part question des rapports entre le cerveau et les états ou les processus psychiques. C’est toujours le corps dans son ensemble (cœur, poumons, vaisseaux, peau, nerfs, estomac, muscles, et ainsi de suite) qui est maintenu en rapport aux états et processus mentaux.

18 Mosso n’a pas assis sa renommée internationale sur les analyses à crâne ouvert, mais sur les exploits (au sens réel, presque athlétique du terme) rendus nécessaires par la physiologie en haute montagne : auto-expériences psycho-physiologiques à 4000 m au- dessus du niveau de la mer, création d’un laboratoire en haute altitude, direction d’une équipe de chercheurs qui profite en particulier du savoir-faire que Mosso a acquis en maniant sphygmographes et pléthysmographes dans les années 70/80 du XIXe siècle6.

BIBLIOGRAPHIE

BINET, ALFRED 1896 Recherches de pléthysmographie, Année psychologique, 2, 576–581.

BINET, ALFRED & COURTIER, JULES 1896 Circulation capillaire de la main dans ses rapports avec la respiration et les cates psychiques, Année psychologique, 2, 87–167.

FELSCH, PHILIPP 2007 Laborlandschaften. Physiologische Alpenreisen im 19. Jahrhundert, Göttingen: Wallstein.

GIACOMINI, CARLO & MOSSO, ANGELO 1876-1877 Esperienze sui movimenti del cervello nell’uomo, Archivio delle scienze mediche, 2, 246– 278.

HALLION, LOUIS 1901 Pléthysmographie, dans Traité de physique biologique, édité par D’ARSONVAL, A. ET COLL., Paris: Masson & Cie, t. 1, 405–421.

KÖNIGSBERGER, LEO 1903 Hermann von Helmholtz, t. 2, Braunschweig: Viehweg & Sohn.

MOSSO, ANGELO 1875 Sopra un nuovo metodo per scrivere i movimenti dei vasi sanguini dell’uomo, Atti della Reale Accademia delle scienze di Torino, 11, 21–81. 1876-1877 Introduzione ad una serie di esperienze sui movimenti del cervello nell’uomo, Archivio delle scienze mediche, 1, 207–244. 1878a Sul polso negativo e sui rapporti della respirazione addominale e torracica nell’uomo, Archivio delle scienze mediche, 2, 401–464. 1878b Sull’azzione fisiologica del’aria compressa, Archivio delle scienze mediche, 2, 148–176. 1878c Sulle variazioni locali del polso nell’antibraccio dell’uomo, Turin: Stamperia reale. 1880 Sulla circolazione del sangue nel cervello umano. Memoria del Socio corrispondente A.

Philosophia Scientiæ, 17-3 | 2013 168

Mosso letta nella seduate del 7 dicembre 1879, Atti della R. Accademie dei Lincei anno CCLXXVII, 5(3e série), 237–358.

NOTES

1. Angelo Mosso est né le 30 mai 1846 à Turin, où il est décédé le 24 novembre 1910; voir aussi les sites italiens comportant des biographies succintes et . 2. Cf. une des premières études que Mosso consacre à la mesure du flux du sang présentée à l’Académie des sciences de Turin [Mosso 1875]. 3. Pour de plus amples détails, cf. [Hallion 1901, 409–410]. 4. Cf. aussi l’article complémentaire relatif à la pléthysmographie, « Sull’azzione fisiologica del’aria compressa [Sur l’effet physiologique exercé par l’air comprimé] », paru la même année [Mosso 1878b]. Cf. aussi les réflexions ultérieures de Binet sur la pléthysmographie [Binet 1896] et [Binet & Courtier 1896]. 5. Voir la figure illustrant l’instrumentation dans l’application au sujet Michele Bertino in [Mosso 1880, 244]. Cf. en outre un autre article de Mosso relatif à la même problématique [Mosso 1876-1877]. 6. La transcription et la traduction de l’inédit de Helmholtz reposent sur l’autographe disponible au Virtual Laboratory de l’Institut Max Planck pour l’Histoire des sciences (Berlin), cf. . Je remercie Julia Kursell (Amsterdam) de m’avoir signalé l’existence du texte helmholtzien, ainsi que Philipp Felsch (Berlin) pour les nombreuses conversations autour de Mosso.

