APA Newsletters NEWSLETTER ON PHILOSOPHY AND COMPUTERS

Volume 08, Number 2 Spring 2009

FROM THE EDITOR, PETER BOLTUC

FROM THE CHAIR, MICHAEL BYRON

PHENOMENAL MACHINE CONSCIOUSNESS

IGOR ALEKSANDER “Essential Phenomenology for Conscious Machines: A Note on Franklin, Baars and Ramamurthy: ‘A Phenomenally Conscious Robot’”

PENTTI O. A. HAIKONEN “Slippery Steps Towards Phenomenally Conscious Robots”

FLORIDI ON FLORIDI

LUCIANO FLORIDI “Understanding Information Ethics: Replies to Comments”

PRÉCIS OF PHILOSOPHY OF COMPUTER ART: A BLOC CO-ORGANIZED WITH THE AMERICAN SOCIETY FOR AESTHETICS (ASA)

DOMINIC MCIVER LOPES “Précis of A Philosophy of Computer Art”

DEREK MATRAVERS “Sorting Out the Value of New Art Forms”

PRÉCIS OF PHILOSOPHY OF COMPUTING AND IT

PHILIP BREY AND JOHNNY HARTZ SØRAKER “A Précis of Philosophy of Computing and Information Technology”

© 2009 by The American Philosophical Association GORDANA DODIG-CRNKOVIC “Comments on ‘Philosophy of Computing and Information Technology’ in Handbook of the Philosophy of Science, Vol. 9 Technology and Engineering Science, Part 6 Philosophical Issues of Engineering Disciplines”

ARTICLES ON AUTONOMOUS WEAPONS

J. P. SULLINS “Telerobotic Weapons Systems and the Ethical Conduct of War”

WILLIAM L. DYER “Keep Autonomous Systems Under Manual Control (for now): A Veteran’s Reflection”

DON BERKICH “Autonomous Robots”

DISCUSSION PAPERS

ROXANNE KURTZ “Ontologically Tough Artifacts and Non-Spooky Intentions – Comments on Baker”

E. DIAZ-LEON “How Many Explanatory Gaps Are There?”

BOOK REVIEWS Christian Fuchs: Internet and Society. Social Theory in the Information Age REVIEWED BY EVELINE LUBBERS

CHRISTIAN FUCHS “A Response to Lubbers” Clayton M. Christenson et al.: Deconstructing the Oracle: A Review of Disrupting Class REVIEWED BY TOM P. ABELES Samir Chopra and Scott Dexter: Decoding Liberation: The Promise of Free and Open Souce Software REVIEWED BY MATT BUTCHER Andy Clark: Supersizing the Mind: Embodiment, Action, and Cognitive Extension REVIEWED BY AMY E. WHITE

CALLS FOR PAPERS “Dialogue and Universalism Informs of a Special Issue Vol. XIX, No. 1–2/2009 Related to the Focus of this Newsletter” “The Philosophy of Computer Science” “Last Call on the Ontological Status” APA NEWSLETTER ON

Philosophy and Computers

Piotr Bołtuć, Editor Spring 2009 Volume 08, Number 2

similarities and dissimilarities in their objectives, focus, even in ROM THE DITOR their organization details. Dodig-Crnkovich also expresses her F E disagreement with Brey’s and Søraker’s somewhat reserved, if not dismissive, attitude towards the computational turn. I hope for more discussion of the structure and scope of philosophy of Piotr Bołtuć computing and information, especially after the main version University of Illinois at Springfield of Søraker’s and Brey’s chapter is out this Fall. In the previous issue of this Newsletter I suggested that Phenomenal consciousness is looming large in the current we should devote a little more time to applied philosophy. We issue, as a follow up on the article by S. Franklin, B. Baars, and are glad to have three papers on robotic weapons in a section U. Ramamurthy [Vol. 08, No. 1]. In his paper, I. Aleksander built around J. P. Sullins’s article. Sullins discusses primarily discusses phenomenal systems and recalls his criteria of the ethical issues of the use of autonomous robots in warfare. phenomenal states. He argues (following Gamez) that It is followed by a piece by W. L. Dyer clarifying some related ‘phenomenology’ is a sine qua non condition of consciousness. practical topics. An article by D. Berkich closes this bloc with In a brief, sharp note P. O. Haikonen criticizes the main thesis analytical reflections on the various notions of autonomy, of Baars/Franklin/Ramamurthy’s article, by pointing out that the and how they apply to robots. Berkich chooses the notion of “loss of stable perception does not lead to the loss of any kind autonomous agency as the right framework for robotics. The of phenomenal consciousness”; hence, phenomenal nature of issue of autonomous robotic weapons, and the responsibility consciousness is a distinct issue from perceptual stability.1 for their actions, is a timely one since this seems the direction in We are proud that two major computer ethicists associated which the military is going. But robot-autonomy, its criteria and with this Newsletter received the Barwize prizes for 2008 and moral ramifications, is a broader topic. We invite our readers 2009; please see M. Byron’s note “From the Chair” for details. to follow up on this debate with submissions. L. Floridi’s article (in Vol. 07, No. 1) resulted in much discussion In our discussion section R. Kurtz refers back to L. Baker’s that lasted throughout both issues of 2008; eight philosophers article (Vol. 07, No. 2). Kurtz constructs a couple of clever participated. In the current issue we gladly host Floridi’s reductios that may turn out to present a challenge to the extensive response to all those commentaries, which may opponents of Baker’s thesis. We are still open to entertaining enhance the general understanding of Floridi’s “metaphysical commentaries on Baker’s eye-opening piece; the author foundation of information ethics,” as T. Bynum has worded it intends to respond to all commentaries in 2010. We also have in his commentary (Vol. 08, No. 1). a discussion piece by E. Diaz-Leon, still referring to Harman’s It is a good idea for a newsletter to publish précis of paper (Vol. 06, No. 2). She recalls Chalmers’s point that we can important works soon to come. In the current issue we have explain intentionality in physical-functional terms, which seems two of those. We are glad to feature the paper by D. M. Lopes, to explain Harman’s main example without an explanatory gap; which is the capstone of the block on the relatively new domain it seems to me that Harman’s response would involve his older of computer art, co-organized with the American Society for notion of functionalism of concepts, missing from the 2007 paper. Aesthetics (ASA). It is a précis of his upcoming book in which Diaz-Leon goes on to argue that there are two explanatory gaps: the author tries to define computer art within the context of she finds the Nagel/Jackson-style explanatory gap to be ‘easy’ analytical aesthetics. It is followed by D. Matravers, who analyzes while the Stoljar-style gap between a phenomenal experience some ontological problems with computer art highlighted (even if familiar) and its description to be the hard one. by D. Lopes’s definition. More discussion on this topic, also After discussion papers the book reviews come, all four of co-organized with ASA, shall appear in the next issue of this them. The topic of war is raised, from a very different perspective Newsletter. We are grateful to Sondra Bacharach and all others than that adopted by Sullins and Dyer, in the review of C. Fuchs’s involved in this project. book by E. Lubbers, and is followed by the author’s response. We happen to have another précis in the current issue—a Book reviews may and should be somewhat opinionated and version of a long chapter by P. Brey and J. H. Søraker on controversial as is also visible in the piece by T. Abeles, quite Philosophy of Computing and Information in a broader critical of C. M. Christensen’s view of online education as a anthology. The authors aim at organizing the subject area disruptive technology. Abeles’s article is the only piece in this of philosophy and computing, list its main topics, concepts, issue devoted directly to online learning but at least we do not developments, contributors—the précis gives us a glimpse come empty-handed on this. of all this. In her commentary, G. Dodig-Crnkovic situates The two remaining book reviews pertain to the main this work in a number of contexts: we learn more about areas of philosophy and computers. In a review of S. Dexter the anthology that the Brey/Søraker chapter belongs to; we and S. Chopra’s book Decoding Liberation: The Promise of learn about the two projects that have had somewhat similar Free and Open Source Software, M. Butcher situates the book objectives in the past (Floridi’s and Moor’s/Bynum’s), the — APA Newsletter, Spring 2009, Volume 08, Number 2 — in a broader historical context and provides an overview of its our own Harriet Baber and Pat Hurley from University of San main ideas. Last but not least, A. White reviews A. Clark’s new Diego, and Wilfrid Sieg from Carnegie Mellon. defense of the extended mind hypothesis in his “Supersizing At the Pacific Meeting in April, in addition to Terry Bynum’s the Mind.” White points out that, while the gist of the argument Barwise Prize address, our own Amy White from Ohio University remains unchanged since Clark and Chalmers came up with chaired a session on gender and technology. The speakers the extended mind argument in 1998, the book does contain included Harriet Baber and Lori Watson from University some helpful responses to critics of this approach. Most of San Diego, and Margaret Crouch from Eastern Michigan philosophers around the world must have heard D. Chalmers’s University. multiple presentations of his introduction to the book, which Congratulations to our award winners, and I look forward brings to the forefront the philosophical meaning of his iPod; to seeing you at our future sessions. those include his keynote speech at the World Congress in Seoul last August and his Barwise Lecture organized by this committee at the Pacific APA. This no doubt enhanced the interest in Clark’s book. Last but not least, I want to thank all those who made it PHENOMENAL MACHINE possible for me to work on this issue: the Dean of the College of Liberal Arts and Sciences at UIS, my colleagues at the CONSCIOUSNESS Department of Philosophy there, as well as the Chair and Members of this Committee. Endnotes Essential Phenomenology for Conscious 1. In an attempt not to publish my own stuff in a newsletter Machines: A Note on Franklin, Baars and where the editor is actively involved in peer review, I Ramamurthy: “A Phenomenally Conscious restrained my own reflection on the questions raised by Franklin/Baars/Ramamurthy to the paper “The Philosophical Robot” Issue in Machine Consciousness,” International Journal of Machine Consciousness 1 (2009). It is fair to add that the issue Igor Aleksander includes a number of authors who took part in the debate in Imperial College, London this newsletter: I. Aleksander, B. Baars, A. Chella, S. Franklin, R. Manzotti, and U. Ramamurthy. Abstract It was right of Franklin, Baars, and Ramamurthy (FBR) in the Fall 2008 issue of APA Newsletters, to draw attention to the existence of the Machine Consciousness paradigm and the fact FROM THE CHAIR that both functional and phenomenal computational models are being attempted. However, the aim of this note is to argue that phenomenal states and a phenomenological mode of analysis Michael Byron are essential if it is argued that a computational model has Kent State University explanatory value in the context of consciousness. Phenomenal States Are Not Add-Ons Recently, the Committee has named the 2008 and 2009 Barwise Prize winners. The 2009 prize will be awarded at the Eastern The machine consciousness paradigm has been summarized Division Meeting in New York, NY, in December. elsewhere (Aleksander 2008) to show that there are two motivations in this work. The first is to add a computational The 2008 Barwise Prize winner is Terry Bynum, from strand to the way in which consciousness may be discussed, Southern Connecticut State University. According to his and the second is to build machines, mainly robots, that in some website, Terry “is Professor of Philosophy at Southern way benefit from being driven by mechanisms which in their Connecticut State University, Director of the Research Center neurological form may be responsible for the consciousness of on Computing and Society there, and Visiting Professor at living organisms. Besides the work associated with the Baars De Montfort University in Leicester, England. He is a lifetime Global Workspace Theory, which led Franklin and his colleagues member of Computer Professionals for Social Responsibility, to build working systems (IDA and LIDA described in FBR), Past Chair of the Committee on Professional Ethics of the the paradigm recognizes that there is a difference between Association for Computing Machinery, and Past Chair of the functional and phenomenal approaches. The prevalent belief Committee on Philosophy and Computers of The American in papers such as FBR is that phenomenal states are added to Philosophical Association.” The Committee chose to recognize functional structures of the Global Workspace kind. his contributions to the area of philosophy and computers, especially his work on computing and human values. Here we argue that adding mechanisms to produce a The 2009 Barwise Prize winner is Luciano Floridi, of coherent perceptual field…might provide a significant the University of Hertsfordshire and the University of Oxford. step forward toward phenomenal consciousness in According to his website, Luciano “currently holds the machines.... research chair in philosophy of information at the University of The thrust of this note is that constructing systems that Hertfordshire, Department of Philosophy. He is also Fellow of St. can be said to be phenomenally conscious requires a structure Cross College, Oxford University.” He is the president of IACAP where maintaining the presence of phenomenal states is the and on the executive board of INSEIT. The Committee chose to principal task which drives the modelling process ab initio recognize his many contributions and pioneering work in the and cannot be seen as an addition to an existing functional fields of philosophy of information and information ethics. model. The Committee sponsored sessions at two recent APA divisional meetings. At the Central Division meeting in Chicago, What Is a Phenomenal System? I chaired a session on online courseware. Panelists included Phenomenology is a study of consciousness considered to

— 2 — — Philosophy and Computers — have been founded by German philosopher Edmund Husserl, important phenomenal characteristics contained in the axioms. who defined it as (1901): “The reflective study of the essence This embraces the methods of virtualism and neural structures of consciousness as experienced from the first-person point as explained below. of view.” A logician by training, Husserl wanted to formalize that which is internally known through experience and treat Virtualism it separately from that which can only be reached by thought. Sloman and Chrisley (2003) point out that the relationship For example, a beautiful painting by Botticelli can be conjured between mind and physical structure may be one where the in the mind, if once seen, and so said to be an intentional same “mind” could run on a large number of computational experience directed at world objects. In contrast, the concept structures. That is, following earlier suggestions by Dennett of all the paintings by Botticelli can be imagined and understood (1991), the relationship between mind and its physical but is unlikely to be directly experienced and is therefore support can be considered as a virtual machine running on not intentional. So the subject of the relationship between a computational host. They describe a system that contains reality and the content of mental states is the phenomenology many functional components that fulfill many behavioral tasks proposed by Husserl and comes into focus by recognizing such as reaction, deliberation, and moderation of the previous different properties of mental states. While this leads to a processes, all in concert with perception, recognition, and consideration of how mental states are acquired, it starts with action. Dubbed “Virtual Machine Functionalism,” they suggest the introspective intuition of how different mental states feel. a computational framework which represents these virtual A phenomenal system, therefore, is one whose reasoning features is on the host machine but do not indicate how it could begins with recognition of the way that mental things differ in any sense be phenomenal. Later, Chrisley and Parthmore rather than the way that world things differ. Therefore, in the (2007) addressed a notion of “synthetic phenomenology” constructive style of explanation embraced in the machine in terms of creating phenomenal states and their formal consciousness paradigm, mechanisms are specified by using descriptions. one’s own introspective classification of mental events, and Aleksander and Morton (2007), while maintaining the stance then asking what mechanisms would need to be present in an on virtualism, pointed out that the creation of phenomenal organism that would naturally make distinctions in a similar states is favored by having dynamic neural networks as the host manner on its own mental events. This implies an examination for a phenomenal mind. and incorporation in the design process of phenomenal Neural Substrates states that are accessible through introspection but (in living organisms) not by investigation by a third person. The point needs to be made here that a neural substrate appears to be necessary to satisfy several aspects of the axioms. First, An Axiomatic Approach to Phenomenal States graininess commensurate with the vividness of phenomenal Since 2003, the author and his colleagues (Aleksander and experience is helped by creating neural state machines with Dunmall 2003; Aleksander 2005) have been proposing a states that use a large number of neurons as state variables. methodology based on phenomenology. The first phenomenal Second, the states of such a machine can acquire the necessary character is obtained by an introspective identification of phenomenal (i.e., intentional) character through a method different types of conscious experience. Five such experiences known as “Iconic Learning” (see Aleksander 2005), which were identified and dubbed “the Axioms of Consciousness”: associates what a neuron state variable learns. In Iconic 1) Presence: the feeling of being in an out-there world. Learning this consists of small elements of sensory experience Note that this is the ‘intentional’ side of phenomenology, which become the state variables of an overall (mental) state. This state variable becomes a function of what the neuron that is, experience directed at the real world. is sensing on neural inputs both from external sources and 2) Imagination: the feeling of visiting previously internal, recursive sources. This creates states that together can experienced intentional states (e.g., “I was in Paris reflect experience (dubbed “depictions”) during perception yesterday and can visualize the Eiffel Tower”), or, (axiom 1) and enable recall of experience (axiom 2) providing in humans, the feeling of constructing objects from a methodological basis for the entire set of axioms. language (e.g., Dr. Dolittle’s ‘push-me-pull-you’ horse To illustrate how large scale neural networks provide the with a head on either end). This corresponds to resolution necessary for survival of the organism, consider a tiny Husserl’s notion of intention-free phenomena. marking on a mushroom which may need to enter the mental 3) Attention: the feeling of choosing that which comes state for poisonous mushrooms to be avoided. The sensitivity into the mental state. The intentional (object-directed) is a property that could be expected from a neural state character of a phenomenal state feels as if it requires machine, but may be difficult to design within a functionalist this choice as direction cannot occur towards many framework. Of course, the functionalist designer could argue objects at once. that phenomenology could be acquired by program branching 4) Planning: the feeling of predicting future intentional caused by small changes found in stored images. This is far states in time depending on imagined actions. less plausible than the neural network which can track and form state sequences as a quasi-continuous process of the 5) Emotion: the feeling of evaluating the prediction of experience. A non-neural functionalist representation would four before determining outward actions. be more like frames in movies on film, which branch only This is not proposed as a definite, sacrosanct list, but more through some very smart recognition of some features in stored as an indication of how one might work from a phenomenal “coherent perceptual fields” rather than a system whose state standpoint. Other areas are currently being investigated at structure directly reflects the dynamic, branching experience. Imperial College, London, through Ph.D. theses on language (Sunil Rao) and various classes of emotion (Rabinder Lee, Conclusion: No Phenomenology No Consciousness Imperial College, 2008, and Mercedes Lahnstein, in progress). I am in complete agreement with David Gamez (2008) when The methodology proceeds by using the phenomenal axioms (on p. 25) he defines consciousness as: Consciousness is to examine known ways of designing mechanisms to select the presence of a phenomenal world. So, in this note, I have those that may be capable of sustaining for themselves the welcomed the suggestion in FBR that phenomenal states are

— 3 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — necessary in any explanatory model of consciousness. But I resemble functional consciousness. This kind of study might be have also argued that much more is necessary. It is very difficult difficult, because in nature there is no such thing as functional to start with a model based in classical cognitive science, the consciousness. mode of expression of which is algorithmic and implies virtual Human consciousness is characterized by qualia, the systems determined through the intention of a programmer. So “phenomenal feel” of every percept; qualia are the way in which the methodology suggested here adopts the phenomenal stance sensory information manifests itself in the mind. Therefore, to from the start and requires the identification of mechanisms, be phenomenally conscious is to have qualia-based perceptual which, through a process of learning, are capable of creating “inner” world. This kind of consciousness is not necessarily phenomenal, experiential state structures as part of their an executing agent or process that produces something; it operation. Dynamic neural networks have such properties may only be a way of representation and inner appearance of under strict and studied conditions. That is the beginning of operation. the story and it is clear that a conscious system must also On the other hand, Merker (2005) has proposed that have mechanisms that do not enter consciousness but act phenomenal consciousness produces a stable perceptual in support of the said dynamic neural system. This has been world by distinguishing real motion from the apparent motion left as unspecified in this note, but is treated in the supporting produced by the movement of the sensors. Franklin reads literature (e.g., Aleksander 2005). Also, this note is not meant this proposition backwards and concludes that phenomenal to diminish the influence of Baars’s Global Workspace Theory consciousness can be produced by the production of stable mentioned in FBR, it just may not be the best place to start, if perceptual world. a true understanding of phenomenal consciousness is to be obtained. Things may not be that easy even if we ignored the faulty logic of that conclusion. As Franklin notes, the loss of stable References perceptual world can easily be demonstrated, for example, by Aleksander, Igor and Barry Dunmall. 2003. Axioms and tests for the the gentle pressing of the lower eyelid of an open eye. However, presence of minimal consciousness in agents. Journal of Consciousness the resulting loss of stable perception does not lead to the loss Studies 10 (4-5): 7-19. of any kind of phenomenal consciousness. The proposed cause Aleksander, Igor. 2005. The World in My Mind, My Mind in the World: has been removed, but the proposed consequence remains. Key Mechanisms of Consiousness in Humans, Animals, and Machines. Thus, stable perception cannot be a cause for phenomenal Imprint Academic, Exeter. consciousness. Merker’s original proposition and Franklin’s Aleksander, Igor and Helen Morton. 2007. Depictive architectures for conclusion must be suspected. synthetic phenomenology. In Artificial Consciousness, edited by A. Chella and R. Manzotti. Imprint Academic, Exeter. From the engineering point of view, stable perceptual world can be produced by compensating the effects of the robot’s own Aleksander, Igor. 2008. Machine Consciousness. Scholarpedia 3(2): 4162. http://www.scholarpedia.org/article/Machine_consciousness. motion and this is not even difficult (see, e.g., Haikonen 2007). The possible phenomenal nature of this perceptual world is a Baars, Bernard. 1988. A Cognitive Theory of Consciousness. NY: different question. Cambridge University Press. Chrisley, R. and J. Parthemore. 2007. Robotic specification of the non- References conceptual content of visual experience. In Proceedings of the AAAI Fall Franklin, Stan, Bernard J. Baars, and Uma Ramamurthy. 2008. A Symposium on “Consciousness and Artificial Intelligence: Theoretical phenomenally conscious robot? APA Newsletter on Philosophy and foundations and current approaches.” AAAI Press. Computers 08(2): 2-4. Dennett, D. 1991. Consciousness Explained. Boston: Little, Brown. Haikonen, Pentti O. 2007. Robot Brains; Circuits and Systems for Gamez, David. 2008. The Development and Analysis of Consious Conscious Machines. UK: Wiley and Sons. Machines. University of Essex Ph.D. Thesis, Computing Department. Merker, Bjorn. 2005. The liabilities of mobility: A selection pressure for Husserl, E. 1901. Logical Investigations. Eng. Trans by J. N. Findlay, the transition to consciousness in animal evolution. Consciousness and London Routledge,1973. Original in German, 1901. Cognition 14: 89-114. Sloman, A. and R. Chrisley. 2003. Virtual machines and consciousness. Journal of Consciousness Studies 10 (4-5): 133-72. FLORIDI ON FLORIDI Slippery Steps Towards Phenomenally Conscious Robots Pentti O. A. Haikonen Understanding Information Ethics: Replies University of Illinois at Springfield to Comments

In their interesting paper Franklin, Baars, and Ramamurthy Luciano Floridi (2008) propose that providing a functionally conscious robot University of Hertfordshire and University of Oxford with stable coherent perceptual world might be a step towards a phenomenally conscious machine. In practice this would take Foreword place by providing the robot with means to shield the spatial Information Ethics (IE) is a lively area of philosophical research percepts from the effects of the robot’s own motion. that has been attracting an increasing amount of interesting What is functional consciousness? Franklin, Baars, and work. A recent special issue of Ethics and Information Ramamurthy answer: “An agent is said to be functionally Technology, edited by Charles Ess (“Luciano Floridi’s Philosophy consciousness (sic) if its control structure implements the of Information and Information Ethics: Critical Reflections and Global Workspace Theory and the LIDA Cognitive Cycle.” the State of the Art,” Ethics and Information Technology 2008 However, this is not a proof. This is a definition and as such (10.2-3)), and two issues of the APA Newsletter on Computers conveniently eliminates the need to study if there were anything and Philosophy, edited by Peter Boltuc (Spring 2008, 7:2 and Fall in this implementation that could even remotely qualify as and 2008, 8:1) testify to the development of several lines of critical