AUTEUR

ALEXANDRE MÉTRAUX Laboratoire d’Histoire des Sciences et de Philosophie, Archives H. Poincaré (UMR 7117), CNRS, Université de Lorraine, Nancy (France)

Philosophia Scientiæ, 17-3 | 2013 169

Ein unveröffentlichter Bericht über verschiedene Arbeiten Angelo Mossos

Hermann von Helmholtz

1 [a] Die Arbeit des Herrn Mosso hat ein Resultat, welches für die Physiologie des Gehirns und für die Psychologie von hervorragender und fundamentaler Wichtigkeit ist; sie zeigt nämlich, dass bei jeder hinreichend kräftigen Empfindung, bei jeder willkürlichen Anspannung des Denkens, wie sie zum Beispiel beim Rechnen eintritt, augenblicklich eine Steigerung des Blutdrucks und eine grössere Amplitude der Pulswellen im Gehirn eintritt, welche nur so lange [b] dauert, wie die erhöhte Thätigkeit der genannten psychischen Vorgänge. Betreffend den Mechanismus dieser Änderungen im Blutlauf hat er den Beweis geliefert, dass dieselben nich[t] von geänderter Action des Herzens, dass vielmehr in anderen Gliedern des Körpers, namentlich dem Vorderarm[,] die entgegengesetzten Aenderungen eintreten, woraus folgt[,] dass die Steigerung der Blutbewegung im Gehirn durch eine örtliche Wirkung auf die Gefässe des Gehirns bedingt sein muss. Er hat gezeigt, dass Sinnesempfindungen bei einem Schlafenden [c] dieselbe Wirkung haben, selbst, wenn derselbe nicht davon erwacht. Er hat gezeigt, dass die schnellen Aenderungen des Blutdrucks durch den Pulsschlag auch in den venösen Sinus des Gehirns wahrzunehmen sind, während langsamere Aenderungen durch Veränderung der Menge der Cerebrospinalflüssigkeit ausgeglichen werden durch Veränderung der Blutmenge im Gehirn.

2 Eine so schnell und regelmässig eintretende Aenderung im materiellen Zustande des Gehirns bei erhöhter Thätigkeit der Aufmerksamkeit war bisher noch nicht bekannt. Was die [d] Frage betrifft, ob diese Entdeckung in die Fristen eintritt, die in dem Programm des Preises festgesetzt sind, so ist zu bemerken, dass die negative Wirkung auf die Gefässe des Vorderarms von Herrn Mosso schon in seinem früheren Werke beobachtet und beschrieben ist. Auch fallen einige der unvollkommeneren Beobachtungen an der Caterina X vor den Anfang des Jahres 1878, die an dem Thron Giovanni sind im Januar 1878 publiciert und können nach dem §§3 des Programms als ein vorläufig mitgetheilter Theil der Arbeit angesehen werden.

Philosophia Scientiæ, 17-3 | 2013 170

3 [e] Der wichtigste, unter den günstigsten Umständen angestellte Theil der Arbeit, welcher namentlich auch alle die Nebenversuche enthält[,] welche nöthig waren[,] um den Mechanismus der Wirkung auf das Blut festzustellen[,] ist in dem vorliegenden Aufsatz zuerst veröffentlicht. Die Ausführung der Versuche unter verhältnismässig schwierigen Verhältnissen an lebenden Menschen zeugt von ungewöhlicher Geschicklichkeit und vollkommenerKenntniss der methodischen Fortschritte der neueren Physiologie und von eingehendem Verständniss der Mechanik des Blutlaufs und aller Umstände, die auf ihn Einfluss haben.

4 [f] Die Majorität der Subcommission der Akademie hat deshalb geurtheilt, dass diese Arbeit des Herrn Mosso vollkommen geeignet ist, um zur Ertheilung des Preises vorgeschlagen zu werden.