— 4 — — Philosophy and Computers — investigations of some of the initial ideas and arguments that Reply to Vaccaro’s “Information Ethics as Macroethics: I have tried to articulate and support during the past decade. Perspectives for Further Research” Indeed, I take it as a good omen that this article, containing the Vaccaro’s article identifies, with insight, several questions replies to the articles published in the APA Newsletter, appears that IE should rightly be called upon answering. They may be exactly ten years after “Information Ethics: On the Theoretical reasonably summarized under three headings: Foundations of Computer Ethics” (Ethics and Information Technology 1999 (1.1), 37-56), a work that I consider the starting a) applicability: whether IE can help in how we deal with point of my research in the field. everyday moral issues. Ten years of sustained, international research of high b) inclusivity: whether IE’s treatment of patients and standards is a long time in any academic field. In Information agents as informational entities can improve our analysis Ethics this is even more so, given the fast-paced and radical and understanding of the full spectrum of applied ethical transformations involving information and communication problems. technologies (ICT). As a result, IE has certainly widened its c) extensibility: whether IE’s ethical principles may be applied scope. It now interacts with many other ethical fields, from to material (as opposed to digital) contexts. business ethics to environmental ethics, from medical ethics to These are very valuable suggestions for further research. the ethics of nanotechnologies, from the ethics of cyberwar to They point towards issues that are challenging and complex. the ethics of e-research. But IE has also deepened its conceptual So it would be preposterous even to try to outline here how insights. These involve now dialogues with other philosophical IE may deal with them fruitfully. What they require is actual and ethical traditions, such as Platonism, Neo-Platonism, work in the field, not further theorizing or a quick and dirty fix. Stoicism, Spinozism, Deontologism, Consequentialism, Nevertheless, they may benefit from a basic clarification, which Contractualism, and Buddhism, as well as analyses and is brief enough to be provided in the short space available for discussions of metaphysical, epistemological, and logical topics, this reply. It is to be hoped that it may further the debate. from digital physics to the method of levels of abstraction, from IE is sometimes understood as if it defended the view structural realism to the philosophy of information. In this rich that artificial, digital, or informational realities should also be and varied context, informed and reasonable debates and taken into account by the ethical discourse, besides human disagreements are welcome. Not only are they clear evidence and biological agents and patients. This is a reasonable but of a healthy market of ideas, open to different and sometimes slightly mistaken view. The thesis actually advocated by IE contrasting views; they above all can foster our understanding is significantly different, and perhaps even more radical (so, and help to guide our sound judgements. The interested reader to a sceptic, it will look even less credible). It consists in will find plenty of such discussions in the comments and in my arguing for a change in the level of abstraction at which the replies to them. ethical discourse may also be fruitfully developed. IE fosters A final explanation, before leaving the floor to the debate. In a development from biocentrism to ontocentrism, where the writing the following replies, I have tried to follow the following latter is expressed in terms of an informational metaphysics. guidelines. To put it simply, according to IE, the effort to be made consists First, I have tried to see points of convergence rather not merely in adding new agents and patients to the list of than seeking confrontation and the sort of wrestling and cut- already ethically qualifiable entities, but to interpret all agents throat duels for which some corners of analytic philosophy is and patients informationally, thus including humans, animals, notorious. When I was younger, I loved a good intellectual fight. social agents, and engineered entities as well. It is an extension I was trained in such a tradition and taught, especially in Oxford, achieved not in terms of addition, but in terms of modification that doing philosophy was a gladiator’s job: only one of the two of the interface through which we analyze moral interactions. philosophers would leave the arena alive. I must be getting It follows that the three fundamental issues of applicability, older because I increasingly appreciate the construction and inclusivity, and extensibility should really be answered after the negotiation of sharable views. Paul of Tarsus docet: “When I was following question: Does it help to adopt an information-based, a child, I spoke like a child, thought like a child, and reasoned metaphysical approach to ethics? As I anticipated, IE defends like a child. When I became a man, I gave up my childish ways” a firm answer in the positive, but the latter only points in the (1 Corinthians 13:11). direction in which further research should be developed. Second, I have tried to avoid repetitions, even when they Reply to Sullins’s “So You Say You Want a Re- were called for, so I am afraid the reader will get a full picture Ontological Revolution” only by browsing through the whole collection of replies and some of the texts to which I have referred there. I hope one day Sullins’s article has many merits, but three of them seem to be to publish a collection. It would certainly facilitate my handling outstanding. I shall list them in no particular order. of EndNote. First, the article finely captures the environmental nature of Third, in order to facilitate the reader’s task, I have tried to IE. This is an uncommon feature among other interpreters of IE. refer only to texts of mine that are freely available online from Second, it is convincing in pointing out the difference between my website: www.philosophyofinformation.net. IE’s philosophy of technology and some of its catastrophist, apocalyptic alternatives. Third, it stresses a crucial difficulty. It By way of conclusion, let me say that I am very grateful is on the latter that I wish to comment briefly. to Peter Boltuc, and to all the colleagues who took their time to write their articles, for this opportunity to acknowledge that To put it simply, Sullins ask whether physis and techne may much more and better work needs to be done in IE before one be reconcilable. The question does not have a predetermined may justifiably claim to have a full outline of the theory. My hope answer, waiting to be divined. It is more like a practical problem, is that I have been able to show that such work is worth all our whose feasible solution needs to be devised. With an analogy, intellectual efforts. we are not asking whether two chemicals could mix but rather whether a marriage may be successful. There is plenty of room for a positive answer, provided the right sort of effort is made. It seems beyond doubt that a successful marriage between physis and techne is vital and hence worth all our efforts. Our

— 5 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — societies increasingly depend upon technology to strive, but by privileging the former as the only authentic dimension of they equally need a healthy, natural environment to flourish. human life. Sullins is right in pointing out that the challenge is to Try to imagine the world not tomorrow or next year, but next reconcile our roles as agents within nature and as stewards of century, or next millennium: a divorce between physis and nature. The good news is that it is a challenge we can meet. techne would be utterly disastrous both for our welfare and for the well-being of our habitat. This is something that technophiles Reply to Buchanan’s “Discursive Explorations in and green fundamentalists must come to understand. Failing to Information Ethics” negotiate a fruitful, symbiotic relationship between technology Buchanan has provided an impressive commentary on some and nature is not an option. Fortunately, a successful marriage key aspects of IE that I have tried to clarify. I doubt that I could between physis and techne is achievable. Admittedly, much add anything more to her outstanding insights. So, in this brief more progress needs to be made. The physics of information comment, I shall limit myself to highlighting a basic and rather can be highly energy-consuming and hence potentially simple thesis, which may help to frame our dialogue. unfriendly towards the environment. In 2000, data centers As I have argued in “Artificial Intelligence’s New consumed 0.6 percent of the world’s electricity. In 2005, the Frontier: Artificial Companions and the Fourth Revolution” figure had risen to 1 percent. They are already responsible for (Metaphilosophy, 2008, 39.4/5, 651-655), the informational more carbon dioxide emissions per year than Argentina or the turn we are experiencing may be described as a fourth Netherlands and, if current trends hold, their emissions will have revolution, in the process of dislocation and reassessment of grown four-fold by 2020, reaching 670m tonnes. By then, it is humanity’s fundamental nature and role in the universe. We estimated that Information and Communication Technologies’ are not immobile, at the center of the universe (Copernican (ICT) carbon footprint will be higher than aviation’s (source: revolution), we are not unnaturally separate and diverse from The Economist, May 22, 2008). The good news is that ICTs will the rest of the animal kingdom (Darwinian revolution), and we also help “to eliminate 7.8 metric gigatons of greenhouse gas are very far from being Cartesian minds entirely transparent to emissions annually by 2020 equivalent to 15 percent of global ourselves (Freudian revolution). We do not know whether we emissions today and five times more than our estimate of the may be the only intelligent form of life. But we are now slowly emissions from these technologies in 2020” (Source: McKinsey’s accepting the idea that we might be informational entities and Information Technology Report, October 2008, “How IT can cut agents among many others, significantly but not dramatically carbon emissions,” by Giulio Boccaletti, Markus Löffler, and different from smart, engineered artefacts. I called this Turing Jeremy M. Oppenheim). This positive (and improvable) balance revolution the fourth revolution. It seems that, in view of this leads me to a final comment. important change in our self-understanding, and of the sort The greenest machine is a machine with 100 percent of ICT-mediated interactions that we will increasingly enjoy energy efficiency. Unfortunately, this is equivalent to a perpetual with other agents, whether biological or artificial, the best way motion machine and the latter is simply a pipe dream. However, of tackling the new ethical challenges posed by ICTs may be we also know that such impossible limit can be increasingly from an environmental approach, one which does not privilege approximated: energy waste can be dramatically reduced and the natural or untouched, but treats as authentic and genuine energy efficiency can be highly increased (the two processes all forms of existence and behavior, even those based on are not necessarily the same; compare recycling vs. doing more artificial, synthetic, or engineered artefacts. This is one of the with less). Often, both kinds of processes may be fostered only fundamental theses defended by IE. by relying on significant improvements in the management of information (e.g., to build and run hardware and processes Reply to Chopra’s “Comment on ‘Understanding better). So here is how IE reinterprets Socrates’ ethical Information Ethics’” intellectualism: we do evil because we do not know better, in Chopra’s article contains a list of questions that are not the sense that the better the information management is the rhetorical, but reasonable requests for further work. He is less moral evil is caused. With a proviso though: some ethical correct in indicating both the need for such a large theoretical theories, especially in the Christian tradition, seem to assume investment and some of the important areas where it should be that the moral game, played by agents in their environments, made. Of course, Chopra’s are also the sort of requests that it is may be won absolutely, i.e., not in terms of higher scores, but fair to make but equally fair to meet in a context different from by scoring perhaps very little as long as no moral loss or error the present one. It would be simply impossible to answer here occurs, a bit like winning a football game by scoring only one any of the questions that close his article, even very succinctly. goal as long as none is received. It seems that this absolute view However, a much more modest goal seems achievable, and has led different parties to underestimate the importance of that is to clarify a potential source of confusion. To put it briefly, successful compromises (imagine an environmentalist unable IE defends an ontological position, not a socio-economical or to accept any technology responsible for some level of carbon technological analysis of our current society (the point is well dioxide emission, no matter how it may counterbalance this captured by Terry Bynum in his article on my work, entitled otherwise). The more realistic and challenging view from IE “Toward a Metaphysical Foundation for Information Ethics,” is that moral evil is unavoidable, so that the real effort lies in which appeared in the following issue of the APA Newsletter, counterbalancing by more moral goodness. Fall 2008, 8:1; see my reply below). The ontological position in Information and Communication Technologies (henceforth itself and the possibility of defending it may both be subject to ICTs) can help us in our fight against the destruction, criticism, but these are different issues. Let me clarify why. impoverishment, vandalism, and waste of both natural and Chopra writes that “The infosphere’s contours, then, human (including historical and cultural) resources, so they continue to be shaped by a variety of forces, not all of which can be a precious ally in what I have called elsewhere synthetic break down barriers to information flows.” There is a risk here environmentalism or e-nvironmentalism. We should resist that the infosphere might be interpreted not as a metaphysical any Greek epistemological tendency to treat techne as the concept (i.e., another way of speaking of Being), but as Cinderella of knowledge; any absolutist inclination to accept some socio-economic or geo-political space. That would be no moral balancing between some unavoidable evil and more a mistake. This is not to say that reality (in the metaphysical goodness; or any modern, reactionary, metaphysical temptation sense) cannot be shaped by a variety of forces, but that such to drive a wedge between naturalism and constructionism forces are not those identified by Chopra. One must agree that

— 6 — — Philosophy and Computers —

“It is socio-economic circumstances that permit the ubiquity debate in terms of a tension between Wiener’s more Aristotelian of this panoply of goods” but also acknowledge that this has and materialist foundation for information ethics, and my nothing to do with the infosphere becoming our ecosystem. approach, which he correctly describes as more Platonic- Imagine criticizing Aristotle’s metaphysics of substance in terms Spinozian and, I would add, idealist (see Soraj Hongladarom, of Greek economic conditions in the fifth century BC. The real “Floridi and Spinoza on Global Information Ethics,” Ethics and point at stake is whether we are modifying our metaphysical Information Technology, Volume 10, Numbers 2-3, September outlook from a materialist one, in which physical objects and 2008). The gap between Wiener and me, however, may be less processes still play a key role, to an informational one, in wide than it seems. For what I have been arguing is that his which objects and processes are dephysicalized, typified, and metaphysics might be too restrictive, not wrong. Even Wiener perfectly clonable, the right of usage is perceived to be at least seems to accept this point sometimes. True, he repeatedly as important as the right to ownership, while the criterion for asserted that he conceived information exclusively in terms existence is no longer being immutable (Greek metaphysics) or of Shannon’s theory, but even the quotations provided by being potentially subject to perception (modern metaphysics) Bynum show that he did not have to limit himself to such a but being interactable. If all this seems a bit too “philosophical,” theoretical frame. When Wiener famously described human let me try to provide an illustrative example, following Chopra’s beings as “patterns that perpetuate themselves” (The Human approach. Use of Human Beings, p. 96) the patterns in question may be Despite some important exceptions (e.g., vases and metal analogue and continuous “persistent information patterns” that tools in ancient civilizations or books after Guttenberg), it was have little to do with Shannon’s concept of information. This the industrial revolution that really marked the passage from a would be good news because Shannon’s information entails nominalist world of unique objects to a Platonist world of types a view of the ultimate nature of reality as necessarily discrete of objects, all perfectly reproducible as identical to each other, and possibly deterministic (this is also known as digital physics). therefore epistemically indiscernible, and hence pragmatically This neo-Pythagorean ontology is hardly tenable nowadays, dispensable because replaceable without any loss. Today, we and in a recent article I have argued that it is actually mistaken find it natural that two automobiles may be virtually identical (“Against Digital Ontology,” forthcoming in Synthese). Instead, and that we are invited to buy a model rather than a specific and more constructively, it is possible to show that a concept “incarnation” of it. Indeed, we are fast moving towards a of information as relational patterns is much more satisfactory, commodification of objects that considers repairment as and provides IE with a structural ontology that is more successful synonymous with replacement, even when it comes to entire philosophically and more easily reconcilable with our current buildings. This has led, by way of compensation, to a prioritization scientific knowledge (“A Defence of Informational Structural of branding, a process compared by Naomi Klein to the creation Realism,” Synthese, 2008, 161.2, 219-253). So, in the rest of this of “cultural accessories and personal philosophies” (No logo, reply, I shall illustrate what informational structural realism (ISR) Canada: Random House, 2000) and of re-appropriation: the amounts to with the help of some classic analogies. person who puts a sticker on the window of her car, which is At least since Plato’s images of the line and of the cave, otherwise perfectly identical to thousands of others, is fighting philosophers have often relied on spatial analogies to explain an anti-Platonic battle. The information revolution has further their theories. References to rooms are particularly popular. exacerbated this process. Once our window-shopping becomes Sextus Empiricus thought that, in our epistemic pursuits, we Windows-shopping and no longer means walking down the are like people in a dark room, searching for gold or shooting street but browsing through the web, the problem caused by at a target: no matter how long the search or the shooting the dephysicalization and typification of individuals as unique proceeds, it is pointless because in principle there is no way to and irreplaceable entities starts eroding our sense of personal establish whether any of us has found a nugget or hit the mark identity as well. We become mass-produced, anonymous (Outlines of Pyrrhonism I.52 and II.325). Turing used different entities among other anonymous entities, exposed to billions rooms for his famous test. And Searle devised a Chinese room of other similar informational organisms (inforgs) online. So for his counterexample. I shall rely on their examples and we self-brand and re-appropriate ourselves in cyberspace by suggest a double box analogy to illustrate ISR. But first, a final blogs and facebook entries, homepages, YouTube videos, and bit of terminology. flickr albums. We use and expose more information about In software engineering, black-box refers to a test-design ourselves to become less informationally indiscernible. We method that focuses on testing functional or behavioral wish to maintain a high level of informational privacy almost as requirements of a program. The methodology treats the if that were the only way of saving a precious capital that can analysandum as a completely opaque and closed system, then be fully invested by us in order to construct ourselves as avoiding using explicit knowledge of its internal nature or discernible individuals. Now, what I’ve been arguing in the past structure to understand the way it works. The opposite is that processes such as the one I have just sketched are part methodology is known as white-box test design. This allows one of a far deeper metaphysical drift caused by the information to “look inside” the system, and it focuses on using specific and revolution. I have also argued that IE is the sort of ethics we detailed knowledge of the program code to guide the selection need to develop if we wish to tackle the moral challenges of test data. A grey-box approach is one that allows only a partial posed by such profound transformations. I believe Chopra view of the internal properties of the system. and I agree that it is a worthwhile investment, perhaps even a According to ISR, any white-box approach to reality is necessary one, given how the world is quickly changing before excluded in principle, given the fact that our knowledge is our eyes. We need to develop a robust Information Ethics. I always mediated by some levels of abstraction (LoA). Forget agree that much work still needs to be done, but that can only about getting out of Plato’s cave. There is no God’s eye be intellectually exciting. perspective from without. We look at the world as if we were Reply to Terrell Ward Bynum’s “Toward a Metaphysical in Sextus’s dark room. This is the first box. We are inside it, but Foundation for Information Ethics” our goal is not mimetic, nor are our (often causal) interactions with the furniture in the room unidirectional, as Sextus assumed. If I were to point out only one major merit of Bynum’s article, I Unlike Sextus’s, ours is only a grey-box. In the best cases, it would certainly choose its perceptive interpretation of IE as a allows the reconstruction of the structural properties relating metaphysical position. Bynum is absolutely right in framing the

— 7 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — the furniture of the room, i.e., our informational objects. These Reply to John Barker’s “Too Much Information: are our second kind of boxes. As in Turing’s test, they are black- Questioning Information Ethics” boxes, not directly knowable, but “epistemically interactable” Let me begin with praise: the article captures perfectly well through LoAs. Sometimes, we can indirectly capture their nature what other, less acute interpreters often fail to grasp, namely, by observing their behavior and mutual interactions, but we the fact that, when I speak of informational objects, I do not do not know their intrinsic (Kant would have said noumenal) mean news, emails, or encyclopaedic entries, but entities properties. How we relate them and use them to build other understood informationally. With a Quinean slogan: to be is black-boxes is our responsibility. This is the right context in to be an informational entity. With a historical comparison, which to talk about a demiurgic power. Informational structural one may think of Leibniz’s monads, or Berkeley’s idealistic realism (ISR) takes our epistemic goal to be constructionist ontology. It is therefore somewhat puzzling that, in the second (mind, not constructivist in any psychologistic or “sociologistic” half of the article, much space is devoted to a digression, sense), not mimetic. Knowledge is not a matter of either (a) namely, the quantification of information Shannon-wise. discovering and describing, or (b) inventing and constructing, The reader is referred to Bynum’s article and my reply to it but of (c) designing and modelling reality, its features, and for a proper understanding of why the mathematical theory behaviors into a meaningful world as we experience it. And of communication represents the wrong approach. Barker one may design and model successfully even in the dark. seems to agree. He acknowledges that “overall complexity, or Intelligibility is the name of the epistemic game, and humanity quantity of information, is a poor measure of intrinsic moral tries to achieve it at any cost, even when this means distorting worth. Now this conclusion, even if true, may not appear to be reality in the most absurd way, from the conception of a flat terribly damaging to Information Ethics, as the latter embodies earth placed at the center of the universe to the interpretation no specific theory of how to measure moral worth.” This is of natural forces and events as anthropomorphic divinities or precisely why we should not be distracted by it. However, the to the assumption of calories, phlogiston, and luminiferous article continues, “It may simply be that some other measure ether. Since we wish to devise an intelligible conceptual is called for.” Or perhaps none at all, which is the alternative environment for ourselves, we do so not by trying to picture or I would prefer. Shannon’s and other similar mathematical photocopy whatever is in the room (mimetic epistemology), approaches to quantitative data, their probability distributions but by interacting with it as a resource for our semantic tasks, and so forth (see my “Information” in the Blackwell Guide interrogating it through experience, tests, and experiments. to the Philosophy of Computing and Information) are simply Reality in itself is not a source but a resource for knowledge. irrelevant here. We should not get lost behind false promises Structural objects (clusters of data as relational entities) work of numbers and formulas. When we speak of more or less evil epistemologically like constraining affordances: they allow or or moral goodness, or higher or lower degrees of flourishing invite certain constructs (they are affordances for the information and well-being, these are qualitative assessments that require system that elaborates them) and resist or impede some others practical wisdom or Phronesis, not a pocket calculator. They (they are constraints for the same system), depending on the may be made more precise by some formalization, but cannot interaction with, and the nature of, the information system that be seriously quantified. processes them. They are exploitable by a theory, at a given LoA, as input of adequate queries to produce semantic information The article continues and concludes with an important (the model) as output. Transforming constraining affordances challenge, which I shall discuss at the end of this reply. Before, into semantic information need not be a metaphysically violent let me stress that the article is correct in stating that, because business (as Bacon thought it might), if reality in itself is indeed of my informational metaphysics, I have argued in favor of an indeterminate or if we are ready to be led by it insofar as it is expansion of the circle of entities that have some (possibly very determinate. From this perspective, semantic concerns (most minimal and often easily overridable) moral claim on us, in order importantly reference, representation, and truth) belong to to include in it every manifestation of Being. Here, however, I the relation among models, that is, among outcomes of LoAs would like to rectify a potential source of misunderstanding. (Kant’s phenomenal world of experience), not to the relation According to the article, “Floridi’s central claim [...], while between models and reality in itself. fascinating, certainly runs counter to most moral theories that have been proposed.” Without starting the usual game of who It turns out that we are like Turing’s interrogator, since said what and whose philosophical ancestors are more or the model of investigation is erotetic: we have indirect (LoA- less respectable, the idea that Goodness and Being (capitals mediated) access to reality and can query it as a database. meant) might be two sides of the same concept, as Evil and Bacon and Galilei shared a similar view. But since our task is Non-being might be, is actually a classic position. Perhaps the not to find out who is who, we resemble Searle in his Chinese objection could rather be that I am defending something hardly room: we get the data input on one side and output semantic new. Without disturbing Eastern traditions within Buddhism, information on the other. The difference, in this case, is that we Hinduism, or Shinto—which I understand attribute intrinsic have some understanding of the rules of the game. It makes little value both to sentient and to non-sentient realities—the reader sense to ask whether the factual information we gain resembles sufficiently acquainted with the history of Western philosophy its source. The Parthenon is as concrete and objective as anyone will recall that many classic thinkers, such as Plato, Aristotle, may wish it to be, but it does not represent marble. Knowing Plotin, Augustine, Aquinas, and Spinoza, have elaborated reality is interpreting it constructively, not portraying it passively. and defended in various ways the fundamental equation Cooking provides a better analogy than photocopying. highlighted above. For Plato, for example, Goodness and Being So the basic idea behind ISR is quite simple: black-boxes are intimately connected. Plato’s universe is value-ridden at its inside a grey-box. The last specification to be added is that very roots: value is there from the start, not imposed upon it these qualifications are LoA-dependent, in the same way as the by a rather late-coming new mammalian species of animals, distinction between being a system and being a component or as if before evolution had the chance of hitting upon homo unit of a system is. A black-box may be opened, but opening it sapiens the universe were a value-neutral reality, devoid of transforms it into a grey-box, in which more black-boxes may be any ethical worth. By and large, IE proposes the same line of found. Whether ad infinitum we simply cannot know. It might reasoning, by updating it in terms of an informational ontology, be Russian dolls (informational objects) all the way in. whereby Being is understood informationally and Non-being

— 8 — — Philosophy and Computers — in terms of entropy. Note that this is not a defense of IE but an or even impossible to win would be unfair. This is a very non- explanation. Although being in the company of Plato or Spinoza, Greek but rather Christian silent axiom. It is not Greek, because for example, might be reassuring, it is not an insurance against Greek culture knows too well the meaning of the tragic: the being mistaken. But it is a rectification of the incorrect remark failure of a good will to do the right thing (see “Information that IE stands rather alone in its defense of what might be called Technologies and the Tragedy of the Good Will,” Ethics and axiological ecumenism. Information Technology 8.4 (2006): 253-62). Greek eyes do My next comment concerns the statement that “the main not fear to see life as intrinsically and irremediably unfair and rationale Floridi provides [for the axiological ecumenism unjust sometimes. The silent axiom is much more in tune with seen above] seems to be an argument from precedent.” This Christian ethics because the latter presupposes a fair Judge and unfortunate oversight is probably due to a fault of mine. In the an Ultimate Redde Rationem (“Redde Rationem villicationis article discussed by Barker I provide only a very general outline tuae,” “give me an account of your stewardship,” Luke 16:2). IE of IE. So let me take advantage of this opportunity to redress the finds a compromise between these two positions by seeking to situation. The reader interested in what I have actually argued interpret the morally good life as a matter of differential score, may wish to check “On the Intrinsic Value of Information Objects as I have tried to explain in my reply to Sullins: we cannot avoid and the Infosphere” (Ethics and Information Technology 2003 doing some evil and this is our tragic predicament, but we can (4.4): 287-304) and “Global Information Ethics: The Importance still be good agents if we do more good than evil, and this is of Being Environmentally Earnest” (International Journal of our heroic chance. We shall inevitably fail many times, but we Technology and Human Interaction 3.3 (2007): 1-11). The could succeed even more times. The question is then: How following should be taken as a mere sketch of the basic line could we know how to strike such an active balance? This is of reasoning. the way I understand the serious challenge posed by the end The actual argument seeks to establish that entities deserve of the article. respect because they have intrinsic value, where this causal The question concerns what sort of metrics one could explanation leads to the interesting problem whether non- use in order to determine whether entities (understood sentient entities too may have some (perhaps very) minimal, as informational objects) have more or less significant (or (perhaps easily) overridable but still intrinsic value. Without negligible) moral value and hence how one could live a rehearsing the whole discussion, I agree that the answer here morally good life. The answer is quite simple: the more an may be difficult to grasp, as it requires a mental frame rather entity contributes to the welfare of the infosphere the higher different from the one any anthropocentric ethics has trained its status is in the chain of morally respectable beings. Once us to adopt. It consists in shifting the burden of proof (a sort of again, this is a classic position for which I claim little originality. Gestaltic shift) by asking, from a patient-oriented perspective, If there is a God, God is the ultimate respectable entity as the not “Why should I care, in principle?” but “What should be taken source of all entities. A biological virus must, unfortunately, care of, in principle?”; that is, whether there is anything that is be destroyed for the sake of the rest of the environment and intrinsically worthless ethically, and hence rightly disrespectable its flourishing. Between these two boundaries we can aim to in this particular sense, i.e., insofar as its intrinsic value is imitate God or run the risk of being worse than a virus. People concerned (again, something might deserve to be disrespected have managed both. for other reasons, e.g., instrumentally or symbolically or for the Nobody could ever argue that a spider’s and a human sake of other entities). In short, one line of reasoning in favor life are equally worthy of respect. Culling, for example, is an of IE’s position (there are others, see references above) is that, ethical duty in environmental ethics. Likewise, the destruction because we lack arguments against the intrinsic value of Being of entities might easily be not only inevitable but welcome in all its manifestations, we are led to expand an environmental in IE. Again, IE is not about respecting a single grain of sand approach to all non-sentient beings. The injunction is to treat as much as one respects the whole earth full of life or other something as intrinsically valuable and hence worthy of moral human beings. It is about placing the threshold below which respect by default, until “proven guilty.” The intuitive idea is that something is morally disrespectable in itself and rightly so. a universe without moral evaluators (e.g., humans) would still With a Cartesian analogy, the mistake lies in thinking that, be morally valuable, and that an ontologically richer universe if one argues that all physical things are extended, then one would be a morally better universe than an ontologically poorer is arguing that they are all of the same size. Of course they one. are not, and nobody could reasonably argue that they are. To At this point, a standard move is to be told that there is at revert to IE, the view that all entities are at least minimally and least one argument against accepting the intrinsic value of Being overridably valuable in themselves should not be confused with in all its manifestations: it would be morally too demanding. the view that they all share the same value. As for IE offering But the objection is only apparently convincing. Age and little guidance once the moral worthiness of all aspects of Being experience teaches that there is probably nothing more difficult (axiological ecumenism) is accepted, this would be equivalent than living a morally good life. Imagine asking what it takes to to saying that, since environmental ethics is based on the value win a gold medal at the Olympic Game, and then objecting of life and of the absence of suffering, then it offers little help that the answer cannot possibly be correct because it would with real-world issues. The truth is exactly the opposite. Having require too much effort, or too many advanced skills, or unusual some general, basic, and robust principles in place helps capacities and gifts. The reply might be that this is not a good enormously when it comes to dealing with complex, practical analogy because a moral life cannot be as difficult to achieve as matters. We should not fear to respect any form of reality too a gold medal at the Olympic Game. And it is exactly this reply much, even if this might be a rather difficult task. Rather, as that unmasks a deeper problem. For it shows that what is at Augustine nicely put it, dilige, at quod vis fac (love/respect and stake is not really a supererogatory issue anymore, but rather do what you wish). the mistaken assumption that lies behind the supererogatory argument: that the moral game is a game sufficiently easy to win, and that it must be so because any human being must in principle be able to win it, and that this is the case because some ultimate salvation is at stake, and a game too difficult

— 9 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Reply to Edward Howlett Spence’s “Understanding scenario in our moral lives, which are full of dilemmas: there Luciano Floridi’s Metaphysical Theory of Information are often good reasons both to adopt a position, a decision, a Ethics” course of action, make a choice, etc. and balancing, equally good reasons not to do so. This is why I have never supported Spence’s article contains many valuable insights and either (b) or (c), and hence I must turn down Spence’s offer to suggestions. It also provides some sceptical remarks. In previous agree with me on a mistake I have never made. replies and many other contexts I have explained why these may be either incorrect or correct but misdirected, as they are The missing philosophical insight is connected to (a). One not relevant to IE (I refer the interested reader to “Information can immediately see that the argument in question is negative Ethics: A Reappraisal,” an article that contains a foreword to or indirect. It consists in reminding historically and showing my replies and the replies to a collection of papers edited by logically that we have nothing to fear from a holistic attitude Charles Ess and published in “Luciano Floridi’s Philosophy of towards the value of Being in all its aspects; that it is fine to start Information and Information Ethics: Critical Reflections and from the presupposition that no entity deserves moral disrespect the State of the Art,” Special issue of Ethics and Information in itself; that anything less than a holistic attitude towards the Technology 10.2-3 (2008)). So, instead of indulging in a critique value of Being would be prima facie unjustified. Throughout of the criticism, in the short space of this reply I would like to the article, this crucial point is entirely ignored. That is why discuss two new points. Spence keeps repeating that he can find no reasons in favor of P: the reasons are against not P. Since this step is never explicitly The first point is scholarly. In an article that has unfortunately grasped nor implicitly followed, unsurprisingly the result is that escaped Spence’s otherwise thorough analysis (“Global Spence and I move in very different directions. This is a pity, Information Ethics: The Importance of Being Environmentally since Spence seems to be interested in reaching conclusions Earnest,” International Journal of Technology and Human that are very similar to IE’s. Interaction 3.3 (2007): 1-11), I have explicitly called attention to the Stoic nature of IE. So Spence is partly right when he writes The second point is more metaphysical: suppose Spence that IE “seems to offer a kind of Stoic Pantheistic Ethics (my had identified and accepted the previous logical step. What [i.e., Spence’s] phrase) that endows everything in the Universe could be the ontological foundation of such an axiological with a moral significance and status through a pre-determined ecumenism? The answer requires the introduction of the divine rational order in which everything is ontologically inter- concept of ontic trust. connected and of which everything forms an ontic part, no Various forms of contractualism (in ethics) and matter how big or small.” However, I would disagree about contractarianism (in political philosophy) argue that moral the fact that pantheism or religious forms of determinism obligation, the duty of political obedience, or the justice of social play any role in IE or in my philosophy in general. I cannot see institutions, have their roots in, and gain their support from a how anyone could believe that they do. This scholarly point so-called “social contract.” This may be a merely hypothetical has an important philosophical implication, which I can only agreement between the parties constituting a society (e.g., summarize here (it is made more fully and repeatedly in several the people and the sovereign, the members of a community, of my writings on IE). IE seeks to break the artificial constraints or the individual and the state). The parties accept to agree of what may count as morally valuable. One way in which it to the terms of the contract and thus obtain some rights in tries to escape from such ethical Chauvinism is by showing that exchange for some freedoms that, allegedly, they would enjoy there is no good reason to raise any barrier. To put it simply, in a hypothetical state of nature. The rights and responsibilities this is like arguing: of the parties subscribing to the agreement are the terms of the a) P is the case because it is not the case that ¬ P. social contract, whereas the society, state, group, etc. is the entity created for the purpose of enforcing the agreement. Both This is classic, elementary logic. Of course, it is also a way rights and freedoms are not fixed and may vary, depending on of reasoning that one may not wish to endorse for equally good the interpretation of the social contract. logical reasons, one only needs to recall Intuitionistic logic or forms of anti-realism à la Dummett. But accepting the logic Interpretations of the theory of the social contract and not its issuing constrains on the validity of the reasoning tend to be highly (and often unknowingly) anthropocentric is mere inconsistency. Let us now turn to the article. When (the focus is only on human rational agents) and stress the Spence writes that coercive nature of the agreement. These two aspects are not characteristic of the concept of ontic trust, but the basic idea of b) “I agree with Floridi that there would be no good a fundamental agreement between parties as a foundation of reason not to adopt such a higher and more inclusive moral interactions is sensible. In the case of the ontic trust, it is moral perspective if there were, in fact, good objective transformed into a primeval, entirely hypothetical pact, logically and independently grounded reasons for adopting predating the social contract, which all agents cannot but sign such a perspective” when they come into existence, and that is constantly renewed a logical mistake is made and a crucial philosophical insight in successive generations.1 The sort of pact in question can be goes missing. understood more precisely in terms of an actual trust. The mistake is the following: (b) is not a version of (a) but Generally speaking, a trust in the English legal system is is rather equivalent to an entity in which someone (the trustee) holds and manages c) if there were good Rs (Spence’s “good objective and the former assets of a person (the trustor, or donor) for the independently grounded reasons”) to adopt P (the benefit of certain persons or entities (the beneficiaries). Strictly thesis concerning the intrinsic moral goodness of speaking, nobody owns the assets, since the trustor has donated Being, or axiological ecumenism) then there would them, the trustee has only legal ownership and the beneficiary be no good Rs not to adopt P; has only equitable ownership. Now, the logical form of this sort of agreement can be used to model the ontic trust, in the but note that while (a) is formally valid, (c) is not (and not just following way: because it is expressed subjunctively): the premise could well be true, e.g., there could well be good Rs to adopt P, while the • the assets or “corpus” is represented by the reality, conclusion still being false, that is, while there could also be including all existing agents and patients (the perfectly good Rs not to adopt P. Indeed, this is a very common infosphere); — 10 — — Philosophy and Computers —

• the donors are all past and current generations of First, purposive or goal-orientated behavior cannot confer agents; any value to entities which have no purpose or goal. This • the trustees are all current individual agents; includes virtually the totality of the universe. At this point, we either have to acknowledge that all the teleological analysis is • the beneficiaries are all current and future individual just another way of speaking about us as the only entities worth agents and patients. moral respect, or we have to expand the teleological analysis By coming into being, an agent is made possible thanks to the in various ways to make sure that it becomes a blanket that existence of other entities. It is therefore bound to all that already does cover all that we wish it to cover in the first place. In the is (the infosphere) both unwillingly and inescapably. It should former case, the position is unsatisfactory as just another case be so also caringly. Unwillingly, because no agent wills itself of the old-fashioned anthropocentrism that IE has been trying into existence, though every agent can, in theory, will itself out to improve. In the latter case, the adjustments are equally of it. Inescapably, because the ontic bond may be broken by an unsatisfactory not just because they are ad hoc but because they agent only at the cost of ceasing to exist as an agent. Moral life end up being increasingly expensive metaphysically: we start does not begin with an act of freedom but it may end with one. having to accept counterfactual cases (“yes, John is just a corpse Caringly, because participation in reality by any entity, including now but if he were alive he would be able, etc.” you get the an agent—that is, the fact that any entity is an expression of picture) and end up endowing with purposive or goal-oriented what exists—provides a right to existence and an invitation to behaviors any manifestation of Being (but what is the purpose respect and take care of other entities. The pact then involves of the moon?). But a fully purposeful universe can only be the no coercion, but a mutual relation of appreciation, gratitude, and result of a fully purposing Architect. I remain therefore sceptical care, which is fostered by the recognition of the dependence of about the value of Spence’s proposal, which seems to me to all entities on each other. A simple example may help to clarify be either unfeasible or more demanding, metaphysically, than further the meaning of the ontic trust. IE. It is often the case that one philosopher’s use of Ockham’s Existence begins with a gift, even if possibly an unwanted razor is another philosopher’s chainsaw massacre. one. A fetus will be initially only a beneficiary of the world. Once Endnotes she is born and has become a full moral agent, she will be, as 1. There are important and profound ways of understanding an individual, both a beneficiary and a trustee of the world. this Ur-pact religiously, especially but not only in the Judeo- She will be in charge of taking care of the world, and, insofar Christian tradition, where the parties involved are God and as she is a member of the generation of living agents, she will Israel or humanity, and their old or new covenant (διαθήχη) also be a donor of the world. Once dead, she will leave the makes it easier to include environmental concerns and values world to other agents after her and thus become a member of otherwise overlooked from the strongly anthropocentric the generation of donors. In short, the life of an agent becomes perspective prima facie endorsed by contemporary a journey from being only a beneficiary to being only a donor, contractualism. However, it is not my intention to endorse or passing through the stage of being a responsible trustee of the even draw on such sources. I am mentioning the point here in world. We begin our career of moral agents as strangers to the order to shed some light both on the origins of contractualism world; we should end it as friends of the world. and on a possible way of understanding the onto-centric approach advocated by IE. The obligations and responsibilities imposed by the ontic trust will vary depending on circumstances but, fundamentally, the expectation is that actions will be taken or avoided in view of the welfare of the whole world. The ontic trust is what is postulated by the approach PRÉCIS OF PHILOSOPHY OF supported by IE. According to IE, any form of reality (any instance of information/being), simply by the fact of being what COMPUTER ART: A BLOC it is, enjoys a minimal, initial, overridable, equal right to exist (be CO-ORGANIZED WITH THE AMERICAN left alone) and develop (not to be interfered) in a way which befits its nature. Nothing is too humble to deserve no respect SOCIETY FOR AESTHETICS (ASA) at all, i.e., to be rightly disrespectable in itself, not even a rock on the moon. In this way, IE brings to ultimate completion the process of enlargement of the concept of what may count as a center of a (no matter how minimal and overridable) moral Précis of A Philosophy of Computer Art claim, which now includes every instance of Being. Dominic McIver Lopes, A Philosophy of Computer Art The acceptance of the ontic trust requires a disinterested (London: Routledge, 2009). judgment of the moral situation from an objective perspective (or Level of Abstraction), i.e., a perspective which is as non- Dominic McIver Lopes anthropocentric as possible. Moral behavior is less likely without University of British Columbia this epistemic virtue. To argue that any ethical judgement is inevitably human and subjective because it is the result of some From the invention of the press, to the elevator, acrylic paint, human subject’s activity is to fall into the fallacy of believing that and the electric guitar, technological progress has helped drive since a message comes from a given source it must inherit all changes in long-established art forms (in these cases, literature, that source’s properties. This is nonsense. An iPod does not architecture, painting, and music), but entirely new art forms play Mac music. Likewise, just because ethics is humanity’s (such as the movies) also spring from new technologies. As business it does not have to be the business of humanity only. everybody knows, computers are having a profound impact Objectivity can be the outcome of a subjective process. The on the long-established arts, and this impact certainly warrants world of mathematics and logic is there as a proof. study, but A Philosophy of Computer Art proposes that computer By way of conclusion, let me briefly comment on Spence’s art is an entirely new art form. It bets that making a case for Information Ethics without Metaphysics. It seems to me to run this thesis sheds light on computer art—and perhaps on the against several difficulties, two of which are worth mentioning arts more generally. here. — 11 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

The book begins by using two conceptions of computer and songs. We normally appreciate a digital image like Wall’s A technology to distinguish what may be called “digital art” from Sudden Gust of Wind in a contrast class that includes arbitrarily “computer art” and then to argue that digital art is not an art any digital image, and we normally appreciate a sonata by EMI form. in contrast with arbitrarily any digital song. However, we do not Since computers handle information in a common digital normally appreciate A Sudden Gust of Wind with digital songs code (usually binary), they are all-purpose representation like the sonata by EMI in mind. We do not appreciate it as digital devices. We use them to make, manipulate, transmit, and art in the most generic sense. The union of digital images and display text, music, sound, and images, whether alone or songs is not a contrast class, so digital art is not an appreciative combined in multimedia. Many scholars explore the varied art kind, so it is not an art form. and far reaching implications of this for the established arts. This argument does not imply that the fact that A Sudden Yet digital stories are still stories, digital images remain images, Gust of Wind has a digital display is irrelevant to our appreciation and digital music is a kind of music. “Digital art” names the of it. On the contrary, we are right to appreciate it as a digital disjunction of digital stories, digital images, digital music, and image. Yet we lose nothing by doing without the idea that we the like. are to appreciate it as digital art. More importantly, the point According to another conception, computers compute. of the argument is to bring out what is needed to defend the That is, they are designed to run computational processes—to claim that computer art is an art form. The task is to show that carry inputs into outputs by following formal rules, or algorithms. it is an appreciative art kind. (If it is an appreciative art kind, Works of “computer art” take advantage of computational then it is an art form, because it is not plausible that it is any processing to achieve interactivity. For example, Sustained other kind of appreciative art kind. For example, it is not likely Coincidence by Rafael Lozano-Hemmer detects the location to be a genre.) of its users and controls a series of lights to ensure that they No work is interactive unless its display can vary depending cast overlapping shadows on the gallery wall (see http://www. on what its user does, and that means its display differs from lozano-hemmer.com/english/projects/sustainedcoincidence. user to user. That is why a hunk of marble or a written text htm). According to the artist’s statement, “the piece is inspired are not interactive. Moreover, this variation means something. by phantasmagorias on the one hand and surveillance and Stepping into Sustained Coincidence, I notice first that the lights digital analysis on the other.” Its operation relies on a computer inevitably follow me and then that, no matter what I do, they that gathers information on the work’s users and follows an bring my shadow into contact with the shadows of others. I get algorithm to maintain an environment with certain features. this far only by gauging the effect of my actions on the current In this way, the actions of users help to shape how the work display of the work and thereby scoping out the space of possible goes. displays that the work generates. Once I do that, I begin to The main elements of a good description of Sustained appreciate the work itself—as what has those possible displays Coincidence show up in a theory of computer art. An item is a (Fig. 1). Of course, anyone is free to appreciate each display as work of computer art just in case (1) it is art, (2) it is run on a unique, without regard to the computer, (3) it is interactive, and (4) it is interactive because fact that it realizes one of many Figure 1. The user appreciates it is run on a computer. Clauses (3) and (4) distinguish works possible faces of the work. the work by generating some of of computer art like Sustained Coincidence from works of However, to do that is to fail to the different displays that it can digital art like Jeff Wall’s photograph, A Sudden Gust of Wind appreciate the work as using have and by appreciating them (http://nymag.com/arts/art/reviews/28478) or the musical computational processing to as belonging among its possible compositions of David Cope’s EMI (http://arts.ucsc.edu/ achieve interactivity. displays. faculty/cope/experiments.htm). Only Sustained Coincidence Computer art users is interactive. What does that mean? A work is interactive just markedly differ from the in case it prescribes that the actions of its users partly generate audiences for traditional art its display. Its display? The display of any work of art is some works. Members of these pattern or structure that results in part from the artist’s activity audiences experience and and that we attend to in order to ascertain the work’s meaning interpret stories, images, and aesthetic features. In Starry Night, the display is a marked and songs, but computer art surface, in Star Wars, it is any of a number of screenings, and users do something more: in “Tears in Heaven,” it is any of a number of performances. they appreciate works by The display of Sustained Coincidence includes many varying generating their displays. This patterns of illumination and cast shadows. Since these are suggests to some that users generated in part by its users, the work is interactive, and this are artists and it suggests to interactivity is mediated by computational processing. Sustained others that users perform Coincidence is a work of computer art, if it is a work of art. computer art works. Neither To see why computer art is a new art form, consider suggestion is correct, though why digital art is not. A kind of art is not an art form unless it each contains a grain of truth. is an appreciative art kind. Works in an art kind share some Whereas users help generate features in common. Works in an appreciative art kind are the work’s displays, only the artist creates the item that has normally appreciated for having those features: they make those many, variable displays (she often does this in part by up a contrast class for purposes of appreciation. Viewed in writing some code for the computational process that the work the context of twentieth-century abstract painting, Broadway runs on). And whereas a performer uses his knowledge of a Boogie-Woogie is restrained, but it is ebullient when viewed work to generate a performance of it, a computer automates against the background of other paintings by Mondrian, so display-generation for users, so that they can discover the work twentieth-century abstract painting and the Mondrian oeuvre by exploring its many displays. That said, both artists and users are different contrast classes for appreciation (for a visual make displays of computer art works. More interestingly, users demonstration, see http://lopes.mentalpaint.net/bbw). They are are like performers in being objects of appreciative attention. different appreciative art kinds. Now consider digital images We appreciate On the Waterfront in part by attending to Marlon

— 12 — — Philosophy and Computers —

Brando’s performance in the role of Terry Malloy. Likewise, I appreciate Sustained Coincidence in part by attending to myself Sorting Out the Value of New Art Forms in the act of generating some of its possible displays. Unless I do this, I don’t really appreciate it as an interactive work. Derek Matravers There now exist thousands of works like Sustained British Open University Coincidence. Some are installations, but others are more closely allied with sculpture, imaging, story-telling, poetry, music, or In his forthcoming book on computer art, Dominic Lopes writes architecture. What they share in common is that they take that “few generations in all of human history have been lucky advantage of computational processing to achieve interactivity. enough to witness the birth of a new art form.”1 We are that They comprise an art form if we appreciate these works, in the lucky, of course, and that form is computer art. Lopes defines way just described, as works of computer art. computer art works (Computer Art Forms) as follows: An account of computer art should provide a framework CAF: an item is a computer art work just in case (1) it’s for criticism specific to computer art. The book suggests that art, (2) it’s made by computer, (3) it’s interactive, and (4) it’s such a criticism refers to facts about a work’s computer-based interactive because it’s made by computer. interactivity as reasons for its having merits or flaws. The book In an illuminating journey, he then explores various also answers several widespread and rather influential critiques ramifications of this definition, including what he calls “the of computer art. One of these is that good art promotes active value question”: What is the value of computer art works? This thought, but computer art inhibits active thought, so computer is intriguing ground because there is a view, among many, that art fails artistically. At first glance, this argument is surprising, computer art forms are not valuable. This could, of course, for interaction often seems to require quick witted responses, be simply a conservative resistance to change, or it could and if the user does nothing, there will be no display from be grounded in solid argument. Lopes gallantly grants his them to enjoy. However, “active thought” is often defined opponents the more respectable position, and considers four narrowly as distanced contemplation. The complaint is that arguments to the effect that computer art is not of value. He users who interact with computer art works get so caught up finds all of these wanting, and thus concludes that computer in the moment—in the feedback loop with the machine—that art is not in principle valueless, but rather—like all art—needs distance is hard to achieve. Indeed, the very cognitive burden to be considered on a case-by-case basis. that interactivity places on the user precludes distance. In this Lopes’s argument focuses on the whether computer respect, computer art compares with video games, which art, as he defines it, is such as to block works in the medium seem to demand real-time responses that block opportunities being of value. In this short paper I would like to take a slightly for leisurely reflection. However, this critique misses a point broader view than those represented in the four arguments he stressed above. Granted that users get caught up in the flow of considers. There are two reasons for doing so. The first is that their interactions, appreciating the work as generating a range I have a hunch that this broader view is, in part, what underlies of displays requires stepping back from any single interaction. the intuition (if one has that intuition) that computer art cannot Immersion in a single display is consistent with—and might be valuable. The second is that it throws some light on the even be needed for—distanced contemplation of the work itself. question (also considered by Lopes) as to whether computer art Other critiques of computer art can be answered in a similar is genuinely a new art form, or merely a vehicle for presenting way, by a careful study of its nature. works in traditional art forms. The problem is not to do with The book closes by discharging the assumption built the computer aspect of computer art, but simply to do with it into the definition of computer art that works like Sustained being a new art form. Coincidence are works of art. The case is not terribly hard to The argument is this. Objects that consist of paint smeared make, but the level of difficulty ratchets up if we take seriously over stretched canvas can, in our tradition, be vehicles for the comparison in the previous paragraph between computer meaning. Is there something about this type of object that art and video games. Video games are run on computers, they made it inevitable that it should play this role, or is it a matter are interactive, and they are interactive because they are run of historical contingency? Richard Wollheim calls this “the on computers. So if they are art, then they are computer art. bricoleur problem” (Wollheim 1980, sec. 22, 23, 63). There are The trouble is that the easy case for counting computer art certainly some qualities of the matter of paintings that make as art then applies to video games and it turns out that video it suitable for the task: it is fairly stable, it endures, it is not games are art. One response is to retrace our steps and reject consumed during appreciation, and so forth. However, more the assumption that works like Sustained Coincidence have art is needed if we are to throw light on the fact that paintings are status. A more interesting response recognizes that Grand Theft vehicles for meaning: namely, that we are able to place this Auto stands to Sustained Coincidence as graffiti tagging stands painting in the context of other, previous, paintings. to Henri Matisse, as Harry Potter stands to Anna Karenina, or as There are both more specific and less specific versions of Feist stands to Richard Strauss. If video games are the popular this argument. The more specific is famously associated with end of computer art, computer art is not a niche phenomenon, Ernst Gombrich. Our understanding of Mondrian’s Broadway and it might be worthwhile considering what we can learn about Boogie-Woogie is informed by it being the case that “the name computer art from video games. Mondrian conjures up the expectation of severity, of an art of Computer art is not just for nerds with an interest in straight lines and a few primary colours in carefully balanced aesthetics. The argument for the thesis that computer art is a rectangles.” This being so, “the boogie-woogie picture gives new art form brings out how much it differs from other art forms. indeed the impression of gay abandon.” However, were we It takes advantage of computational processing to achieve user to believe the picture were by Severini—“who is known for interaction with a highly variable display, and its users engage his futuristic paintings that try to capture the rhythm of dance in appreciative activities that go beyond those of traditional music in works of brilliant chaos”—our understanding would spectators. A philosophy of computer art is an opportunity to fit be very different (Gombrich 1977, 313). Thus, the question of together many of the components of a complete philosophy of which oeuvre we should use as a background is going to arise an art form: a definition of the art form, an ontology, a framework for any individual work of computer art. for criticism, and an account of its status as art.