Philosophia Scientiæ, 17-3 | 2013 171

Un rapport inédit sur divers travaux d’Angelo Mosso

Hermann von Helmholtz Traduction : Alexandre Métraux

Note du traducteur: La transcription rend fidèlement l’orthographe de l’inédit autographe de Helmholtz, et cela jusqu’à l’oubli du « t » dans le mot « nicht », erreur que l’auteur ne semble pas avoir remarquée ou à laquelle il ne devait pas accorder d’importance, vu qu’il s’agissait, probablement, d’une ébauche très avancée du rapport qui aurait dû, ou pu, être copiée, après correction, pour l’envoi en Italie. Dans la transcription ainsi que dans la traduction, l’espace laissé en blanc d’une largeur correspondant à quatre ou cinq mots aurait dû mentionner le titre de l’un des ouvrages de Mosso auquel Helmholtz fait allusion dans cet alinéa. Le texte est conservé sur six feuillets volants, marqués dans la transcription ainsi que dans la traduction par les lettres [a] à [f]. Enfin, la transcription comporte d’autres ajouts entre [ ] afin d’en faciliter la lecture.

1 [a] Le travail de Monsieur Mosso atteint un résultat d’une importance notable et fondamentale pour la physiologie du cerveau aussi bien que pour la psychologie. Il démontre en effet qu’une augmentation de la pression sanguine et de l’amplitude de la pulsation dans le cerveau se produit instantanément à chaque sensation suffisamment forte et à chaque effort d’attention intellectuelle, tel qu’il a lieu par exemple dans le calcul mental. La durée de cette augmentation est équivalente [b] à celle de l’activité accrue des processus psychiques mentionnés. Concernant le mécanisme de ces modifications du flux du sang, l’auteur a apporté la preuve qu’elles ne provenaient pas de l’action altérée du cœur. Au contraire, dans d’autres parties du corps, notamment dans l’avant-bras, des modifications contraires se produisent, d’où il suit que l’augmentation des mouvements du sang dans le cerveau est nécessairement due à un effet local sur les vaisseaux cérébraux. L’auteur a montré en outre que les sensations produisaient le même effet sur un sujet endormi, même lorsqu’elles ne provoquent pas son réveil. Il a montré que les modifications rapides de la pression sanguine causées par le pouls peuvent encore être observées dans les sinus véneux du cerveau, tandis que les

Philosophia Scientiæ, 17-3 | 2013 172

modifications plus lentes sont compensées par la quantité du liquide cérébro-spinal variant en fonction de la quantité de sang dans le cerveau.

2 Une altération si rapide et si régulière dans l’état matériel du cerveau, suite à l’activité accrue de l’attention, n’était pas connue préalablement. Quant à la [d] question de savoir si cette découverte entre dans les délais mentionnés dans le programme du concours, il faut noter que l’effet négatif sur les vaisseaux de l’avant-bras a déjà été observé et décrit par M. Mosso dans son précédent ouvrage . En outre, quelques-unes des observations moins abouties concernant Caterina X sont intervenues avant le nouvel an de 1878, et celles portant sur Thron Giovanni ont été publiées en janvier 1878. Conformément à l’art. 3 du programme, elles peuvent être considérées comme faisant partie de l’ouvrage tout en ayant été publiées à titre provisoire.

3 [e] La partie la plus importante de l’ouvrage, qui a été élaborée dans des conditions optimales et qui englobe toutes les expériences subsidiaires nécessaires à la détermination du mécanisme de l’effet sur le sang, a été publiée pour la première fois dans le présent article. Les expériences menées sur des sujets humains vivants dans des conditions particulièrement difficiles témoignent de l’extraordinaire habileté, de l’entière maîtrise des progrès méthodiques accomplis par la physiologie récente, de la compréhension approfondie de la mécanique de la circulation sanguine et de toutes les circonstances qui peuvent exercer une influence sur ladite circulation.

4 Aussi les membres de la sous-commission de l’académie ont-ils jugé à la majorité que le présent ouvrage de M. Mosso était parfaitement digne d’être proposé pour l’attribution du prix.

Philosophia Scientiæ, 17-3 | 2013