— 13 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

The question itself does not seem unanswerable. If the In its secondary occurrence the question [Why is this work is one in a number by the same artist, we can use that particular stuff or process an accredited vehicle for artist’s oeuvre as a background for comparison. The oeuvre art?] is raised in a context in which certain arts are might be limited, which might detract from the content of already going concerns. It will be apparent that, when the work, but some oeuvre (no matter how limited) will help. the question is raised in this second way, the answer Alternatively, one might look to the genre of computer art as it receives will in very large part be determined by a background. Within this background, some works will be the analogies and disanalogies that we can construct relatively dynamic, some relatively static, and so on. between the existing arts and the art in question. The more interesting issue is why computer art (considered In other words, the question will benefit from the as computer art—I will from henceforth take this qualification comparatively rich context in which it is asked. It is, as given) is meaningful. Which of Lopes’s clauses allows for instance, in this way that the question, Is the film an meaningfulness in? The two places are either (1), or (2) and art? is currently discussed. (Wollheim 1980, 152) (3). I shall take the last of these first as it is clearly the more However, that might be a bit too quick. What if, instead of trying interesting. Why should our interactivity with a computer, to describe the property in such a way that we pick out only which is grounded in the work being made by a computer, those possessed by past works of art, we import clause (1) of make that interactivity meaningful (as opposed to simply being CAF? That is, the context is ways in which works of art (and not, interactivity)? The broader argument is that what makes an for example, holes) have solved the problem of exemplifying our interactive sequence meaningful is its historical context. being hopelessly trapped. Then one would be considering the Let me consider one of the pieces of work described in way in which this work of art (Boundary Functions) exemplified Lopes’s book: the sense of being hopelessly trapped as compared to how Scott Snibbe, Boundary Functions, 1998. When you and other works of art have exemplified that sense. How it was done your companion step onto a slightly raised platform, a line is could exhibit a great deal more flexibility, as the description projected onto the floor halfway between you. As more people of the property is no longer being used to specify the relevant join in, more lines are drawn, creating an irregular tiled pattern. context. Hence, the way Boundary Functions does it could Try as you might—and some children who have just stepped in be compared (perhaps) to how Berthe Morrisot’s and Mary are trying very hard—you cannot step outside your boundary. Cassatt’s paintings, with their constricted pictorial space, convey Someone remarks that Boundary Functions was the title of the a sense of women being trapped in the domesticity and mores doctoral thesis of Theodore Kaczynski, the Unabomber. of late eighteenth-century bourgeois society (Pollock 1988); or Let us say that what the interaction consists in is trying to how Arthur Miller’s Death of a Salesman conveys a sense of (and failing) to step outside your boundary. What makes this hopeless entrapment in the ever decreasing circles of Willie meaningful in the way in which it is meaningful? In other words, Loman’s options and opportunities. In addition to the successes, what makes it different from (for example) trying to get out of we could also bring into the context some less successful a hole into which one has fallen? Put crudely, the argument is attempts: perhaps Richard Serra’s Matter of Time installation at as follows. This is only one way of putting it; the point could be the Guggenheim in Bilbao, which was just too big to be clever. made in other ways (and let me apologize now for the crude art The moral here is that the value question and the art question criticism). The work is trying to solve a problem: how to convey are not distinct; we need an answer to the second to sort out a sense of being hemmed in, in ways one cannot escape. It an answer to the first. I am sure Lopes will have interesting does this by making it the case that our moving has the effect of things to say on this connection in his final chapter, “The Art redrawing the boundary in a way in which we are still hemmed Question.” Sadly for me, it remains unwritten in the draft from in in our new position. The first move in explaining what makes which I have been working, which means, of course, that I will this reflective and interesting, as opposed to merely irritating, is need to buy the book. Clever fellow, that Lopes. that this “solution” to the problem is seen in the context of other Endnotes solutions to the same problem. However, this is not enough, as 1. The book is in its early stages, and quotations are from the included in such solutions would be those which would enable typescript that Lopes was generous enough to show me. I one to get out of a hole where the rim collapses every time you hope any readers of my piece might be spurred into buying get enough of a grip to haul yourself upward. Such comparisons what looks to be an excellent treatment of the subject. are not going to invest Boundary Functions with meaning. We 2. A debate memorable for the fact that Emin and Scruton had could try to specify the problem more precisely, such that the a fascinating (albeit short) exchange, before Emin wandered context of comparison includes only works of art. However, off the set, having clearly had a great deal to drink. the specification would have to be precise enough to pick out References precisely the way in which past works of art achieved this; in Gombrich, E. 1977. Art and Illusion: A Study in the Psychology of Pictorial other words, it would have to talk about forms specific to art. Representation. London: Phaidon. However, the problem with new media being meaningful Pollock, G. 1988. Modernity and the spaces of femininity. Art in Modern then becomes apparent. If the properties that make the work Culture: An Anthology of Critical Texts, edited by F. Frascina and J. Harris. in the new medium meaningful are the properties the new 121-35. London: Phaidon, 1992. medium has in common with the art of the past, the meaningful Wollheim, R. 1980. Art and Its Objects. Cambridge: Cambridge University properties of a new medium are not those which make it new, Press. but rather those it takes over from older media. This would incline us to think the burden of meaning for computer art is borne by its traditional elements; the value of computer art lies in its being a vehicle for works in more traditional media. This argument seemed to underpin the point Roger Scruton was trying to make to Tracy Emin in the 1997 Turner Prize debate,2 and can be found explicitly in Richard Wollheim, writing in 1968, who makes the same point about film, the then current “new medium” (he was writing in 1958):

— 14 — — Philosophy and Computers —

In our introductory section, we provide a concise yet RÉCIS OF HILOSOPHY OF comprehensive introduction to the field, which has a target P P audience of not only philosophers and computer scientists, COMPUTING AND IT but anyone interested in the intersection between the two. We here debate a proper name for the field (including proposals like “philosophy and computing,” “philosophy of computing,” “computational philosophy,” and “cyberphilosophy”) before A Précis of Philosophy of Computing and settling on “Philosophy of computing and information Information Technology technology.” We here also propose and defend our previously mentioned classification of five subfields, and briefly discuss Philip Brey their historical development. In the context of this historical University of Twente survey, we also discuss the hypothesized emergence of a “computational turn” in philosophy and the rise of Johnny Hartz Søraker “computational philosophy,” and argue that the idea of a University of Twente computational turn in philosophy has perhaps been overstated. Most current work in the field, we believe, studies philosophical Abstract issues in computing and IT. There is much less work that studies The authors recently finished a comprehensive chapter on the implication of computing for philosophy, and in this way “Philosophy of Computing and Information Technology” for qualifies as computational philosophy and contributes to a the forthcoming (fall 2009) Philosophy of Technology and “computational turn.” Engineering Sciences (Ed.: A. Meijers), Volume IX in the Elsevier In the second section, we turn to the fundamental nature series Handbook of the Philosophy of Science (Eds.: D. Gabbay, and basic principles of computing and computational systems, P. Thagard and J. Woods). The purpose of the chapter is to which we refer to as “philosophy of computing.” Although review and discuss the main developments, concepts, topics, this term is often used to denote any philosophical issue and contributors in the intersection between philosophy and related to computers, we have chosen to narrow this section computing, as well as provide some suggestions on how to to issues particular to the nature, possibilities, and limits of structure the many subcategories within what is loosely referred computation. This is arguably the origin of the entire field, to as philosophy of computing. In this short synopsis, we will and includes the early philosophical attention given to notions give an outline of the kinds of issues raised in this chapter. such as computatibility, computational complexity, and symbol grounding, many of which were raised by pioneers like Turing, Church, Minsky, Rice, Searle, Harnad, and others. In this section, The field of philosophy and computing is the topic of a chapter we begin by giving an outline of what a computer is, focusing entitled “Philosophy of Computing and Information Technology” primarily on the abstract notion of computation developed by in the forthcoming book Philosophy of Technology and Turing. We then consider what it means for something to be Engineering Sciences (ed. A. Meijers), which is Volume IX in computable and outline some of the classical problems that the Elsevier Handbook of the Philosophy of Science series and cannot be computed by a Turing machine. Having considered is expected to come out in the fall of 2009. In this 27,000 word which kinds of problems are Turing non-computable in chapter, we aim to provide an extensive overview of topics and principle, we then consider problems that are so complex approaches in the philosophy of computing and information that they cannot be solved in practice. In both cases, one of technology (IT). The work in this area is diverse, with topics our aims was to show that oft-discussed problems like the ranging from the nature of computational systems to the political halting problem and Rice’s theorem are not of mere theoretical regulation of cyberspace, and from the ontological status of interest, but that they point to problems that still lie at the core virtual worlds to the limitations of artificial intelligence. Our of computer science. Finally, computing is always computing of chapter follows in the footsteps of one previous introduction something; hence, we conclude this section with a brief outline to the field (Floridi 1999) and two previous anthologies (Floridi of central notions like data, representation, and information. 2004; Moor and Bynum 2002). Although much of the section on philosophy of computing In setting up our chapter, we have invested much effort draws on classical problems and discussions, we also draw into providing an appropriate structure to the field as a whole, on more recent philosophical work on computation, including which respects the conceptual and historical development discussions of hypercomputation and some recent approaches of the field. We ended up distinguishing five subfields within in philosophy of information. the loosely structured field of computing and IT: philosophy Following the section on philosophy of computing, we then of computing, philosophy of computer science, philosophy consider “philosophy of computer science.” This was for many of artificial intelligence, philosophy of the Internet and new years a neglected field of study, perhaps explained by the fact media, and computer and information ethics. Conceptually, that philosophy of science has tended to focus on sciences that these areas have distinct subject matters and involve distinct aim to represent reality, not on fields that model and design philosophical questions, as we try to show in the corresponding artifacts. In recent years there have, however, been a number sections in our chapter. Historically, the philosophy of AI and of substantial contributions to the field, a field that, in analogy to philosophy of computing are the oldest fields, having found the philosophy of science in general, aims to analyze, interpret, their footing in the 1960s and 1970s, respectively. In the 1980s, and clarify the central concepts, aims, assumptions, methods, a separate field of computer ethics takes shape that studies scientific status, and the role in society of computer science ethical and professional issues in computing. The 1990s see the and its various subfields. Following a comprehensive subsection emergence of philosophical studies of the Internet and other on the nature, scope, and methods of philosophy of computer new information and communication media, and an increased science, we give special attention to philosophical issues raised interest in social, cultural, and political issues. Finally, the late within some particular computer professions, including issues 1990s see the emergence of a distinct philosophy of computer concerning programming languages, software engineering, science, as opposed to a philosophy of computing. data modeling, information systems management, computer simulation, and human-computer interaction.

— 15 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

In the fourth section, we turn to philosophy of Artificial Undoubtedly, many will find that their niche of specialty Intelligence (AI). AI is arguably the area of philosophy of computing has been underrepresented in this chapter, but one of our that has received the most attention from philosophers, in part main challenges was to condense as much of the philosophy due to its close ties with numerous subdisciplines of philosophy, of computing and information technology field as possible such as philosophy of mind and philosophy of language. It is within a limited number of pages, as well as making it clearly not possible to give full justice to this field, but we start out accessible well beyond the community of scholars who are by discussing some of the fundamental assumptions, challenges, already doing philosophy of computing and IT. Our aim and and contributors in philosophy of AI, ranging from symbolic AI to hope is to introduce the philosophy of computing and IT field connectionist and anti-representationalist approaches, as well as to anyone interested in this intersection, be they philosophers, the closely related approaches that can be found in philosophy of computer scientists, or from other disciplines, be they students artificial life and dynamic systems theory. Given the purpose of our or professionals. We also want to showcase the field as a chapter to reach both philosophers and computer scientists, we good example of interdisciplinary research, and to give an continue by addressing a number of philosophical issues raised impression of what philosophers and computer scientists alike by particular applications of AI, including knowledge engineering, have brought to the table. Still, in trying to show how the many robots, and artificial agents. We conclude this chapter with a different subcategories are related and how they do, after all, discussion of the (potential) social impact of AI as well as the way form a coherent yet many-faceted discipline, we hope and in which AI research has shed new light on notions like “risk,” believe that the chapter will also be of interest to those who “responsibility,” and formalization of morality. are already acquainted with the field. The rise of the personal computer and multimedia References technology in the 1980s and the Internet and World Wide Floridi, L. 1999. Philosophy and Computing: An Introduction. London Web in the 1990s ushered in a new era in which the computer and New York: Routledge. became part of everyday life. This has brought along major Floridi, L. (ed.) 2004. The Blackwell Guide to Philosophy of Computing changes in society, including changes in the way people and Information. Oxford: Blackwell. work, learn, recreate, and interact with each other, and Moor, J. and T. Bynum. (eds.) 2002. Cyberphilosophy. Oxford: in the functioning of organizations and social and political Blackwell. institutions. It has even been claimed that these technologies are fundamentally changing human cognition and experience. These social and cultural changes have prompted philosophers Comments on “Philosophy of Computing and to reflect on different aspects of the new constellation, ranging from the epistemology of hyperlinks to the ontology of virtual Information Technology” in Handbook of the environments and the value of computer-mediated friendships. Philosophy of Science, Vol. 9 Technology and In our fifth section, we tie these different investigations together Engineering Science, Part 6 Philosophical under the rubric “philosophy of the Internet and new media.” Whereas most of the contributions in the other sections have Issues of Engineering Disciplines been in the analytic tradition in philosophy, a large part of the research in this area is taking place in the Continental Gordana Dodig-Crnkovic Mälardalen University, Sweden tradition, and includes phenomenological, poststructuralist, and postmodernist approaches. Additionally, philosophical work in this area is often affiliated with work in social theory I. The Context: Handbook of the Philosophy of Science, and cultural studies. We start this section by giving a broad Technology, and Engineering Science outline of new media and discuss theories on how society has First of all, the précis written by Brey and Søraker is an excellent increasingly become an information society. We also consider presentation of their work. Therefore, I will continue by studying epistemological and ontological issues related to the Internet how this work reflects today’s state-of-the-art of the dynamic and other new media, as well as their status as a platform for field of “Computing and Information Technology.” communication and virtual communities. Closely related, we To set the stage, let us start with the project of Handbook of conclude this section with a discussion of how our identity the Philosophy of Science (HPHS) (D. Gabbay, P. Thagard, and J. might be affected by the disappearing barriers between body Woods, eds.), of which the chapter “Philosophy of Computing and technology and between real and virtual selves, through and Information Technology” is a part. The HPHS consists such notions as “cyborgs” and “avatars.” of the following volumes: 1. General 2. Physics 3. Biology In the sixth and final section, we turn to computer 4. Mathematics 5. Logic 6. Chemistry and Pharmacology 7. ethics and the shared concerns of both computer scientists Statistics 8. Information 9. Technology and Engineering Science and philosophers about social and ethical issues related to 10. Complex Systems 11. Ecology 12. Psychology and Cognitive computers. Throughout the other sections, in particular in Science 13. Economics 14. Linguistics 15. Anthropology and sections four and five, we consider a number of ethical issues Sociology 16. Medicine in the contexts in which they naturally arise. In this section, Of special interest for us are volumes 8. Philosophy of however, we focus on a number of more general issues in Information (where “Information” stands for “Science of computer ethics and outline its main topics, contributors, and Information” along with other sciences) and 9. Philosophy meta-ethical challenges. Some of the topics raised in this section of Technology and Engineering Sciences (as it includes the are privacy, security, computer crime, freedom of expression, “Philosophy of Computing and Information Technology” equal access, and intellectual property, as well as a number of chapter, where “Information” points specifically to “Information foundational issues. Again, with an eye to the audience in mind, Technology Science,” among other Technology and Engineering we conclude the section with a discussion of “value sensitive Sciences. design,” which is an inherently interdisciplinary approach that aims to analyze embedded values in computer software and Volume 8. Philosophy of Information (Science), (P. systems, and to devise methodologies for incorporating values Adriaans and J. van Benthem (eds.), includes: Concepts into the design process. and History; Epistemology and Information; Information and Philosophy of Language; Modern Trends in Philosophy of

— 16 — — Philosophy and Computers —

Information; Philosophy of Learning; The Quantitative Theory coming editions of the HPHS will be accommodating for of Information; Logico-Semantic Theories of Information; new developments, and it is highly probable that those new Algorithmic Complexity; Learning, Epistemic Logic and developments will also affect the whole structure and the Information Update; Information Structure and Belief Revision; relationships on the map we see now. That is nothing to worry Information, Processes and Games; Information and Beliefs in about; on the contrary, the dynamics of the development Game Theory; Information in Computer Science; The Physics in sciences and philosophy and all surrounding and related of Information; Information in Social Science; Information in knowledge-generating fields is just a sign and the consequence Artificial Intelligence; Information and Cognitive Science; and of the transition to a new paradigm of knowledge production, Information in Biological Systems. equally unavoidable as epoch-making. Volume 9. Philosophy of Technology and Engineering So much for the background to the chapter, and so much Sciences, A. Meijers (ed.), addresses six fields: Technology for the Handbook of the Philosophy of Science contents and and the Sciences; Artifact Ontology and Artifact Epistemology; organization—now a few words about the Brey-Søraker chapter Philosophy of Design; Methodological Issues; Normativity and itself. Values in Technology; and Philosophical Issues of Engineering Disciplines. III. Brey-Søraker on Philosophy of Computing and Brey-Søraker’s chapter is included in the Philosophical Information Technology Issues of Engineering Disciplines, along with articles on The chapter heading “Computing and Information Technology” philosophical issues of agricultural technology, medical itself is essentially ambiguous as it might be read as (Computing technology, nanotechnology, and architecture. and Information) Technology but also as Computing and (Information Technology). In the chapter we will find both views Why is this HPHS context for the Brey-Søraker chapter so of Computing—as science in Chapters 1-2 and considerable important? In the first place because it says something about parts of Chapter 3, and as technology in Chapters 4-5. The five our current understanding of science and its relations with subfields: Philosophy of Computing, Philosophy of Computer technology and philosophy. Science, Philosophy of Artificial Intelligence, Philosophy of II. A Major Ongoing Paradigm Shift the Internet and new media, and Computer and Information We are living in the midst of a major paradigm shift in sciences Ethics, are chosen according to criteria well explained in the caused by information and communication technologies, Synopsis. This chapter necessarily gives a condensed overview which may be termed “info-computational turn.” Changes of a huge field of knowledge in the context of the HPS, and its in the knowledge production process, based on dramatically choice of subfields can be judged against two groundbreaking increased information storage, processing, and communication books in the field, Blackwell Guide and Cyber Philosophy. For capabilities, are broad and sciences are just one among the comparison, here is how they are structured. domains of knowledge production they affect (Dodig-Crnkovic The Blackwell Guide to Philosophy of Computing and 2003). Philosophy is also influenced, on the level of working Information, L. Floridi, (ed.) Blackwell, 2004 is organized as tools, conceptualizations, and models, as clearly visible from follows: CAP research community activities. We are witnessing an 1. Four Concepts (Computation, Complexity, System, interesting development that makes information and knowledge Information) (including scientific and philosophical knowledge) increasingly 2. Computers in Society (Computer Ethics, Computer- available. This is related with the simplicity of access and mediated Communication and Human–Computer easiness of publishing on the World Wide Web. In the past, the Interaction, Internet Culture, Digital Art) barriers between different fields of knowledge and information production were high and physical. In order to get information 3. Mind and AI (The Philosophy of AI and its Critique, you would have to visit the institution, talk to scholars, borrow Computationalism, Connectionism, and the Philosophy books, and similar efforts that were often cumbersome and, of Mind) every so often, practically impossible. 4. Real and Virtual Worlds (Ontology, Virtual Reality, The Nowadays as a rule no such barriers exist. Browsing Physics of Information, Cybernetics, Artificial Life) between different fields of enquiry is frictionless, borders 5. Language and Knowledge (Information and Content, invisible. In the course of writing an article one can without Knowledge, The Philosophy of Computer Languages, effort reach references, find source information, quotations, Hypertext) study relations to other fields, etc. Automated intelligent 6. Logic and Probability (Logic, Probability in Artificial searches and knowledge management techniques are rapidly Intelligence, Game Theory: Nash Equilibrium) developing. As a consequence, interdisciplinary research is 7. Science and Technology (Computing in the Philosophy becoming not only possible but also very attractive because of Science, Methodology of Computer Science, it explores white spots on our knowledge maps—domains virtually impossible to reach in the past. This process of vivid Philosophy of Information Technology, Computational cross-disciplinary, inter-disciplinary, and trans-disciplinary Modeling as a Philosophical Methodology) communication establishes new connections and new research Of course, more than 350 pages allow for a detailed conceptual fields that make the traditional division of sciences (and then analysis and the format affects even the choice of themes. Here consequently of their corresponding philosophies) appear more science and technology are interlaced. and more artificial. The Blackwell Guide focuses on conceptual analysis of In this moment of exuberant activity it might look like an basic information and computation phenomena and their impossible task to map the territory of philosophy of science manifestations in knowledge generation and communication and set the clear cut borders between the fields as landscape mechanisms and practices, including technology. is changing before our eyes. Nevertheless, it is both historically CyberPhilosophy: The Intersection of Philosophy and necessary and significant to find out the present state-of-the- Computing, J. H. Moor and T. W. Bynum (editors) is organized art in philosophy of science, the outcome resembling a high into following chapters: speed photograph of a dynamic system. We may expect that

— 17 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

1. Minds and Computers [Synthetic Neuroethology; In [Dodig-Crnkovic 2003 and 2006] I discuss the question Computer Modeling and the Fate of Folk Psychology; Philosophy of outlining the field of Computing by comparison of what then of Mind, Cognitive Science, and Pedagogical Technique; existed and what seemed to be needed in the Philosophy of Phenomenology and Artificial Intelligence] Science perspective: 2. Agency and Computers [Adaptable Robots; A Radical According to ACM/IEEE 2001 [Computing Curricula], Notion of Embeddedness: A Logically Necessary Precondition Computing can be described as encompassing for Agency and Self-Awareness; Building Simple Mechanical Computer Science, Computer Engineering, Software Minds: Using LEGO Robots for Research and Teaching in Engineering and Information Systems. Philosophy] 3. Reality and Computers [What Is the Philosophy of The German, French and Italian languages use the Information?; The Substantive Impact of Computers on respective terms “Informatik,” “Informatique,” and Philosophy: Prolegomena Computational and Information- “Informatica” (Informatics in English) to denote Theoretic Metaphysics; Computation and Causation] Computing. 4. Communication and Computers [Philosophy for Computation is the study of the fundamental characteristics Computers: Some Explorations in Philosophical Modeling; The of the process of information processing. Computation Stanford Encyclopedia of Philosophy: A Developed Dynamic and information can be understood as representing two Reference; Cultures in Collision: Philosophical Lessons from complementary phenomena—information as a structure and Computer-Mediated Communication] computation as a process on that structure (its dynamics) 5. Ethics and Computers [Heuristic Methods for Computer [Dodig-Crnkovic 2006]. Ethics; Lilliputian Computer Ethics; Deontic Logic and I wonder if computation, computer science, and even Computer-Supported Computer Ethics] artificial intelligence really are best seen in the context of Cyberphilosophy revolves around the computer as a cultural philosophy of technology or rather should be recognized as a artifact, and each of its chapter titles contains “and Computers,” subject of philosophy of science. This division in subfields by so it addresses classical philosophical themes in the context Brey and Søraker follows the HPHS global structure, in which of computer technology, which clearly indicates that the book Information is seen through the lens of Philosophy of Science belongs to the Philosophy of Computing Technology. while Computing is sorted under Philosophy of Technology. Well aware of the above work, Brey and Søraker choose Computing, however, has strong reasons for being placed their own strategy in the choice of subfields of their chapter and among sciences and studied by Philosophy of Science at least in they focus on what already exists as developed philosophical question of computation, computer science, and computational fields. One would say, if Moor and Bynum start from central aspects of artificial intelligence as equally important scientific philosophical themes and explore how those are affected by discipline as Information Science. Or else if “Information” would computers, Brey and Søraker, appropriate for the handbook of be seen as corresponding to “Informatics,” Computation has its philosophy of science format, focus on what there is of existing, given place in the book on Information Sciences. Computation well-developed philosophy on the intersection of computing stands in tandem with Information, as accurately given in and philosophy. Blackwell Guide conceptual framework. In the Synopsis, they accurately explain the approach they Computing Curricula 2005 identifies the following five have: “The purpose of the chapter is to review and discuss sub-disciplines of Computing: Computer Science, Computer the main developments, concepts, topics and contributors in Engineering, Information Systems, Information Technology, the intersection between philosophy and computing, as well and Software Engineering. as provide some suggestions on how to structure the many In light of the above, it is good to see that Brey and Søraker subcategories within what is loosely referred to as philosophy make a clear distinction between Philosophy of Computing of computing.” (which, understood as the study of the nature, possibilities, and limits of computation in effect is a Philosophy of Computation) IV. Computing vs. Informatics vs. Informatics and Philosophy of Computer Science. Similar to the way Technology Computer Science is a subfield of Computing in Computing In this initial phase of the development of Philosophy of Curricula, Philosophy of Computer Science is a subset of Computing field (less than ten years old) conceptual jungle is Philosophy of Computing, while Philosophy of Computation the major obstacle to overcome. has a broader scope of understanding basic mechanisms in The recognition of this fact is clearly visible from the Brey- its addressing fundamental questions of computation process Søraker Synopsis, as they say: in general. Traditionally, Computer Science is focused on present-day computers and their features and less on possible In setting up our chapter, we have invested much substantially different computing mechanisms, which typically effort into providing an appropriate structure to the are the interest of Theory of Computation (e.g., computation field as a whole, which respects the conceptual and beyond Turing Limit and bio-computation). In short, Computing historical development of the field. We here debate is interested in the existing computation, while Computation is a proper name for the field (including proposals more about computation as it can be. like “philosophy and computing,” “philosophy of computing,” “computational philosophy” and V. The Significance of Computational Turn “cyberphilosophy”) before settling on “Philosophy of Finally, one position of the Synopsis I definitely want to point computing and information technology.” out and comment on is the following: “we also discuss Of the mentioned suggestions, the accepted one is surely the the hypothesized emergence of a ‘computational turn’ in best in this context because it clearly declares its focus on philosophy and the rise of ‘computational philosophy’, and technology. We are talking about the philosophy of science argue that the idea of a computational turn in philosophy has approach to technology and engineering. perhaps been overstated.” Allow me to disagree. In my view “computational turn” is both clearly evident in philosophy today, and it is setting new — 18 — — Philosophy and Computers — ground for increasingly far-reaching changes in philosophy Dodig-Crnkovic, G. 2006. Investigations into information semantics and in coming decades. In case of emerging new kinds of ethics of computing. Mälardalen University Press. http://www.diva-portal. Natural Philosophy, the role of computational turn cannot be org/diva/getDocument?urn_nbn_se_mdh_diva-153-2__fulltext.pdf. overrated. It is absolutely essential. Nascent computational Indiana University School of Informatics student information. http:// Natural Philosophy will play a fundamental role in unification www.indiana.edu/~ocmhp/110901/text/informaticist.html. of sciences in a network with complex architecture, providing communication means and meaning-production mechanisms—something that the scientific community feels is not only desirable but even absolutely necessary for the future ARTICLES ON development of sciences and their corresponding philosophies. In the context of a Handbook of Philosophy of Science, this fact AUTONOMOUS WEAPONS is very important. Conclusions The main role of the very significant project of Handbook of Telerobotic Weapons Systems and the Ethical the Philosophy of Science and its volume on the Technology Conduct of War and Engineering Science is to inform and orient the future comprehension of sciences. We may see the process of thinking J. P. Sullins through and capturing the present state of understanding Sonoma State University of sciences and their philosophies as a process of learning and gaining new insights by disclosing implicit and hitherto Abstract uncovered structures and relationships. I will argue that telerobotic weapons systems need to be more In this context Brey-Søraker’s work makes an important carefully designed and deployed in order to minimize the original contribution in establishing the fact of increasing negative impacts they could have on the already shaky ground importance of philosophical understanding of Computing and that supports the ethical conduct of military actions. Using just Information Technologies. It also addresses several fields of war theory as a background for discussion, I suggest that the Computing which belong to fundamental sciences together design of telerobotic weapon systems can foster values that with Information Science (Computation, Computer Science, either extend just war ideals or not. While recent instances of the theoretical aspects of AI). The work builds on the pioneering use of this technology have appeared to be morally nebulous, I books of Floridi and Moor & Bynum and it prepares the ground suggest some ways in which this technology may be designed for the establishment of the future understanding of computing and used in ways that will enhance just war practices. For and information technologies, including basic computational example, telerobotic weapons might serve to limit casualties mechanisms. It follows analytic and continental traditions, and increase command and control of pilots and soldiers. Also, social theory, and cultural studies in relevant parts. the same technology used for military reconnaissance can, and Several times the readership is mentioned as “philosophers” should, be used by the media to extend the abilities of war and “computer scientists,” while it is probable that it will be much reporters in order to enhance accountability and scrutiny of broader. “Computer scientists” may well be “informaticists,” in military affairs. Finally, I address the problems that occur when the following sense of Informatics, from the Indiana University the operators of these weapons become too distant and isolated School of Informatics, website: from the violence at the other end of their video monitors. Informatics students choose their areas of Introduction specialization from a number of diverse fields, such On November 2, 2002, one hundred miles east of Sanaa, the as: new media, health sciences, design, biology, capital of Yemen, a car sped along a desert road. Inside were six chemistry, cognitive science, computer science, suspected al-Qaeda terrorists, including Abu Ali al-Harithi, who communication and culture, economics, fine was suspected by U.S. intelligence as the leader of the bombing arts, geography, instructional systems technology, of the USS Cole in 2000 and also the attack of a French oil journalism, linguistics, mathematics, political science, tanker off the coast of Yemen in October 2002. High above, the psychology, public and environmental affairs, and car was identified and tracked by a Predator unmanned aerial telecommunications. vehicle. The Predator was flown by remote control, possibly One shall keep in mind that we are in the midst of a through a satellite hookup, by CIA pilots many hundreds of paradigm shift in which huge changes in knowledge production miles away. It is likely that senior CIA officials decided that the practices in general, and scientific knowledge in particular are identification of the target was clear and that there would be going on. minimal potential for civilian damage, so they authorized the pilots to fire a small, but powerful, Hellfire missile. From the References video and telemetry data on their remote control panels, the Floridi, L. (ed.) 2004. The Blackwell guide to philosophy of computing pilots set up and executed the attack, killing Abu Ali al-Harithi and information. Oxford: Blackwell. and the five other passengers in the car. For decades prior to Moor, J. and Bynum, T. (eds.) 2002. Cyberphilosophy. Oxford: this operation, unmanned aircraft had been used as tools for Blackwell. reconnaissance and surveillance by the U.S., Israel, and other Computing Curricula 2005: The Overview. http://www.acm.org/ governments, but this attack was the first successful use of the education/curricula-recommendations. Predator in a combat role that caught major media attention. Handbook of the Philosophy of Science, edited by Dov Gabbay, Paul Prior to this mission, a few unsuccessful attempts had been Thagard, and John Woods. Elsevier 2006. http://www.elsevier.com/wps/ made to use the Predator drone in a combat roll both in the find/bookdescription.cws_home/BS_HPHS/description. Balkan conflict and another few dozen successful minor combat Dodig-Crnkovic, G. Shifting the paradigm of the philosophy of science: operations in Afghanistan (Defense Update 2007 and Pincus The philosophy of information and a new renaissance. Minds and 2002). A new chapter in armed conflict has now begun that Machines: Special Issue on the Philosophy of Information 13 (2003). requires ethical analysis. http://www.springerlink.com/content/g14t483510156726. — 19 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

The ethical conduct of war is a murky and complex issue Predator Series Unmanned Aerial Vehicles (UAVs) but it is important to notice the special concerns that remotely These are the most successful of the unmanned aerial vehicles. operated robotic weaponry raises. We must attempt to analyze At the time of the writing of this paper, more than one hundred these emerging military technologies and determine if anything are in service with the U.S. military and intelligence agencies, constructive can be said about the ethical use of such machines. and six are in use by the Italian Air force.2 In one year of the The development of unmanned fighting vehicles for use in the war in Iraq, “…June 2005 to June 2006, Predators carried out air, sea, and land is currently a top priority for defense research. 2,073 missions, flew 33,833 hours, surveyed 18,490 targets, and Unmanned aerial vehicles have proven to be quite successful participated in 242 separate raids” (Singer 2009, 35). The maker in the U.S. War on Terror being useful both militarily and in the of the aircraft, General Atomics Aeronautical Systems Inc., favorable press they receive. They have been used to kill or aid in claims on its website that a Predator has a range of 400 nautical the killing of suspected terrorists and those who happen to be in miles, a ceiling of 25,000 feet (7,620m), is equipped with and can close proximity to these targets in numerous locations around the carry 450 lbs (204kg) internal payload of weaponry, and much globe, occasionally even hitting a number of high profile targets. more can be carried externally. It is 27 feet (8.23m) long, and Under development or just recently deployed are telerobotic has a wingspan of 48.7 feet (14.84m). It has an impressive array weapons systems, such as the SWORDS robot, that are designed of laser range finders and cameras with both visual and infrared to aid frontline combat operations from the ground, and various light sensing ability. Civilian and scientific communities have guided torpedoes and missiles have been developed for use at built variations on this design to increase the endurance of the 1 sea. In this paper I will analyze these telerobotic weapon systems machine up to 49 hours of continuous flight, and they have also from the perspective of the philosophy of technology and the just altered the imaging capabilities for their specific needs. war tradition in ethics. Just war theory is nowhere near perfect The pilots of these aircraft can be located thousands of but it is preferable to no holds barred violence, especially given miles from the actual machine. Pilots located in Nevada in the that a country may find itself involved in violence against its better U.S. have flown Predators in combat actions in Iraq. In recent judgments and/or interests due to political contingency beyond hearings in the U.S. capital, General Deptula reported that: its control. What I am asking in this paper is whether telerobotic technologies enhance or degrade the practice of just war in both When a sniper was pinning down Marine ground the move towards war, and the conduct of the war itself. After a forces in Iraq, a Predator UAV flown by the Air Force brief overview of current telerobotic military technologies, this from Nevada, spotted and identified the insurgent. The paper will identify and address the most important ethical issues Predator UAV delivered video of the sniper’s location surrounding the design and use of this technology. directly to a Marine controller in the fight, and he used that video to direct a Navy F/A-18 into the vicinity. Then The Technologies in Question the Navy jets’ laser bombs were guided to the enemy Telerobots are one of the most successful applications of position by the Predator UAV laser designation of the recent robotics technologies. A telerobot is any machine that target, eliminating the sniper. This engagement took is controlled by an operator at a distance and used to perform less than two minutes. (As reported by AF News)3 actions that otherwise would have been performed by a human agent. Telerobots have a decades long history and have been Not only are these long-range drones piloted out of Nevada, successfully employed in many roles running the gamut from but the Air Force UAV Battle Lab there is also designing the next planetary exploration to sewer repair. These robots allow generation of these machines in Indian Springs, Nevada, just human agents to extend their presence into arenas that are too outside of Las Vegas. It has been reported that they are working distant, dangerous, or repugnant for humans to easily operate on the Reaper, which is a much larger unmanned fighter aircraft, in. One job that fits all three of these criteria is the modern other longer ranged drones, and tiny insect sized UAVs for use battlefield. Because of this, telerobotics is generating a lot of in surveillance, border patrol, and extra terrestrial exploration interest as a force multiplier, since it promises to be a way to on Mars (Knapp 2005; Singer 2009, 40). These are just a small allow a small number of professional operators to remotely sample of the large number of different kinds of UAVs in the control robots in both support and direct combat functions. The design phase and in operation. They are certainly going to be political motivation for the pursuit of these technologies is clear. a commonplace technology in the near future that will spin If successful systems can be built, then they will provide a way off from battlefield operations for use by police, firefighters, for warfare to be conducted with a greatly reduced casualty rate. researchers, and scientists. The various bomb removal machines already seeing action are Battlefield Extraction-Assist Robot (BEAR) a prime example of this phenomenon. But as these machines The Battlefield Extraction-Assist Robot, or BEAR, is designed to move from support roles to direct combat, important ethical assist the removal of casualties in dangerous combat situations, questions are quickly raised. which would be too risky for human medics to attempt. Built In warfare, telerobots can be relatively simple machines by Vecna Robotics in the USA, this robot has passed its proof such as a roadside bomb remotely operated by a cell phone. of concept phase and is nearing its final testing phase. The They can also be quite complex, such as the military robots that Vecna website and press releases claim that the machine is are being designed and used on the battlefields of today by large a “hyper agile” humanoid form robot that can balance and industrial nations. In this paper I will be mostly addressing the carry a human sized casualty to safety.4 This is an amazingly upper end of this technology, as the ethical issues raised by a designed robot that can reconfigure itself into a number of roadside bomb are not much different from those found in any forms to facilitate numerous other tasks besides casualty other conventional weapons system used in the conflicts of the removal. A human operator located at a safe distance from the hour. But, as we will see, there are some interesting differences robot controls the machine wirelessly. Built under numerous found in the advanced telerobots and weapons platforms defense research grants, this machine is likely to see service that are semi-autonomous in their remote actions. These are in the next few years. more interesting from an ethics standpoint, as at least a part of the decision to use lethal force is provided by the design and iRobot PackBot and Autonomous Ground Vehicle programming of the machine itself. First, let’s briefly look at The iRobot corporation has a very successful robot called some paragon examples of these technologies. the PackBot (sometimes also referred to as “PacBot”) that

— 20 — — Philosophy and Computers — has been used for reconnaissance, bomb removal, and law Ethical Use of Technology enforcement. The iRobot corporation describes the PacBot as A technology is used ethically when it is intelligently controlled a “battle-tested Unmanned Ground Vehicle (UGV) designed to further a moral good. The philosopher Carl Mitcham explains for Military Operations Urban Terrain (MOUT) and other 21st- that the intelligent control of technology requires: century battle missions. This lightweight, rugged robot can be hand-carried and deployed by a single soldier.”5 This robot is (1) Knowing what we should do with technology, the built to be highly durable and its small size and innovative track end or goal toward which technological activity ought design allow it to enter many areas that are inaccessible to its to be directed; (2) knowing the consequences of human operators. The manufacturer claims that the machine technological actions before the actual performance is both telerobotic and semi-autonomous: of such actions; and (3) acting on the basis of or in accord with both types of knowledge—in other words, PackBot Scout’s software architecture supports translating intelligence into active volition. (Mitcham multiple modes of operation ranging from basic tele- 1994) operation through full autonomy. Assist behaviors allow operators to focus on mission success instead We can easily extrapolate that the ethical use of robotic of the robot. For example, PackBot Scout’s intelligent weapons technology occurs only when that technology is power management system constantly monitors intelligently controlled and advances a morally just military the health and status of the batteries to ensure long action. This leaves us with a tall order. If there are no situations mission life. Its all-digital architecture, with full data in which warfare is an ethical act, then there are no situations in and video logging, aids in post-mission analysis.6 which robotic weapons are ethical technologies. This is a real possibility that must be considered. For the sake of argument, In addition to this proven technology, iRobot, in conjunction and to advance the discussion in a useful direction, I will with John Deere Tractors, is designing autonomous ground assume that there are at least some plausible situations in which vehicles with a hope of providing a robotic vehicle that can military force might be used in an ethical manner. To advance “serve numerous important roles, acting as unmanned scout, this position I will use the Just War tradition in philosophy, ‘point man’, perimeter guard, pack/ammo/supply carrier and especially that tradition as it is transmitted to us via Michael 7 more for soldiers, marines and airmen.” While this is still in the Walzer’s expertly argued book, Just and Unjust Wars (2006). I design phase, it is of obvious value for combat support roles. know that just war theory is not unproblematic but it serves as TALON for Armed Reconnaissance (SWORDS) an excellent starting point for this discussion. The TALON robot, built by Foster-Miller, has been a workhorse For the purposes of this paper I will take the stand that for the U.S. military for some time now as a bomb removal and robotic weapons technologies are used in an ethical manner in reconnaissance aide, completing tens of thousands of missions. only those situations where the operators (and in some cases It is small in size, similar to the PacBot, and can operate in the machine itself) know what morally justifiable outcome they highly contaminated areas and as such was used to locate are seeking by using lethal force, and what the consequences possible survivors at the World Trade Center. The engineers at of those actions are most likely to be, and that these actions do Foster-Miller claim that it is fast and easy to control.8 What is not contradict the execution of a principled and just war. Much different about the TALON is that it is the first ground telerobot can be said about this, but the best course of action will be to to be fitted with lethal weaponry and was deployed in Iraq in look for potentially troubling uses of this technology and see if 2007. The TALON for Armed Reconnaissance or SWORDS is they run afoul of the maxim I just described. fitted with “M240 or M249 machine guns or Barrett .50 caliber The Distancing Offered by Telepresence rifles for armed reconnaissance missions,” and “[a]lternative The first ethical issue I want to address is the effects of weapons, including 40 mm grenade launchers and anti-tank distancing the operator of the robot far from the location of rocket launchers, continue to be evaluated by the U.S. Army.”9 military activity. This capability is the obvious driving motivation The present status of the use of these machines is in doubt; still, for building these technologies in the first place. A successful the ethical use of this new development in ground telerobotics military robot is one that accomplishes the actions of a human is worth looking at, and we will now turn to that topic. soldier, sailor, or pilot in a way that is comparable or superior Ethical Issues Raised by These Technologies to what a human would have done in the same situation. If this The Yemini incident in 2002 is a good example to use to illustrate can be achieved, then it seems perfectly correct to remove the some of the more difficult ethical issues that can be raised by irreplaceable human agent from danger and replace that person the use of these technologies in modern combat operations. The with a more expendable robot. robot’s operators were at a great distance from the action; it is From the point of view of the individual that might have possible that the pilots of this drone were located in the state of been harmed or killed if the robot were not there to stand in as Nevada in the U.S. and, as such, these pilots form a kind of shift a proxy, this is quite possibly an unmitigated good. These robotic work warrior who may be flying via cyberspace in a combat systems allow the user to be well away from the dangers the zone for eight hours, then home by the pool for R and R in the robot is facing. The amount of distance from the combat varies. evening. There is also a blurring of the battlefield to include The operators of ground combat and support telerobots in use countries we are not technically at war with, such as Yemen. today are usually, at the most, just a few hundred yards from More subtly, there is the unknown factor of the oversight of the the machine, but some UAVs allow the pilot to be thousands use of these weapons, and the fact that they make a big hit with of miles away. The support teams for these machines do have media, which fosters little reason for the public to question or to be within the flight radius of the aircraft or the operational critique the use of these technologies. We need to look at each theater of the ground robot, but they can still be well away from of these, and other, issues in more detail, which I will do in the the most dangerous areas of the modern battlefield. following sections. But first we need to discuss in general terms It seems like distancing from combat is a moral good that what it means to ethically use a piece of technology. these machines provide. However, when we look at it from a different perspective we can see that there are a number of potential ethical challenges to be faced when we distance

— 21 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — human combatants from the battlefield. The most disturbing tradition and lead to a war without limits, which cannot be potential is that distancing helps facilitate political arguments philosophically justified in any way (see Walzer 2006, 203-04). that propagate the impression that modern warfare can be But are we fighting well when we risk significant double effect a sanitized and surgical affair, and that this illusion could be through an act, not of open combat, but of assassination? a powerful disincentive towards accountability and media Assassination of key figures in an oppressive regime might be scrutiny into the affairs of military and intelligence forces. justifiable in extraordinary circumstances, but what if we are not Already, the successful strikes made by these weapons are dealing with political figures but instead shadowy criminals? The compelling media fodder; examples include the Yemini attack Iraqi insurgents are some sort of cross between criminal and described at the beginning of this paper, the strike on a hotel paramilitary organization, and it is not clear that it is possible outside of Kabul where the al Qaeda military commander to conduct war as it is traditionally conceived against these Mohammed Atef and one hundred other people were killed kinds of foes. by an air strike after a Predator tracked him and his party going We are stumbling on a larger ethical issue here as well: into the building, Predators destroying Iraqi anti aircraft positions Can one prosecute a just war against terrorists, or are we prior to the invasion and coordinating air strikes on insurgents in doomed to having to fight with the reduced sense of morality Iraq and Afghanistan. All of these have provided ample exciting of our enemies? There have been notable failures in the use video images that play well on YouTube and in the twenty-four of telerobotic weapons: NBC News reported that, in January hour news cycle (Kaplan 2003, BBS 3/2003, A F News 4/2007, 2006, thirteen civilians were killed in a failed Predator strike on Singer 2009). We have all seen the grainy infrared or black and Ayman al-Zawahri an al Qaeda deputy (Miklaszewski 2006). This white videos captured by the cameras of UAVs as they help question is a bit more than can be addressed here, but we are identify, target, and eliminate the insurgents and their vehicles left with something that is directly relevant to our discussion. moving on the ground. The moral use of robotic weapons systems must not help to These images provide compelling evidence that we can mask the complexities of the situations they are being used in; eliminate our enemies from the air with ease and precision, if they mediate our interpretations of these events in a simplistic fostering an illusion of military omnipresence and omnipotence. way, then their use is unethical. The distant and omnipotent The videos we see are highly selective and focus on the big viewpoint provided by the unmanned aerial vehicles in use successes. This selective sample will obviously result in a today sometimes serves to distort the complexities of modern skewed opinion of the technology. Admittedly, if surgical strikes war and make it more difficult to fight these conflicts in a just are indeed possible, then they are more permissible under and proportionate way. the just war tradition than other indiscriminant forms of aerial It can be argued that these technologies facilitate a greater attack. It is important to remember that mistakes can still be understanding by the individual soldier of the context of the made with this technology and other unintended casualties can military action that would be lost by forcing that soldier to fight in result from their use. In the just war tradition certain instances the conventional manner. This is indeed true and it is foolish to of so-called double effect (civilian casualties caused while argue that we intentionally limit our soldiers from accessing all attacking military targets) are allowed, if the act is a legitimate the information that they can get, much of which is increasingly act of war, the primary effect is intended to destroy legitimate provided by telerobots. P. W. Singer, in his book Wired for War military targets, and the means of attack are acceptable and not (2009), reports that the technology we are using now does not excessively cruel or violent (Walzer 2006, 153). So we want to provide flawless telepresence that allows the distant operators use these machines in a principled and just manner, then we of these machines to completely understand the situation their can divide the use of robotic weaponry into three categories: robots are encountering; voice, text, and email are not enough acceptable risk of double effect, questionable risk of double to prevent miscommunications between the pilots, their effect, and unacceptable risk of double effect. We must work commanders, and the troops on the ground (336-37). to ensure robotic weaponry such as UAVs help us make more What I am arguing is that we have to work hard to make accurate assessments of the acceptability of the double effects sure that the information remote operators are receiving does of our aerial bombardment. If we accomplish this, then they not inappropriately remove or alter information that allows are used in a just manner, if they only make it seem like we are the soldier to place the action in its proper ethical context. better at these judgments, because they provide compelling Imagine, for instance, technology that gives the soldier but inaccurate video, for instance, then they are not being used detailed information on her enemies for targeting purposes but in a morally justifiable way. The example described above, obscures information on the gruesome impact of those actions. where a UAV helps target and eliminate the threat posed by Increasingly we live in a cloud of information created by other an enemy sniper in a combat situation, is a case of the moral humans and our smart artifacts. These technologies can color, use of UAVs. or spin, the information coming into a specific user and this However, using them in the assassinations of suspected is a phenomenon we ignore at our peril. Next, we will look at terrorist leaders such as Abu Ali al-Harithi and Mohammed Atef some specific ways distancing might alter our perceptions of represents a gray area. We have the problem in both cases of moral situations. significant double effect. In the Mohammed Atef affair, one hundred others, including civilians, were killed in the raid. It is Shift Work Soldiers also difficult to determine the status of insurgents such as Atef. The extreme distancing afforded by some telerobots is leading Was he the same as a military commander or a dangerous to the creation of shift work soldiers. Pilots who live and work criminal? If the FBI cornered a killer in a building it would not in the Western United States generally fly the larger UAVs such be moral for them to call in an air strike to kill him because of as the Predator. They commute to work each day, fight the War the likelihood of killing many civilians in the process. Is this on Terror in three-hour shifts, and return home to their everyday action moral only because the collateral damage were Iraqis lives (Knapp 2005). Right now, it is neither technically feasible or was it okay because they were in close proximity to a known nor desirable to set up a similar situation for ground support terrorist? This situation is far from clear. The actions of criminals and combat robots, but there is no reason why this technology like Mohammed Atef and Abu Ali al-Harithi are unjustifiable in could not develop in the future. Other future innovations that the just war tradition as acts of terrorism of the sort we see in are being worked on now will allow pilots to fly up to four the modern world are the conscious breaking of all military UAVs at once, greatly reducing the number of pilots needed for — 22 — — Philosophy and Computers — these operations (Kelly 2005). With innovations such as these, of continued warfare and decrease the chance of peace, thus might we one day see large portions of our military consist of conflicting with the concept of a just war. It is important that telerobotically operated fighting machines controlled by pilots when we design telerobotic weapons systems we mitigate this and soldiers who never have to deploy far outside of their home possible outcome by making sure that the pilots are military country? This would greatly reduce the casualty rate amongst professionals who have the full comprehension of what they the forces operating these machines and would not require are doing. One final issue is that this normalization process much sacrifice even in the lifestyle of the pilots and soldiers implicates nearby civilian centers and the families of the pilots themselves since they could live in close proximity to their in the military actions they are propagating. Does this make friends and family and live a somewhat normal life while at them legitimate military targets? If this is indeed the case, then the same time fighting the enemies of their country. This would the military is putting people in jeopardy of retaliation in a greatly reduce the political liability warfare represents now and morally unjustifiable way. It is important that the international certainly would have a positive effect on the morale of these community address this troublesome issue, hopefully working future soldiers. Only someone who is not a soldier would find to end this practice. this next question compelling, but if there was little cost in the lives of its soldiers, I wonder, what would be the incentive not All the World’s a Battlefield to go to war on a regular basis, as long as one’s opponents As the Yemini case we started our discussion with above were significantly less powerful and could not reciprocate in illustrates, the war we find ourselves in now has an ill-defined this kind of warfare? This evolution in warfare could very well combat zone. Small hostile actions can break out anywhere result in a kind of nominalization of this activity so that it is not on the globe at any time. Prosecuting our enemies requires much different from any other job. us to find them wherever they are operating. This has the UAV pilots do experience a lot of stress just like combat unfortunate consequence of turning the entire world into a pilots, and it can be difficult for them to deal with being mentally potential battlefield. The UAVs we have been talking about are in combat one minute, then at home the next. CBS reporter extremely useful in these situations since they have a large Robert Riggs, who was following this story, asked a Predator range and can loiter unobserved for long periods of time high pilot about this phenomenon and he wrote: above just about any location on earth. As we saw in the Yemini case, lethal force can be delivered to these locations as well. When I asked if Captain Platt had witnessed ground The temptation to use that force is high. For instance, it would troops suffering casualties through her Predator have been difficult to send in human agents to arrest Abu Ali viewfinder, she took an emotional pause, “like any al-Harithi. Instead of just letting him and his compatriots go to soldier you press on with the mission and you get the potentially cause future mayhem, CIA operatives instead chose work done. And you don’t let those things get to you.” to assassinate him and the others in his vehicle. This seems (Riggs 2004) like a highly rational choice, but we may find it expedient to At this time the vast majority of UAV pilots are military use extreme force from high above when it is more risky to professionals, most of whom have much experience flying use ground-based agents. This could increase the chances conventional aircraft. This is slowly changing as the number for casualties amongst non-combatants. This is hard to justify of UAVs rapidly grows. We are likely to see more and more from an ethical standpoint. Not to mention the tricky political pilots of UAVs who fly only these aircraft. Also, the intelligence implications of prosecuting our enemies with military force in community flies a certain number of missions and the civilian countries we are technically at peace with. For these reasons, or military status of their pilots is hard to determine. Just as cases such as the Yemeni affair have made politicians in Europe private security agencies are used as force multipliers by the U.S. nervous and are far from unambiguously acts of just war. military in Iraq, so too might we expect private security firms The Illusion of Autonomous Warfare to provide pilots and operators for future telerobotic military Distancing in telerobotic warfare might give critics pause in support and weapons platforms. This eventuality stands to that they seem too much like deadly video games. If we are greatly complicate our ability to propagate just wars. to use these weapons in a just manner, silencing this criticism We have to be careful if these innovations lead to a is incumbent on those who would use these telerobots. normalizing of warfare, if we want to remain principled and Telerobotic weapons systems place a tremendous antiseptic just in our military actions. In the just war theory warfare layer of technology between the combatants that may help each inhabits a special place and diverging from that position quickly side to dehumanize the other. The operator of the machine will nullifies any claim to morality a combatant might claim. Just see his or her enemy as little more than thermal images on a war theory attempts to outline the ethical conduct of war based screen, and the human combatant will see only the animated on the maintenance of certain preconditions that prescribe the mechanical weapons of his or her foe. This type of warfare could morality of any given war or action committed in the progress ferment an even deeper hatred than that already fostered by of a war. One of these preconditions is the requirement that current modes of armed conflict. the use of force be limited to actions that aim at concluding the This eventuality would run counter to just war theory conflict and regaining a peace that is more stable than what because it might lead to the excessive use of violence and it existed prior to the conflict (Walzer 2006, 120-26). The only limits the combatants’ ability to see their opponents as moral reason to prolong a war is if this goal cannot yet be achieved. equals. The moral equality of soldiers is a key component of If the conduct of warfare becomes equivalent to a day at the Michael Walzer’s theory on just war (Walzer 2007, 34). This is a office, it is conceivable that we might lose motivation towards difficult concept, but if one is to fight a just war, it is permitted its speedy conclusion. Given that only large technologically to use lethal force on an enemy in order to secure a better advanced countries such as the United States can produce lasting peace; however, this must be done without losing sight and maintain telerobots, and that the majority of opponents that one’s enemy is still to be treated as a moral equal. Since it to countries such as the United States do not have telerobots is difficult for each individual soldier to properly work though themselves, then we are likely to see a continual rise in the all the nuances of the conflict he or she might be involved in, blood shed by our opponents with a welcome decrease in they do not lose their status as moral agents simply by being that shed by ourselves. This can tend to make war more of a on the wrong side of a conflict. This especially applies to lower palatable political expedience, which will increase the chance

— 23 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — ranking soldiers who have found themselves fighting in wars we also work to mitigate the many difficulties that the distancing due to conscription or other social forces well beyond their from reality that telerobotic weapons systems cause. control. Since we know that it is a real possibility that soldiers We also saw that using this technology in an ethical manner operating telerobotic weapons, both now and in the near future, will require us to be comfortable that the military objectives are only telepresent to the situations on their screens, we have and rules of engagement we are implementing are ethical. to make sure that proper rules of engagement are enforced to Assuming that we are in agreement on that point, then, next prevent potential abuse. we need to know the probable consequences of the use of the The Ethical Use of Telerobotic Weapons technology and then use them intelligently to advance our goals and mitigate negative side effects. Finally, I would like to end on a hopeful note. Along with the potential pitfalls of the use of these weapons, there is also the At the stage we are in with the technologies of telerobotic great possibility that they can be used in a just and principled warfare the ascription of moral agency is not too difficult given manner. Evidence of this can already be seen in the use of that there are human operators who tend to still be in control UAVs. In an incident in 2006 a Predator tracked a group of of the morally charged questions of the use of deadly force. Taliban fighters in Afghanistan who were massing around a This situation is rapidly changing as more and more of these vehicle. After observing the fighters for a time they could see decisions are becoming automated or at least influenced by that roughly two hundred fighters and their commanders were automated systems. This makes the situation much more lined up in seven rows as if on parade, an easy target waiting to difficult to analyze. For work that addresses these eventualities, be eliminated. It was believed that they were at a funeral burying see Arkin 2007; Sullins 2005, 2006, 2008; and Wallach and Allen their casualties suffered in fighting earlier that day. Commanding 2009. officers of the Predator pilots decided that attacking this group Endnotes would go beyond the rules of engagement that the U.S. military 1. As of the writing of this paper, the SWORDS weapons system is holding itself to in this conflict and the group was not attacked has been deployed by the U.S. military and then removed while it was in the cemetery. Explaining this action in a news from use in the Iraqi conflict. statement, military officials said, “Coalition Forces do not 2. See http://www.airforce-technology.com/projects/predator/. discuss rules of engagement; however, they hold themselves 3. See http://www.af.mil/news/story.asp?id=123050533. to a higher moral and ethical standard than their enemies,” 4. Vecna Robotics, http://www.vecnarobotics.com/robotics/ and even though it frustrated other military personnel, since index.shtml. the military actions in Afghanistan are not seen as “all out war” 5. iRobot Corporation, http://www.irobot.com/index.cfm. the decision not to fire has been properly seen as the ethically 6. Ibid. correct thing to do (Miklaszewski 2006). 7. Ibid. This technology does allow for closer scrutiny of combat 8. Foster-Miller, http://www.foster-miller.com/lemming.htm. actions by commanding officers and also allows for other 9. Ibid. parties not caught up in the heat of combat to voice an opinion 10. There are significant privacy and prudential constraints that in a timely manner. This kind of oversight could work to limit must be followed in the use of these technologies but they the possibilities of excessive force. What must be avoided are are beyond the scope of this particular paper. situations where this oversight does not exist, or the use of these weapons by forces that may not be operating under any sort of References rules of engagement. A F News. 2007. Predator strikes insurgents, Airforce Link. AF.mil. We must realize, though, that technologies such as Arkin, Ronald C. 2007. Governing lethal behavior: Embedding ethics in telerobotic weapon systems can also serve to decentralize a hybrid deliberative/reactive robot architecture, Technical Report GIT- traditional command and control (Singer 2009, 348). This GVU-07-11, Mobile Robot Laboratory, College of Computing, Georgia Institute of Technology. http://www.cc.gatech.edu/ai/robot-lab/online- distributed command and control can also serve to distribute publications/formalizationv35.pdf. moral and ethical decisions in such a way that no one has to BBS. 2003. First Predator strike takes out anti-air threat. BBSNews.net. take final responsibility or that removes all officers so far from the military action that they have no way to properly make these Defense Update. 2007. Smart weapons for UAVs. http://www.defense- update.com/features/du-1-07/feature_armedUAVs.htm. important decisions responsibly. Kaplan, David. 2003. The inside story of how U.S. terrorist hunters Right now a telerobotic reconnaissance craft such as the are going after al Qaeda. U.S. News and World Report. http://www. Predator costs tens of millions of dollars to buy and operate, but usnews.com. as the technology becomes more ubiquitous the costs will fall, Kelly, Mary Louise. 2005. The Nevada home of the Predator aircraft. and soon it may be in the price range that news organizations NPR, All Things Considered. NPR.org. may be able to afford. In this scenario, the same technology Knapp, George. 2005. Predator UAV ‘battle lab’ just north of Las Vegas. used for military reconnaissance can, and should, be used by http://www.klas-tv.com/Global/story.asp?S=3001647CBS. the media to extend the abilities of war reporters. If this were Miklaszewski, Jim. 2006. U.S. passes up chance to strike Taliban. MSNBC accomplished, then it may greatly enhance the accountability News. msnbc.msn. and scrutiny of military affairs that reporters can provide.10 Mitcham, Carl. 1994. Thinking Through Technology: The Path between Conclusion Engineering and Philosophy. Chicago: University of Chicago Press. We have seen that telerobotic weapons systems need to be Pincus, Walter. 2002. U.S. missiles kill al Qaeda suspects. The Age. carefully designed and deployed in order to minimize the theage.com. http://www.theage.com.au/articles/2002/11/05/1036308 311314.html. negative impacts they could have on the already shaky ground that supports the ethical conduct of military actions. But we have Riggs, Robert. 2004. Predator TV: Eye in the sky protects soldiers in Iraq also seen that these systems do allow for increased command and Afghanistan. CBS News. and control with the attendant potential for increasing individual Singer, P. W. 2009. Wired for War. New York: Penguin Press HC. accountability and review when, indeed, the cause for conflict Sullins, John. 2005. Ethics and artificial life: From modeling to moral is just and necessary. So as long at this technology is knowingly agents. Ethics and Information Technology 7: 139-48. utilized in a just military action, it can be used in an ethical way if

— 24 — — Philosophy and Computers —

Sullins, John. 2006. When is a robot a moral agent? International Review own UN plane with a faulty IFF transmitter and, as a result of of Information Ethics. Vol. 6. http://www.i-r-i-e.net/inhalt/006/006_ the visual sighting, the aircraft was escorted and landed. Sullins.pdf. Being present during both incidents, I have often wondered Sullins, John. 2008. Artificial moral agency in technoethics. In Handbook what could have happened had those systems been set to “run of Research on Technoethics, edited by Rocci Luppicini and Rebecca on full auto.” Here, we have two unique events, during two Adell. Idea Group, Inc. different conflicts, with two different weapon systems capable Wallach, Wendell and Allen, Allen. 2009. Moral Machines: Teaching of operating automatically. Either system, once set by its human Robots Right from Wrong. Oxford: Oxford University Press. operator to operate on its own, could detect, track, load, and Walzer, Michael. 1977. Just and Unjust War: A Moral Argument with fire with minimal human interaction. Battle doctrine software Historical Illustrations, 4th ed. New York: Basic Books. was already loaded and, at the press of a button, would have Acknowledgements begun execution. I would like to thank the many people who commented on this paper Now, there was no need to have these systems set to and with whom I had helpful discussions at the Computer Ethics automatic. In fact, very specific circumstances and events must Philosophical Inquiry conference in 2007 and the North American take place before authorization would be granted for the change Computers and Philosophy conference in 2008. Also, I would like to in state. The military is quite happy with the overseeing of thank the anonymous reviewer for this Newsletter for helpful comments on the final draft. weapons by humans. This policy of manual oversight, I believe, played a major part in the fact that no lives of friendly forces were lost in these specific events. Human decision-making took Keep Autonomous Systems Under Manual precedence and, in taking certain possibilities into account, prevented the needless firing of a weapon either manually or Control (for now): A Veteran’s Reflection automatically. While my history is old, this view continues today. J. P. Sullins writes of an event where robots were not allowed to William L. Dyer attack a funeral procession. In 2006 a Predator tracked a group University of Illinois at Springfield of Taliban fighters in Afghanistan. Closer observation led U.S. commanders to believe that they were at a funeral burying Abstract casualties suffered in fighting earlier that day. They decided that I will argue that robotic weapons, while useful, should continue to attack this group would be outside the rules of engagement to be monitored and operated by humans. Depending on their and the funeral was allowed to go on to completion.2 designed function, such systems can reduce human casualties, Today, we have unmanned aerial vehicles (UAV) that can increase the air and land superiority, and can work in harsh fire weapons. SWORDS, a robotic vehicle weapon, is also on environments—basically giving a commander a means to the ground and in use. This weapon can be fired by a soldier in continue the fight—or humanitarian mission without the need a different location. Then we have the TALON family of bomb- for additional personnel. Ultimately, I foresee the use of fully sniffing robots, HAZMAT robots, SWAT/security robots, and autonomous systems on the battlefield, but until they can be the like.3 These systems can be controlled remotely, placing proven to act in an ethical manner, they should remain under the burden of proper ethical conduct on the operator of the human control.1 system under their control. Such machines are called Tactical Autonomous Combatants (TAC). TACs have the ability to work Introduction on the ground, in air, space, or under water, as well as in severe I spent several years in the Navy, much of it on patrol in the conditions such as extreme heat or cold and in chemical, Persian Gulf region. In April of 1988, I was assigned to a ship that biological, or radiological contaminated environments.4 was present during Operation Praying Mantis, a U.S. offensive In a 2003 interview for About.com, Gordon Johnson, the against the Iranians. Along with naval guns, air strikes played a Unmanned Effects Team leader for Project Alpha, had this to large part in the fight. Eventually, the planes needed to refuel say: and re-arm. As a returning sortie was returning to their carrier for resupply, they passed over my ship, which had its Close-In We call them tactical autonomous combatants Weapon System (CIWS) in a non-firing, but automatic tracking because they’ll operate largely autonomously with mode. Despite repeated requests for permission to place the some limited human supervision. We’re talking about, system in manual (by pilots and the gun’s crews), the on- where we can and where we have the capability watch Tactical Air Officer ordered the weapons to remain in of replacing humans. We’re not talking about the the automatic tracking state, according to the ship’s published operational level or strategic level, but at the tactical doctrine. The system dutifully tracked each plane as it flew level, still using humans where we need to. Using overhead. Needless to say, the pilots were more than a little adjustable autonomy or supervised autonomy, nervous at having Gatlin guns following their movements. humans will still have to interact with the machines Soon afterward, an order from higher echelons had the guns and help guide them.5 placed in a stowed position, where they remained for the rest A TAC’s purpose and operation may be better described of the conflict. this way: a bomb disposal robot and operator are assigned to Fast forward to Desert Shield (1990) and the Iraqi no-fly zone disable and destroy bombs in a partially destroyed building is in place. During one of my watches as a missile engagement which must be cleared in order for troops to advance safely. The console operator, unidentifiable IFF from an aircraft in the operator may be able to control the robot into the building, but air, over Iraq, was detected. The weapon system was set to problems with line of sight due to the damage to the building recommend actions, but loading of the missile launchers was may require that the operator allow the robot to navigate on set to manual. My orders to the pilot to land were ignored, but its own. Once the robot reaches an explosive, the operator can the ship’s launchers were not loaded. Conversations about the reassume control and guide the robot through the disarming track, between various combat stations aboard the ship, led to of the bomb. a request of a visual sighting by the Air Force, which launched two F-16s. It turned out that the plane I was tracking was our

— 25 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Types of Autonomy with the use of systems designed to compute the fire control Typically, a robot’s level of autonomy is fixed. This state, problem rapidly and continually as the situation changes. known as tele-operation, has the human operator as the sole Incoming missiles are generally traveling too fast for an optimal controller of the system. During an operation, however, there firing solution to be made by a human. are a number of elements that are likely to change, such as the In addition to speedier computations, robots, depending mission, environment, workload, and a host of other variables. on their purpose, are more mobile and survivable. They have Such changes can be so rapid that the human operator or robot increased reaction times and greater sensing capabilities. They may not be able to respond in a timely manner under a tele- are fearless, do not need to eat or sleep, and can function operative mode. in climates dangerous to humans. They can be designed in Various types of autonomy exist for certain systems. Many numerous ways, suited to their purpose and mission. For terms are used to describe these states, but I have chosen three example, some look like vehicles while others look like planes. that I think best summarize them. In addition to the fixed mode Some have no physical form at all as intelligent software 10 of operation, the most common are: supervised, adjustable, agents or cyberbots. Robots are continually improving. and full. Advancements in mobility, dexterity, longer lasting power supplies, miniaturization, weaponry, and artificial intelligence Supervised autonomy allows the human to remain in are continual and rapid. AI and power supplies pose some of control while using the system’s advanced computational the greatest challenges. properties. Here, once certain parameters and goals have been defined, the robot is allowed to execute some tasks, such as Yet there is a sense of distrust toward the fully autonomous perform initial evaluations, initiate actions, and implement weapon. Mechanical and electrical parts fail, causing some plans, without the micro-managing, step by step, human unpredictable (most are predictable) results. In the case of operator input. Supervision and communication are more the UN plane’s faulty IFF signal, I suspect that, had the ship’s constant in this state of autonomy as human operators provide weapons been set to automatic, it would have performed as assistance to the robot.6 it should have and taken down the plane. The system was designed to shoot down enemy targets and, not knowing the Adjustable autonomy is a state in which one agent chooses IFF signature, could have seen the plane as an enemy target. the appropriate entity, whether it is itself, the human controller, No other means of verifying the plane’s true identity (that I can or another agent, to make a decision at times when immediate recall) was possible and so the weapon would have acted action is necessary. It is situation dependent. This type of properly under its rules of combat. autonomy allows human operators to adjust the autonomy of the robot as well as allowing the robot to adjust its own A system that is intentionally set to operate and fails autonomy. This reduces human interaction with the robot, brings about a different problem. How do we punish the robot but interactivity is maintained. Human operators can interact that experienced a malfunction and causes damage to other at any time with the system at any level of control and most equipment and personnel? If it is proven that the operator was appropriate for any situation.7 not at fault, the robot is blamed and held causally responsible for the action and to repair the damaged subsystem. We tend Full autonomy allows the robot full control of its operations. to fault a specific failure in a subsystem rather than the entire A fully autonomous system requires minimal communication system accurately. Repairs are easy enough, but experience with its human operator, can evaluate its situation, develop and has shown me that there is often an additional opinion held by initiate actions, and return back to base upon completion of its others that, if the system was unreliable, all other such systems mission. Communications are reestablished when the robot are also unreliable. uploads its data once the mission has been completed. No outside processing or decision-making is needed; intelligence Human training and experience is another concern. During resides entirely on the robot itself.8 the UN plane incident, the missile system was in a manual state. Would I, as the operator, have been blamed for the downing Beneficial but Not Trusted of a friendly aircraft or the weapon? In this case, I would have Robots have numerous benefits—enough to cause their been blamed, and rightly so. Under a manually controlled continual development. They can process collected data system, the operator would bear some, or all, of the weight of much faster and more accurately than humans, reaction the situation’s results. Going back to the CIWS event, during the times are often much faster, and the fire control problem is offensive against the Iranians, the guns could have been placed solved much more quickly. To give an idea as to the speed of a in manual mode to allow for a safer passage for the returning computer in solving a fire control problem, the problem takes planes. It is a reminder that experience is also a concern. into account many variables: speed and direction of the robot, Many other illustrations could be presented here, but speed and direction of the target, location of itself (the robot) Russ Richards, Project Alpha’s director, effectively sums up and the target, during travel, the jagged movements—and the this way: angles of the robot and target—are also computed. If a new, more threatening target appears, the problem is adjusted. A The greatest hurdle is likely to be overcoming military fire control computer can solve and figure for all variables in culture. Just getting present-day decision makers mere seconds whereas a human will need one to three hours to allow robots to perform some functions that are (depending on the variables to be taken into account). A person currently being performed by humans will be difficult. with a calculator, pencil, and paper will not solve the problem What is interesting is that we are already doing this. For of an actual incoming missile. example, Patriot missile batteries, close-in-weapons systems, cruise missiles, and other “smart” weapons This is not to say that humans cannot successfully calculate are already pretty autonomous. a firing solution in a short time, nor is this to say that the need for well-trained crews is no longer necessary. A highly trained It will be difficult to overcome the resistance crew can use their weapons in a manual state and successfully to replacing human pilots, soldiers, sailors, and hit fast moving targets. In addition, history continually shows Marines with robots. Or, to allow machines to make that mankind has successfully used advanced weaponry and decisions. The case will have to be made based on tactics.9 The point here is that some incidents are better solved the imperatives.11

— 26 — — Philosophy and Computers —

The military does not like unexpected surprises and favors Conclusion structure and order. This distrust may dissipate as robots Robots are changing the way wars are being fought. Not only is improve in stability and prove themselves in the field. Radar warfare changing, it is changing in ways that can reduce human was widely distrusted over eyesight for several years after its casualties. Robots have to be programmed, however, and it is introduction. Today, it is a necessary set of eyes on planes, important that this be done using strict ethical guidelines. With ships, and ground units. the anxieties of commanders in the field and disagreements Ethics among experts, one is a little worried that robots can ever act ethically at all. Yet, Moor’s idea that AI applications of ethics Robot technology is constantly advancing toward greater may lead to valuable clarifications within ethical theory is capability and autonomy. At present many of the smaller robots encouraging. Additionally, once released, these devices should are geared toward surveillance, security, and bomb disposal be closely monitored. Sullins may have said this better: while larger autonomous weapons, such as CIWS and weapon direction systems aboard ships, are destructive devices. So, for the purposes of this paper I will take the The role of some of the smaller robots, such as aerial drones, stand that robotic weapons technologies are used is changing, however. Previously used for reconnaissance, they in an ethical manner in only those situations where have advanced to the point where they can carry munitions. the operators (and in some cases the machine In September of 2007, the U.S. Army’s MQ-5B/C Hunter armed itself); know what morally justifiable outcome they aerial drone made history as the Army’s first-time use of a drone are seeking by using lethal force, and what the aircraft to kill enemy fighters in Iraq by targeting and dropping a consequences of those actions are most likely to be, precision bomb on a team of insurgents implanting a roadside and that these actions do not contradict the execution 19 bomb near Qayyarah.12 of a principled and just war. Research and development in robotic systems is In this paper I try to show the importance of the ethical use progressing. The Army’s Future Combat Systems initiative13 is of autonomous systems in combat. Robots are already in use one such example; Foster-Miller, Inc., also plays a large part.14 and, while we may imagine them being constantly monitored As technological advances are proven stable, the demand for from a console, as a tele-operative system, many of them can more, greater technology increases. The use and deployment be given more autonomy at the press of a button. Many systems, of autonomous technology demands careful thought and such as CIWS and missile systems, are capable of detecting, consideration, especially if ethics and technology are to work tracking, and engaging targets with very little supervision. hand in hand.15 These are normally controlled in a manual mode, and should In a recent New York Times article,16 the writer, Cornelia continue to be operated in that state as much as possible. Dean, asks about ethical issues surrounding autonomous Robots, whether tele-operative or with advanced autonomous battlefield robots: Is it possible to have killing robots that states, should be used such that their actions do not perform behave ethically? Even better, will they act with greater morality acts counter to just war principles. than their human comrades? Can they be designed to act in Systems designed to operate on a higher level of autonomy such ways that they will never violate the principles of proper should be allowed greater freedom to perform their unique battlefield rules of engagement or the tenets of the Geneva tasks, but little more. I do not think that we have complete ethical Convention? programs yet, but work should continue. As robots advance, Dean also interviewed Colin Allen and Daniel C. Dennett, the need to build more and more ethical considerations into pointing out that the two men agree with one another in that them exists. they believe that the creation of ethical robots is possible and the From my days in the military, up through the present day, ethical issues concerning the use of them should be considered robots have not been allowed to go completely autonomous. now. Noel Sharkey, on the other hand, holds that robots will A human is still required to monitor and operate the device. not be able to act morally, but if they do achieve this, and until Perhaps robots will be allowed to run autonomously in the they have proven themselves to do so, autonomous systems future (I think that they will), but until ethical routines can be should be banned. coded and the robot can act accordingly, I do not see the military The article also touches on AI research. “My research allowing autonomy, as history so far shows. We’re testing the hypothesis is that intelligent robots can behave more ethically waters and progress is being made, but I like to think that we’ve in the battlefield than humans currently can,” said Ronald C. been so far wise enough to know that, until AI and ethics can Arkin, who is designing software for battlefield robots under catch up, fully autonomous, completely unsupervised robots contract with the Army. The Information Science Directorate will be slow to arrive. of the Army Research Office funds Arkin’s work. Its director, Endnotes Randy Zachery, was also interviewed, stating his hopes for the 1. In this paper, I will be using autonomous weapon, project’s goals in showing how soldiers might use and interact autonomous system, and robot interchangeably. with these systems and how software might be developed to 2. J. P. Sullins. “Telerobotic Weapons Systems and the Ethical 17 enable such a system to act within ethical bounds. Conduct of War.” Class handout (no date given). Admittedly, Arkin’s work is difficult at best and, to some, 3. TALON Family of Military, Tactical, EOD, MAARS, Hazmat, perhaps impossible. James Moor provides examples as to how SWAT and Dragon Runner Robots. http://www.foster-miller. this can be done. One approach to create ethical knowledge com/lemming.htm. Last Accessed: 30 December 2008. Foster- might be accomplished through the use of good old-fashioned Miller, Inc., is also the producer of SWORDS. AI. For the computer to produce sound ethical judgments, it is 4. About.com. “Military Robots of the Future (from U.S. Joint programmed with a large script that selects the information Forces Command, for About.com).” http://usmilitary.about. appropriate to making ethical decisions and then processes the com/cs/weapons/a/robots.htm. Last Accessed: 10 December data fittingly to produce an ethical course of action. Neural nets 2008. or evolving genetic algorithms are other possibilities in which 5. While I generally do not like to use about.com as a source, I ethics can be learned and acted upon by a machine.18 had long been aware of Project Alpha and its goals. The U.S. Joint Forces Command site (https://jfcom.mil) has archived

— 27 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

the Project Alpha pages and taken them offline, but they can be ordered. Autonomous Robots* 6. Odest Chadwicke Jenkins, Monica N. Nicolescu, Maja J. Mataric. “Autonomy and Supervision for Robot Skills and Don Berkich Tasks Learned from Demonstration.” http://www.cs.brown. Texas A&M University–Corpus Christi edu/~cjenkins/papers/aaai-sc-camera.pdf. Last Accessed: 27 December 2008. 7. Meghann Lomas de Brun, Vera Zaychik Moffitt, Jerry L. Abstract Franke, et al. “Mixed-Initiative Adjustable Autonomy for “Autonomy” enjoys much wider application in robotics than Human/Unmanned System Teaming.” http://www.atl. philosophy. Roboticists dub virtually any robot not directly external.lmco.com/papers/1595.pdf. Last Accessed: 27 controlled by a human agent “autonomous” regardless of the December 2008. extent of its behavioral repertoire or the complexity of its control 8. Idaho National Laboratory. “Dynamic Autonomy: Control mechanisms. Yet it can be argued that there is an important Architecture.” http://www.inl.gov/adaptiverobotics/ difference between autonomy as not-directly-controlled and dynamicautonomy/controlarchitecture.shtml. Last Accessed: autonomy as self-control. 30 December 2008. With all due respect to the enormous achievement Mars 9. To name just a few: artillery and flame throwers from the ancient times, a successful anti-submarine attack in the rovers Spirit and Opportunity represent, in this paper I argue Revolutionary War, a successful sinking of a surface ship by that the roboticists’ conception of autonomy is over-broad in the a submarine in the Civil War, and successful anti-air attacks sense that it is too easily achieved, uninteresting, and ultimately from World War I to the present day—a good many of these not especially useful. I then propose a theory of autonomous without the use of a computer or robot. agency that capitalizes on the roboticists’ insights while 10. About.com. http://usmilitary.about.com/cs/weapons/a/ capturing the philosophical conception of autonomy as self-rule. robots_2.htm (page 2). Last Accessed: 10 December 2008. I close by describing some of the capacities autonomous agency 11. Ibid. so construed presuppose and explaining why those capacities 12. “Drone Chalks Up First Iraq Kill.” http://www.defensetech. make it a serious challenge to achieve while arguing that it is org/archives/003714.html. Last Accessed: 12 December nonetheless an important goal. 2008. Introduction 13. U.S. Army. Future Combat Systems Website. https://www.fcs. army.mil/. Last Accessed: 10 December 2008. In a 1986 manuscript Rod Brooks set the path for robotics 14. Foster-Miller, Inc., website: http://www.foster-miller.com/. development by announcing that he wished “to build completely Last Accessed: 30 December 2008. autonomous mobile agents that co-exist in the world with humans, and are seen by those humans as intelligent beings 15. The Army’s Future Combat Systems website introduces itself in this way: “The Future Combat Systems (FCS) is in their own right” (1999, 86). Brooks astutely side-stepped the cornerstone of Army Modernization. FCS is the Army’s difficult questions about intelligence and intentionality with promise to provide Soldiers the best equipment and the decidedly dennettian phrase, “are seen by,” yet showed not technology available as soon as practical. FCS is not just a the slightest hesitation in declaring the goal to be autonomous technology development program-it is the development of agency (Dennett 1987). To be sure, the field of situated robotics new Brigade Combat Teams-these new brigades, with more Brooks arguably initiated has achieved far more than the infantry, better equipment, unmatched situational awareness commercial success of the floor cleaning Roomba: Mars rovers and communications allowing complete domination in Spirit and Opportunity are particularly spectacular examples. asymmetric ground warfare while allowing the Army to build Nor should Brooks, writing what amounts to more of an a force that can sustain itself in remote areas.” advertisement than a closely argued essay which explicitly 16. Cornelia Dean. “A Soldier, Taking Orders From Its Ethical shrugs off philosophy—he admits to having “no particular Judgment Center.” New York Times. http://www.nytimes. interest in the philosophical implications..., although clearly com/2008/11/25/science/25robots.html?_r=2&ref=science. there will be significant implications” (1999, 86)—be held to Last Accessed: 12 December 2008. the exacting standards of philosophical analysis. 17. Georgia Institute of Technology’s Mobile Robot Laboratory professor Ronald Arkin, under Army contract, is attempting Philosophical implications notwithstanding, perhaps to define algorithms for ethical behavior in machines that can Brooks ought to be interested in his philosophical assumptions. follow norms such as the Geneva Convention. A thorough His sweeping assumption that autonomous machine agency is paper: Technical Report GIT-GVU-07-11: “Governing Lethal possible strikes one as desperately needing justification in light Behavior: Embedding Ethics in a Hybrid Deliberative/Reactive of the intuitive rub, if not outright conflict, between mechanism Robot Architecture” can be found at: http://www.cc.gatech. on the one hand and autonomy on the other. It is one thing to edu/ai/robot-lab/online-publications/formalizationv35.pdf. suppose that a machine can act—a questionable assumption Powerpoint slides can also be located here: http://www.agi- in its own right—quite another to assume that a machine, the 08.org/slides/arkin.ppt. paradigm of causal pawns, can act of its own accord. 18. James H. Moor. “Taking the Intentional Stance Toward Robot Ethics.” APA Newsletter on Philosophy and Computers In this paper I examine the capacities Brooks’s mobile 06:2 (Spring 2007). http://www.apaonline.org/documents/ agents would have to enjoy to be completely autonomous in any publications/v06n2_Computers.pdf. interesting sense of the phrase. I begin by contrasting the use 19. Sullins. of “completely autonomous” in computer science circles with its philosophical use and argue that its use in computer science is by comparison over-broad and uninteresting. I then explain how one might go about analyzing “completely autonomous” in such a way as to capture its philosophical use. I leave as an open question whether any agent, human or robot, could be autonomous in this sense, but in closing I explain why roboticists should care.

— 28 — — Philosophy and Computers —

Two Conceptions of Autonomy the behavior of the system it describes is translation invariant Philosophers have no monopoly on the term “autonomous,” yet over time and in that sense independent. computer scientists—who more frequently come to computer Autonomy-as-independence contrasts sharply with the science from mathematics than philosophy—appear to have a more etymologically consistent philosophical conception of much broader conception of autonomy. To help clarify his goal autonomy-as-self-rule. As Dworkin puts it, of achieving completely autonomous mobile agents, Brooks What I believe is the central idea that underlies the goes on to explain that his robots, which he calls “creatures,” concept of autonomy is indicated by the etymology would be adaptive, robust, opportunistic, and active. Thus of the term: autos (self) and nomos (rule or law). quoting, The term was first applied to the Greek city state. • A Creature must cope appropriately and in a timely A city had autonomia when its citizens made their fashion with changes in its dynamic environment. own laws, as opposed to being under the control • A Creature should be robust with respect to its of some conquering power. There is then a natural environment; minor changes in the properties of extension to persons as being autonomous when their the world should not lead to total collapse of the decisions and actions are their own; when they are Creature’s behavior; rather one should expect only a self-determining. (1988, 12-13) gradual change in capabilities of the Creature as the To be sure, “self-determining” needs unpacking and, since our environment changes more and more. interest is in completely autonomous robots, the unpacking • A Creature should be able to maintain multiple goals must be computationally tractable. Suffice it to say at this point and, depending on the circumstances it finds itself in, that many more things are autonomous in the autonomy-as- change which particular goals it is actively pursuing; independence sense than the autonomy-as-self-rule sense. thus it can both adapt to surroundings and capitalize Bacteria, bump-and-turn toy cars, and Roombas are equally on fortuitous circumstances. independent of the direct control of some other agent. A • A Creature should do something in the world; it should suitably equipped bacterium is capable of independent motion, have some purpose in being. (1999, 86) yet it is completely at the mercy of the stimulus and response mechanisms for which it has been selected by evolution. It Two points bear emphasis. First, situated roboticists seek cannot but follow a chemical gradient, for example. Similarly, to create AI on a par with insects by constructing mobile robots the Roomba can navigate and optimize floor-cleaning patterns, with sensors for immediate feedback about the results of their but in doing so it merely follows a somewhat more general motions. The idea is to construct robots that behave in such a set of rules than those implemented in the construction of the way that they are able to negotiate their environment; the goal in bump-and-turn toy car. situated robotics is not the intelligence of the situated robot per se but its ability to succeed in its environment. The successful The point is that a mobile agent can be independent of any situated robot adapts to its environment by “learning” strategies direct control, yet be at the complete mercy of internal control for accomplishing tasks. It thereby uses the environment, not its mechanisms. Put simply, it is not enough that a robot be able to program, as its source of information about the environment. react in an appropriate way to a stimulus, since the reaction and, Presumably, the more adaptive and successful robot will also presumably, its appropriateness may simply be an artifact of its be the more intelligent robot. design and programming (even with the addition of learning algorithms or bayesian decision algorithms). A narrower and, Second, the situated robot is not pre-programmed with it seems to me, much more interesting sense of autonomy detailed instructions for completing its tasks. Unlike industrial is one in which the agent’s internal control mechanisms are roboticists, who spend nearly as much time specifying the themselves subject to the agent’s control, if only because of the environment in which their robots will operate as designing vastly enhanced adaptive control such an agent presumably the robots themselves, the situated roboticist in all likelihood exercises. has no way of knowing in advance what the situated robot will encounter and do in a given environment apart, perhaps, from Autonomy-as-Self-Rule global directives like “move from position P to position P*.” The It may be, of course, that complete autonomy-as-self-rule roboticist has no way of knowing exactly how the situated robot is simply not achievable by any organism. Perhaps human will end up solving the problem of a rock placed in its path, for agents at best approximate autonomy-as-self-rule. Precisely instance. An excellent example of this was the successful use of what capacities would an agent have to have to be completely situated robotics techniques in the Mars mini-rover, Sojourner autonomous in the self-rule sense? Rover (SR). Because of a substantial delay in communication time between earth control and SR, global objectives like “move Consider first how actions might be distinguished from to rock C” could be given to SR while the problem of just how mere occurrences. The simplest approach is a recursive to get to C around intervening rocks A and B was left up to SR specification of actions, as follows: itself. Mission control could not always anticipate the path SR Let S be an organism, let ‘u’, ‘v’, and ‘w’ be replaced would end up taking. uniformly by verbs, and let t, t*, and t^ be (not necessarily A robot, in short, is completely autonomous provided it distinct) times. is not under the direct control of a human operator. Thus, a The Base Axiom bomb-disposal Remotely Operated Vehicle is not autonomous Ax1: S’s deliberately u-ing at t is an action. inasmuch as it is tethered by joystick to a human operator, while the SR and the Roomba presumably are autonomous because The Induction Axiom they do what they do without direct human intervention. Call Ax2: If S’s u-ing at t is an action, then S’s v-ing at t* is an autonomy of the not-directly-controlled variety “autonomy- action if either as-independence.” Perhaps unsurprisingly, autonomy-as- S v’s at t* by u-ing at t independence is consistent with the mathematical notion of or autonomy, whereby a differential equation is autonomous if S u’s at t by v-ing at t*.

— 29 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

The Closure Axiom in short. Deliberately raising my arm, however, presupposes Ax3: Nothing else is an action. a capacity to form arm-raising relevant beliefs and desires. What distinguishes beliefs from desires, however, is not just So, what distinguishes an action from a mere occurrence, on the difference in attitude they represent but their admissible this account, is its having been sourced in deliberation. Although representational content. To wit, no desire is ever directed at deliberation begs many questions, an agent is distinct from a an actual state of affairs. Whereas beliefs can be about actual mere causal mechanism in having the capacity to shape its and possible states of affairs, desires are exclusively about contributions to its causal environment. In giving an account of counterfactual states of affairs. Counterfactual representations autonomy-as-self-rule which bears on how the agent shapes its can also, however, include as part of their content the very causal contributions, it will be useful to further stipulate beliefs and desires which rationalize the agent’s actions. So, The Transitivity Axiom just as an agent has the capacity to deliberately raise her arm Ax4: If S u-ed at t by v-ing at t* and v-ed at t* by w-ing at or not, the autonomous agent has the capacity to deliberately t^, then S u-ed at t by w-ing at t^. form, or not, arm-raising relevant beliefs and desires. The agent’s capacity to reflectively or recursively deliberate upon The By-History Definition any of the agent’s own propositional attitudes provides at least Df1: A by-history of S’s action a = any ordered n-tuple n df one way to understand complete autonomous agency in the of S’s actions a ,..., a such that, for each a , a , 1 n 1 n j k interesting sense of autonomy-as-self-control. S a ’s by a -ing. k j In sum, a mobile agent is completely autonomous if The Pathway Definition its actions are n-order autonomous, where the n-th order

Df2: A pathway =df any segment of a by-history of an exhausts all the agent’s action relevant beliefs and desires. The action. autonomous agent is something like a deliberative Neurath’s Boat, constantly evaluating and revising beliefs and desires in Autonomy-as-self-rule presupposes that the agent exercise light of other deliberatively evaluated beliefs and desires. control over the ways in which it shapes its causal contributions. Thus, The capacity to form counterfactual representations of possible states-of-affairs, which can include the agent’s own The Axiom of Autonomy states-of-affairs, together with the capacity to form belief, Ax5: S autonomously v’s at t* by u-ing at t iff desire, and perhaps other attitudes towards representations a) S deliberately desires v-ing at t*, so-formed, are minimally necessary for autonomous agency on this view. Setting aside the (admittedly huge) problem b) S deliberately believes that v-ing at t* is probable to of attitude formation and robot attitudes for another time, some non-zero degree d of u-ing at t via a pathway p, counterfactual representation presents its own special and problems. At the very least we need a computationally tractable c) if not (a) or not (b), then S would not have u’ed at t. account of counterfactual representation. One possibility (cf. On this view, Cummins 1996; Swoyer 1991) is to account for counterfactual representation in terms of embedded isomorphic relational [A]utonomy is conceived of as a second-order capacity structures, but using the theory for counterfactual representation of persons to reflect critically upon their first-order raises several crucial questions: preferences, desires, wishes, and so forth and the • Complexity: How do we optimize embeddings so as capacity to accept or change these in light of higher- to minimize computational overhead? order preferences and values. By exercising such a capacity, persons define their nature, give meaning • Formability: If possible states of affairs are represented and coherence to their lives, and take responsibility for by transformations on embedded isomorphisms, the kind of person they are. (Dworkin 1988, 20) then precisely how and by what factors are those transformations constrained for relevancy? Complete autonomy will not, however, have been established • Attainability: What metric can be used to assess until the agent has the capacity to shape the deliberative basis the action-relevant, agent-relative realizability of by which it exercises control over its own deliberations. That counterfactual states of affairs represented by is, an agent cannot be said to be completely autonomous if it embedded isomorphisms—i.e., in Lewisian terms is at the mercy of immutable higher-order preferences and (Lewis 1973), how does the agent determine the values (and beliefs). If our goal is complete autonomy-as- self-rule, then this is at best first-order autonomy. Echoing closeness of possible worlds so represented? Frankfurt (1971), an agent enjoys first-order autonomy when it In light of these puzzles, perhaps no agent is ever completely has deliberative control over the deliberative basis it employs autonomous, but only approximates complete autonomy. in determining its course of action. Moving beyond Jackson, The Value of (Complete) Autonomy-as-Self-Rule we may say that the second-order autonomous agent has in addition deliberative control over the deliberations it brings So why should roboticists, who can achieve autonomy- to bear on the deliberative basis it employs in determining its as-independence with off-the-shelf technologies, seek course of action, and so on. even first-order autonomy-as-self-rule quite apart from complete autonomy-as-self-rule? Likewise, why should moral To be sure, deliberation bears a great deal of weight in psychologists fret over whether human actions are ultimately this account. Without becoming lost in particular deliberative rooted in complete autonomy-as-self-rule if the requirements strategies, all deliberations share in common the capacity to of complete autonomy make it, if not unattainable, then at contrast alternatives. The capacity to contrast alternatives in turn best exotic?1 presupposes the capacity to represent counterfactual states of affairs, and it is this capacity, I submit, which ultimately secures Consider again the industrial robot toiling away in its well- autonomous agency. That is, my raising my arm is not my arm’s defined factory environment. The distinction between this robot raising since it, my raising my arm, has a causal history involving and the Mars Rovers or even the humble Roomba lies in the my intention to raise my arm. I deliberately raised my arm, capacity of the latter to negotiate unpredictable environments.

— 30 — — Philosophy and Computers —

The behavior of the industrial robot is precisely scripted to Lewis, David. 1973. Counterfactuals. Cambridge, Mass.: Harvard within millimeter tolerances. Any change in the environment University Press. not anticipated by the industrial robot’s programmers at best Swoyer, C. 1991. Structural representation and surrogative reasoning. results in complete failure. At worst, it leads to the injury or Synthese 87: 449-508. death of human factory workers. The industrial robot is thus brittle with respect to its environment: it cannot suffer the least change in its environment without malfunction. Unlike the industrial robot, the designers of the Mars Rovers DISCUSSION PAPERS and the Roomba have only the most general notions of what to anticipate in their robots’ respective environments. Moreover, the problem of brittleness in robots is compounded by their own agency. Thus, Brooks saw quite clearly that robots by their Ontologically Tough Artifacts and Non- very nature make changes to their immediate environment, Spooky Intentions – Comments on Baker whether by moving from one location to another or by directly manipulating objects. This creates something of a feedback loop Roxanne Kurtz of environmental dynamism, which is perhaps why brittleness University of Illinois at Springfield in the smallest regard so quickly amplifies into catastrophe. In “The Shrinking Difference Between Artifacts and Natural Yet, the Mars Rovers and the Roomba are themselves brittle Objects,” Lynne Rudder Baker denies that artifacts are in comparison to a first-order autonomous robot. Neither the ontologically deficient compared to natural objects.1 I agree. Mars Rovers nor the Roomba has deliberative control over In this short comment, I offer further support for her thesis by its action-directed deliberations, however rudimentary those pursuing a point that she makes in passing: though we can action-directed deliberations may be. The Roomba cannot distinguish artifacts from natural objects based on their intention but return to its charging base, even if just how it navigates its dependence, the makers of artifacts are themselves part of way back is subject to measured responses to its environment. nature.2 I assume we are to conclude that intention-dependence Similarly, the Mars Rovers cannot but traverse from one gives us no good reason to believe that the products of our geological feature to another, even if how they negotiate intentional labors are ontologically weak compared to natural impediments is not determined in advance by mission control. objects. The first-order autonomous robot, on the other hand, has the capacity to assess, revise, reject, and even create new goals, Generally, I will be talking about ordinary material artifacts giving it an altogether higher order of flexibility and adaptability (arrowheads, computers, pencils, ranch houses) and ordinary over its more rudimentary predecessors. material natural objects (acorns, feathers, gold, rivers). I take it that these objects are the primary objects of concern in the If brittleness is the principle liability of agency, the utility debate at hand, rather than, for instance numbers, human of autonomy-as-self-rule lies in its inverse relationship with concepts, or animal souls. brittleness. That is, the greater the degree of autonomy-as-self- rule, the less brittle the agent insofar as the degree of the agent’s The participants in this debate share two basic assumptions. autonomy is key to its adaptability. First, artifacts necessarily depend on intentions for their existence, while natural objects do not. Second, natural objects Much the same is true of human agents. Our reflective, are ontologically robust.3 Taking these assumptions as givens, as opposed to reactionary, deliberative capacities have made the debate concerns whether artifacts are ontologically on a par us the technological savants of the world’s ecosystem. Few with natural objects. Let us put this in a bit of context. Suppose environments on or off the Earth have proved thoroughly we have an arrowhead and an acorn. The relevant question here inaccessible to our adaptations. Indeed, our interest in is: Is the arrowhead as ontologically robust as the acorn in that completely autonomous robots is itself a manifestation of our each partakes of full-fledged existence?4 Or is the arrowhead own creative autonomy. ontologically deficient in some way? Perhaps, for example, it Endnotes only subsists, or “exists” in some attenuated, derivative sense. * I presented material from this paper at the North American Or perhaps the arrowhead fails to exist altogether because only Computing and Philosophy 2007 Conference held at Loyola a chunk of flint rests where it is supposed to be. University–Chicago. I’m grateful to conference participants for Baker takes a clear position. She contends that despite their their puzzled looks and probing questions. I’m also indebted intention dependence, artifacts are as ontologically robust as to Vere Chappell, Lynne Rudder Baker, Bruce Aune, and Rod Grupen, who contributed much to the development of these natural objects. Arrowheads and acorns both enjoy full-fledged ideas. existence. But suppose that we oppose Baker. Suppose, like 1. I am grateful to Luciano Floridi for pointing out that the utility one sort of opponent that Baker considers, that we think that of autonomy-as-self-rule may not be entirely obvious. for an object to be ontologically robust, it must satisfy at least one criterion: Works Cited Intention Independence: An object is ontologically Brooks, R.A. 1999. Intelligence without representation. In Cambrian robust only if it could exist in a world that lacks beings Intelligence: The Early History of the New AI. Cambridge, Mass.: MIT 5,6 Press. A Bradford Book. with intentions. Cummins, R. 1996. Representations, Targets, and Attitudes. Cambridge, Would accepting this criterion allow us to consistently Mass.: MIT Press. A Bradford Book. maintain that natural objects are ontologically more robust Dennett, D. C. 1987. The Intentional Stance. Cambridge, Mass.: MIT than artifacts? Press. A Bradford Book. Perhaps at first blush—it gives us reason to regard artifacts Dworkin, G. 1988. The Theory and Practice of Autonomy. Cambridge: as ontologically deficient, given that artifacts necessarily depend Cambridge University Press. on intentions for their existence. Consider, for instance, Anasazi Frankfurt, H. G. 1971. Freedom of the will and the concept of a person. cave dwellings. These consist of connected cells constructed The Journal of Philosophy 68: 5-20. by the Anasazi people on the sides of cliffs in shallow caves from shaped sandstone, mud, and water.7 The existence of

— 31 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Anasazi dwellings necessarily depends on the intentions of and other such criteria.11 Thus, we should ask: What is the best the Anasazi people, who built them to provide shelter. If some place to draw the line between acceptable and non-acceptable random series of geological events blew sandstone, mud, and criteria to find an attractive and defensible position? water into shallow caves on the sides of cliffs, and the result was One move here is to bite the bullet and accept all the physically just like an Anasazi cave dwelling, it would not be an various criteria in all their ontologically debilitating glory (and Anasazi cave dwelling. Instead, it would be an interesting case of freely embrace most other such criteria that we might devise). accidental geological mimicry. Thus, Anasazi cave dwellings fail If we do so, then we may consistently hold that artifacts are to satisfy the Intention Independence criterion for ontological not ontologically deficient compared to most natural objects robustness. So far so good for Baker’s opponent.8 because it turns out that most natural objects are deficient as But on what grounds should we accept the Intention well. Someone who holds the view that only naturally occurring Independence criterion for ontological robustness, yet simples exist might find this strategy quite congenial. I do deny other candidate criteria? Consider several possible not. Nor, I think, would those who wanted to settle what the contenders: difference is (if any) between artifacts and natural objects in Instinct Independence: An object is ontologically robust terms of ontological robustness. only if it could exist in a world that lacks beings with Accepting a simples-only view requires rejecting our initial instincts. assumption that ordinary natural objects are ontologically If we accept this criterion, then we will have to downgrade robust, which seems to be non-controversial in the context of the ontological status of some natural objects, namely, those the debate. So, while acknowledging that a simples-only view objects such that their existence entails that beings driven by may well be a stable position, let us proceed hereon with our instincts created them. The homes of mud daubers come to initial assumption about ordinary natural objects intact. mind. These consist of connected cells constructed and shaped Given we want to press on with our attempt to defend by mud daubers from mud and water onto, for instance, the Baker’s opponent’s position, we still need a good reason to sides of buildings and into shallow recesses in trees. Sometimes maintain the robustness of natural objects while we demote the daubers carry the mud in from offsite; sometimes they mix artifacts to some lesser ontological status. At this point, to do it on-site. The mud daubers shape the mud with their mouths so, we need to explain what makes intentions special such and the vibrations of their wings.9 that we should embrace the first criterion above and reject The existence of mud dauber homes necessarily depends the others. on the fact that mud daubers instinctively built them to provide Being supernatural might be just the thing to mark intentions shelter. As with the Anasazi housing, if a random series of as special in that way. A dualist might be attracted to such a view. geological events (or intended events put in motion by a If instinctive behaviors, learning mechanisms, and biological careful scientist with a hankering to create mud dauber housing and geological processes are all natural, and intentions are models) created dauber-like homes, they would again merely not, this metaphysical distinctiveness would surely make for an mimic mud dauber homes. interesting difference between them. Moreover, it is a difference Learning Independence: An object is ontologically robust that could arguably explain ontological disparities. We might only if it could exist in a world that lacks beings that can say that objects that necessarily depend on the supernatural learn. are less connected to their natural existence than those that are wholly grounded in the material world. The disconnectedness Going along with this criterion would require us to might make them ontologically watered down in some way. Fair downgrade the ontological status of more natural objects, enough, if we are willing to go in for supernatural intentions. But given that the existence of some natural objects entails that one cautionary note—on this view, if God exists, the ontological beings learned to create them. Here, the example of woven status of everything would be hostage to her intentions. For on gorilla nests seems relevant. Woven nest making is a learned this view the ontological robustness of any object—including gorilla behavior. Is it necessary to the existence of a real gorilla me, you, a dog named Boo, and the land on which we travel nest that it be constructed by a gorilla who learned to make and live12—ultimately hinges on the intentions of its creator, such nests? Plausibly so. For, it is a distinctive feature of gorilla should it have one. nests that instinctive behavior alone does not account for their 10 But if we are of a more naturalist bent, if we agree that existence. intentions are simply another natural phenomenon of our world, Biological Independence: An object is ontologically what to do? Finally, this is where I see Baker’s passing comment robust only if it could exist in a world lacking biological coming into play and where I see Baker’s opponent hitting an processes. ontologically robust brick wall. This very demanding criterion would cast all biological As noted at the start, Baker briefly suggests that we entities (assuming they are essentially biological) and things should reject something like the Intention Independence that necessarily depend on organic matter for their existence criterion because we, the bearers of the intentions on which (such as fossil fuels and some kinds of diamonds and pearls) the existence of artifacts necessarily depends, are ourselves into the pile of ontologically deficient objects. simply part of nature.13 Geological Independence: An object is ontologically Now, I argue that if we are part of nature, then our intentions robust only if it could exist in a world without geological are natural, and the process by which intentions lead to the processes. creation of artifacts is natural, just like the process by which instincts lead to the creation of natural objects is natural. If this With this final suggested criterion for ontological is so, then it is difficult to see why intentions, but not the other robustness, we hit rock bottom (pun intended). Joining the natural processes picked out by the other criteria we considered world of ontologically weak objects will be all those objects that should detract from the ontological oomph of objects. necessarily depend on geological processes for their existence, such as mountains and pebbles, diamonds and gold. Recall that Baker’s opponent needs to draw the line below the Intention Independence criteria but above the We can easily see that our stock of ontologically robust other criteria. But such a move now looks arbitrary. It requires ordinary natural objects dwindles as we accept more of these

— 32 — — Philosophy and Computers — us to pick and choose among natural processes, all of which 11. For instance, we might embrace the following additional involve the creation of objects that necessarily depend on the criteria for ontological robustness, in no particular order: relevant process for their existence. But it is reasonable for us Species Independence, Consciousness Independence, to expect the ontological robustness criteria that invoke these Molecular Independence, Emotion Independence, Physics processes to stand and fall together in virtue of the metaphysical Independence, Art Independence, and so on. similarities of the involved processes. It should be an all or 12. Baker 2008, p. 4. nothing choice that we make here.14 Thus, I conclude that 13. Recalling, of course, Lobo’s fine song, “Me, You, and a Dog unless we are willing to downgrade almost all natural objects Named Boo,” which you may find on The Best Of Lobo, 2005, to an ontologically deficient status, then we should draw the and which you may sample for free at Amazon: http://www. amazon.com/gp/product/B00123G152/ref=dm_mu_dp_ line above the first criteria. We should reject all of the criteria trk1. for ontological robustness considered herein, including the criteria of Intention Independence. 14. Minimally, I contend that the burden of proof now falls on those who want to hold that intentions are natural, but have Rejecting all the criteria yields a stable position that remains ontologically undermining consequences, while other natural consistent with two of Baker’s significant commitments: processes do not, even though they too bear on existence the intention-dependence of artifacts and the ontological conditions for some objects. robustness of artifacts. It is a position that eliminates the original reason that we found while posing as Baker opponents to think that artifacts like Anasazi cliff dwellings are ontologically below How Many Explanatory Gaps Are There? par compared to the natural homes mindlessly made by instinct- driven mud daubers. So much the better for housing (a good E. Diaz-Leon result in this real estate market I guess). University of Manitoba

Endnotes According to many philosophers, there is an explanatory gap 1. Lynne Rudder Baker. “The Shrinking Difference Between between physical truths and phenomenal truths. Someone Artifacts and Natural Objects.” APA Newsletter on Philosophy could know all the physical truths about the world and, in and Computers 07 (Spring 2008): 2-5. particular, all the physical information about the brain and the 2. Baker (2008, p. 4) writes: “… it is tendentious to claim that neurophysiology of vision, and still not know what it is like to the existence of artifacts depends not on nature, but on us. see red (Jackson 1982; 1986). According to a similar example, Of course the existence of artifacts depends on us: but we are part of nature.” someone could know all the physical truths about bats and still not know what it is like to be a bat (Nagel 1974). We can conceive 3. Clearly there is room to offer coherent objections that deserve of an individual that is physically identical to me, molecule per responses to both assumptions, but I put such a task aside here. molecule, but does not have any phenomenally conscious state whatsoever (Chalmers 1996). Some philosophers have 4. Baker would say that ontologically robust (my term) objects are “genuine substances.” She explains (2008, p. 3), “what argued that the explanatory gap shows that we cannot explain I mean by saying that things of some kind (e.g., hammers, consciousness in physical terms (Levine 2001), or even that dogs, persons)—Fs in general—are genuine substances phenomenal consciousness is not physical and therefore is that any complete account of what there is will have to physicalism is false (Chalmers 1996; 2002). include reference to Fs.” Gilbert Harman has recently suggested (Harman 2007) that 5. For a very similar criterion, see the fifth condition Baker (2008, we can see the explanatory gap regarding consciousness as a p. 4) considers that might justify distinguishing between particular case of a more general phenomenon. He claims: “A artifacts and natural objects on the basis of ontological somewhat different explanation of the explanatory gap appeals status. to a difference...between two kinds of understanding, objective 6. I have not mentioned human intentions here. Surely the and subjective. Objective understanding is characteristic of the species of intender is irrelevant to how intentions bear on physical sciences. …These sciences aim to discover laws and ontological status. other regularities involving things and their parts, in this way 7. Ann G. Harris, Esther Tuttle, Sherwood D. Tuttle. Geology of achieving an understanding of phenomena ‘from the outside’” National Parks (Kendall Hunt, 2003), 95. (2007, 1). On the other hand, he adds, subjective understanding 8. Baker (2008, p.4) aims to cause trouble for this view, mainly is characteristic of the sciences of the mind broadly conceived, in the section prior to her conclusion. There, Baker devotes such as social and psychological sciences, which are concerned herself to making the Intention Independence criterion less with phenomena that cannot be completely explained within appealing to natural object ontological elitists. She does so by introducing artifacts that seem like natural objects and vice an entirely objective approach, and require an understanding versa, thereby blurring the line between artifacts and natural “from the inside.” For instance, he claims that “there are aspects objects. This is all well and good, but I wonder why Baker of reasons, purposes, feelings, thoughts, and experiences that does not target the Intention Independence criterion more can only be understood from within, via sympathy or empathy directly based on her view that humans are part of nature or other translation into one’s own experience” (2007, 2). too. Harman discusses several examples of this explanatory 9. Mike Arduser. “Missouri’s Mud-daubers.” Missouri gap between the objective and the subjective aspects of reality. Conservationist Online (August 2002). http://mdc.mo.gov/ In what follows, I will discuss these examples and examine conmag/2002/08/40.htm. Also, Jeffrey K. Barnes, “Organ- whether the explanatory gap between the physical and the pipe mud-daubers,” Arthropod Museum Notes 51 (July 31, phenomenal can really be seen as a phenomenon of the same 2007). sort, or whether it has any special features that require special 10. See, for instance, Nold Egenter, “The Nestbuilding Behaviour treatment. I will offer some motivation for this second option. of Higher Apes: Foundation for an Architectural Anthropology.” http://home.worldcom.ch/negenter/081NestbApes_E.html. The first example concerns intentionality. Harman argues Article originally appeared in Nr. 14 (Sep. 1990) in the that a purely objective approach to the behavior of a certain International Semiotic Spectrum - A Publication of the Toronto community could not tell us, say, what certain morning rituals Semiotic Circle as “Evolutionary Architecture: Nestbuilding that certain individuals are performing mean (Harman 2007, 2). Among Higher Apes.” Indeed, a purely behavioral description of those activities would — 33 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — not be enough for finding out what they are doing, that is, what crucially, this is something that a pure objective account sort of actions they are. There is some additional information cannot provide: this is where a subjective approach is needed, that is missing, namely, information regarding the mental states according to Harman. of the individuals that perform those movements. If we had Why is it the case that in order to know what seeing sufficient information concerning the content of their beliefs red is like, you must have had the experience of seeing red? and desires (and perhaps other mental states), we would be Because, we can assume, in order to know what seeing red in a position to know whether they are performing a religious is like, you need to be able to entertain the relevant thought. ritual, practicing a dance, or just doing some morning exercise. As Harman says: “Knowledge that P requires being able to However, I contend, it is widely accepted that a purely objective represent its being the case that P” (2007, 4). And in order to approach would be able to provide a complete account of the entertain propositions about what it is like to see red, you need content of their beliefs and desires. to possess the constituent concepts. One crucial concept is the To illustrate that point, we could follow Chalmers’s phenomenal concept “red experience,” which, arguably, you distinction between the psychological and the phenomenal can possess only if you have had that experience. aspects of the mind (1996, Ch. 1). The psychological aspect I agree that this model can explain the existence of an has to do with those properties of mental states that are explanatory gap in the case of experiences that we have not independent from consciousness, and which, plausibly, had ourselves or that we cannot imagine having. It is pretty clear can be explained in physical-functional terms, whereas the that a physical account of the world will not put us in a position phenomenal aspect of the mind has to do with what it is like to to know what certain organisms are experiencing, if we have undergo conscious mental states (and it is more controversial not had those experiences before, because in that case we whether those features can be explained in physical-functional will not possess the required concepts. However, I think that terms). Chalmers, among many others, considers intentionality even when we solve this problem, there is still an important to be a psychological aspect of the mental, since it seems very explanatory gap remaining. This further explanatory gap can plausible to think that we can explain intentionality in physical- be stated as follows: imagine someone who possesses all the functional terms. If this is so, then there are no good reasons relevant concepts, both physical and phenomenal. This person to think that a purely objective approach is not enough for could know all the physical truths about the world, and still not explaining actions. be in a position to know what experiences someone is having, Harman discusses a second example of the gap between say, that I am feeling pain at time t. This is not due to the fact objective and subjective understanding, namely, the problem that this subject does not possess the concept “pain,” because of providing a translation of the language of a certain group of by hypothesis she does. The problem is that she would not be people into one’s own language. In this case, he argues, it seems able to infer a priori the phenomenal truth that I am feeling pain that a purely objective account would not be enough either. If at t from a complete physical description of the world, since, we want a correct translation of a certain language into our own arguably, physical truths do not entail phenomenal truths a language, two additional requirements will be needed: priori (Chalmers 1996, Ch. 3). First, one must be able to identify a certain objectively In my view, we can distinguish two different explanatory described language as one’s own language, an gaps concerning consciousness, which require different identification that is itself not fully objective. Second, treatment. On the one hand, we have the (“easy”) explanatory one must have in one’s own language some expression gap, which manifests itself in examples such as Nagel’s bat that is used in something like the same way as the example or Jackson’s Mary thought-experiment: the idea here expression in the other language. (2007, 3) is that a complete physical description of the world does not necessarily put someone in a position to know what certain That is to say, if there is some expression E in the original experiences are like. This is due to the fact that we might lack language for which there is no expression E* with a similar the relevant phenomenal concepts. function in my own language, then I cannot provide a translation On the other hand, there is a remaining explanatory gap of E into my own language. which manifests itself in the fact that knowing a complete Using this analogy, Harman has offered a possible physical description of the world would not put us in a position solution to the problem of explaining the experiences of other to know what different experiences different organisms are creatures. Let’s see how it works: suppose that we could draw having, even if we possess phenomenal concepts to refer to a correspondence between the possible experiences of one all these experiences.1 The problem is that such a physical creature, say, a bat, and the possible experiences of another description of the world would not entail a priori a phenomenal creature, say, a human being (where this correspondence is description of the world. described in a purely objective manner, for instance, in terms of The explanatory gap in this second sense, that is, the fact physical-functional relations). In order for this objective account that a complete physical description of the world (P) would to provide an account of what it is like to be a bat that we can not entail a priori a phenomenal description of the world (Q) is understand, two additional criteria have to be satisfied: (i) we invoked by recent formulations of the conceivability argument have to identify the second system of experiences as our system against physicalism. For instance, Chalmers (2002) formulates of experiences, that is, as the class of possible experiences that the argument as follows: we can have; and (ii) for any possible experience of the bat, there must be an experience of ours that is functionally similar (1) It is conceivable that P&∼Q. to it (2007, 3). (2) If it is conceivable that P&∼Q, it is metaphysically According to this model, how could Jackson’s Mary know possible that P&∼Q. what it is like to see red? First of all, she would need an account (3) If it is metaphysically possible that P&∼Q, then that draws a correspondence between the experiences of materialism is false. normal perceivers when they see red, and her own possible (4) So, materialism is false. (2002, 249) experiences (in purely objective terms). In addition, she The first premise of this argument claims that P&∼Q is would need to have had an experience of seeing red herself. conceivable: by this, it is meant that we can conceive of P&∼Q Otherwise, she could not know what seeing red is like. And, without contradiction, that is, it is not a priori false. Or, in other

— 34 — — Philosophy and Computers — words, P→Q is not a priori true: this corresponds to the second characterization of the explanatory gap that I have sketched OOK EVIEWS above, and this is what is supposed to entail that P→Q is not B R metaphysically necessary and therefore materialism is false. The question now is: Could we use an explanation along Harman’s lines in order to solve the problem posed by the Internet and Society. Social Theory in the conceivability argument? I think that it is clear that we cannot, Information Age and the reason is the following: The problem posed by the explanatory gap in this second sense (that is, that P→Q is not a Christian Fuchs (New York: Routledge, 2007). priori true) cannot be solved merely by saying that we are not in Reviewed by Eveline Lubbers a position to know P→Q a priori because we fail to have some University of Strathclyde, Glasgow phenomenal concepts. Crucially, we are also unable to infer Q from P a priori even when we possess all the relevant concepts, The Internet is obviously here to stay. As self-evident as this both physical and phenomenal, (or so the first premise of the statement seems to be for our everyday life, in the academic conceivability argument says). Therefore, in order to answer the world the Internet has yet to find its place. This process of challenge posed by this argument, we need to either show that settlement is bound to be sped up with the appearance of in fact we are really able to infer Q from P a priori, or argue that Christian Fuchs’s seminal work Internet and Society. Social even if P→Q is not a priori true, that does not entail that it is not Theory in the Information Age. The main goal of the book is metaphysically necessary. And, to the best of my knowledge, to work out a theoretical understanding of the relationship of Harman’s model does not accomplish (and does not even aim Internet and society. The approach worked out in this book, as to accomplish) any of these two tasks. explained by Fuchs in his introduction, is critical in the sense What Harman’s model does provide is a very plausible that it focuses on social problems: “it identifies opportunities response to Jackson’s knowledge argument and Nagel’s bat and risks, sees them related to the larger social structure of argument: it can be argued that the reason that Mary could contemporary society, and understands them as antagonistic not learn what it is like to see red from her black-and-white forces.” The analysis of these forces, more specifically the room, or the reason that we cannot know what it is like to be opposition between cooperation and competition, runs as a a bat, is that we lack the relevant phenomenal concepts given thread throughout the book. Investigating the kind of society we that we haven’t had the corresponding experiences (and this is live in, Fuchs tries to tackle fundamental questions like “Which perfectly compatible with materialism). However, as we have role do networks and knowledge play in contemporary society?” seen, the conceivability argument sketched above poses a and “Which role do cooperation and competition have in the harder challenge for physicalism, which still has to be answered Internet economy, in online politics, and in cyberculture?” (The satisfactorily.2,3 fact that the spelling checker of my version of Word 2003 has Endnotes a problem with the term “cyberculture” can be regarded as an 1. Stoljar (2005) makes a similar distinction. indicator of the issues at stake here…) 2. For the record, I do think that conceivability arguments can The merits of Fuchs’s work lie in bringing together an be answered satisfactorily from the physicalist point of view. enormous amount of recent research, current discussions, A very persuasive response is the so-called “phenomenal and a broad diversity of definitions about the issues at stake concept strategy,” which argues that we can explain the fact here. Although the amount of information might be a bit that P→Q is not a priori, in terms of the special features of overwhelming for the average reader—the dense type area phenomenal concepts. Loar (1997) is an excellent exposition doesn’t really help here—the value of the book is unquestionable of this view. for those interested in the field. The proof of the pudding is in 3. I am grateful to Peter Boltuc for very useful comments. the eating. References Looking for a definition of information warfare, I was Chalmers, D. 1996. The Conscious Mind. Oxford: Oxford University startled by the fact that Fuchs’s starting point was the concept Press. of war, by Carl von Clausewitz’s definition. Then the relationship Chalmers, D. 2002. Consciousness and its place in nature. In Philosophy of war-technology-globalization is analyzed, and a theoretical of Mind: Classical and Contemporary Readings, edited by D. Chalmers. notion of information warfare is introduced. War in the eyes of 247-72. New York: Oxford University Press. von Clausewitz is not only “a mere continuation of policy by Harman, G. 2007. Explaining an explanatory gap. APA Newsletter on other means,” even more important for Fuchs’s fundamental Philosophy and Computers 6(2). idea is the inclusion of the “act of violence to compel our Jackson, F. 1982. Epiphenomenal qualia. Philosophical Quarterly 32: opponent to fulfil our will.” 127-32. I would have expected Fuchs to have taken “information” Jackson, F. 1986. What Mary didn’t know. Journal of Philosophy 83: as the starting point for his quest, following in the wake of Alfred 291-95. Toffler and Manuel Castells. Toffler predicted growing “total info Levine, J. 2001. Purple Haze. Oxford: Oxford University Press. wars” within and between the public and the private sector Loar, B. 1997. Phenomenal states (revised version). In The Nature of about the access and connection of data. The informatization Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. of our society is not just about social and economic changes, 597-616. Cambridge, MA: MIT Press. but about a major power-shift: who controls and regulates Nagel, T. 1974. What is it like to be a bat? Philosophical Review 83: information will be able to consolidate his political, commercial, 435-50. and/or organizational position. Stoljar, D. 2005. Physicalism and phenomenal concepts. Mind and With violence as the starting point for his quest for a Language 20: 469-94. definition, Fuchs curtailed the possible output and got drawn into a one-sided and, therefore, too limited approach of this specific topic. Today the study of information warfare is defined from a military point of view, concentrating basically on the risks of automated attacks on computer networks and related — 35 — — APA Newsletter, Spring 2009, Volume 08, Number 2 — security risks. I consider this a fundamental flaw in this field this presupposition, the thoughts run free. This train of thought of study in general, and in this book to a certain extent. The is of more use—at least to me: growth of corporate-led globalization has prompted the fine- All surveillance is political action no matter if it is tuning corporate communication and its more widespread undertaken by political actors, economic actors, or use with the aim of influencing international politics in favor of private individuals because it aims at influencing the corporations. I am not only talking about PR or propaganda, but decision capacity of others to select certain actions would like to include the manipulation of information on the with the help of information-gathering mechanisms Internet as well, the use of false identities in critical discussions, that are means of exerting and accumulating power. or the introduction of so-called independent third parties to (271) influence opinion making. I think it is essential to include topics like these in information warfare research, and I am afraid Fuchs Although sometimes Fuchs’s strive for completeness, the will miss them because of his focus on violence. We should dedication to fulfillment, results in an overflow of information, not let the military and the intelligence services determine the Internet and Society certainly is a valuable contribution to social definitions and the subsequent fields of research. theory in the information age. However, although the author Exactly that is already happening in this book, when looking aimed at providing answers, for me his work inspired a lot of for a definition of Netwar, for instance. Fuchs’s starting point new questions. But then again, isn’t that what critical thinking is the work of John Arquilla and David Ronfeldt (1996) who is all about? think of Netwar—in short—as an emerging conflict (and crime) at societal level, involving measures short of war, depending on using network forms of organization, doctrine, strategy, A Response to Lubbers and communication. As examples, they mention Hamas, the Zapatistas, the Asian Triads, and Chicago’s Gangsta Disciples Christian Fuchs (263). The problem with this definition is that it includes most University of Salzberg non-state actors, besides military groups, also transnational One grounding proposition that runs throughout my book is criminal organizations and NGOs. Fuchs rejects this assumption, that capitalism is a system of competition, which is a specific extensively explaining why. By describing NGO activities as a institutionalized form of violence that shapes modernity. My form of warfare, they are put into the same line with terrorists contention is that there is a dialectic inherent in capitalism, and war-waging parties: so that the competitive forces that advance capitalism at the As many of the contemporary social-movement same time produce potentials for co-operation that, if used organizations struggle for a global democracy in emancipatory political movements, allows resistance to with peace-full means, it is totally inappropriate to capitalism. employ the image of war here. Social struggle is not I do not see any advantage in conceiving war based on automatically warfare; it does imply conflict and Alvin Toffler’s concept. He has an extremely broad notion of war trying to change the distribution of political power that results in the implication that all forms of contemporary and society as a whole, but this doesn’t imply physical competition are also forms of war. The reason why I distinguish destruction of the lives and infrastructures of political between violence in general and war as a specific form of opponents. (264) violence (war as a systematic killing machine in the interest Although Fuchs seems genuinely annoyed here and does of dominant groups) is that I think that there is a categorial not disavow his own activist background when clarifying difference between certain forms of violence that needs to be his point of view, he fails to assess the roots of the problem. theoretically distinguished. So, for example, the Second World Arquilla and Ronfeldt both work for the RAND corporation, War that included the Shoah and Auschwitz is categorically an American think tank related to the State Department and different from struggles “between the public and the private as such an inspiration for U.S. foreign politics. RAND reports sector about the access and connection of data.” The first is have a solid influence on who is perceived an “enemy of the much more barbarian and dangerous than the second. If one state”—and by defining Netwar as a pre-stage of real war by intermingles such phenomena in the concept of war, then the end of the nineties certainly helped to confuse NGOs with certain phenomena that are singular in their form of horror will real war-mongering groups. be trivialized. I want to avoid such a trivialization. Fuchs’s focus on violence is again limiting him when War is an inherent feature of capitalism as expansion and looking for a definition of surveillance, and more specific defense are temporal necessities of accumulation, but it is not electronic surveillance. Starting point is Foucault’s panopticon the only form of violence that capitalism requires and makes use (and, to a lesser extent, Orwell’s 1984). Although he never of. Hence the need for such a distinction. Certainly, war makes related this to computer technology, Foucault’s analysis is use of various forms of violence that are also present in other still topical in the information society. For him, surveillance subsystems of capitalism than the system of war. Certain forms is a power that is “capable of making all visible, as long as of psychological violence and information manipulation that it…[can] itself remain invisible” (Foucault 1979, 222). Once Eveline wants to include in the category of war surely are part more, violence is the distinguishing factor, as people discipline of contemporary warfare. I use for these aspects the category themselves afraid of possible sanctions. Today, according of cognitive information war (pp. 258-261). But my contention to Fuchs, in symmetric relations like reading blogs or using is that not all manipulation is a type of warfare, only if it has webcams, there is no state of surveillance, but of monitoring. a specific role within systematic societal killing machines, David Lyon defines surveillance as “routine ways in which although all manipulation could be considered as a form of focussed attention is paid to personal details by organisations psychological violence. that want to influence, manage, or control certain persons or For the broad notion of war that Eveline points out with population groups” (2003, 5). Although Lyon doesn’t speak of the help of Toffler, I use the category of social struggles from violence, coercion is an immanent aspect of his notion, Fuchs above that aim at dominating and repressing subordinated determines (269). Again, this focus on the repressive aspects groups. For me it makes sense to start talking about war tends to limit his scope, which is a shame. When he lets go of based on Clausewitz as he is generally considered as the most

— 36 — — Philosophy and Computers — important theorist of war. Alvin Toffler is not so much known causing a change by allowing the entrance of new competition, for his concept of war, but more for his active engagement in such as the emerging virtual universities. But careful neoliberalism and the New Right’s pro-info-capitalist agenda examination shows that all parties are aware of, and have equal in the USA. access to, the same technologies and, like other enterprises, I do not see the theoretical and political advantage of have chosen to selectively develop different markets using the using broad theoretical notions of categories like war or same vehicles and methodologies. Competitive advantages surveillance, and consider such notions rather ideologically include price, curriculum, schedules, and similar traditional dangerous as I have tried to show, for example, in my book for vehicles for institutional differentiation. the broad notion of netwar employed by Arquilla and Ronfeldt. Since educational institutions at the post-secondary A generalized usage of the notion of war is likely to ideologically level, equally, have access to similar facilities, be they a essentialize warfare and to see it as a feature of all societies, physical environment (the traditional campus) or the virtual which ideologically forestalls the idea of a general state of peace environment (hardware and software), they can choose how within an emancipated society. they wish to compete. The balance between campus types Overall, I feel the need to stress that my book is not only a becomes one of choice, such as whether to add apartments book about information warfare and electronic surveillance, but instead of dormitories, or to rebuild the sports complex on that these are two of many notions that are discussed. The book campus. Thus, the idea that the introduction of e-learning as theorizes a broad range of ecological, technological, economic, being disruptive to education seems to be a mis-direction as political, and cultural issues of contemporary transnational the technology competes more with the construction industry informational capitalism. for a share of the education marketplace. Can the institution leverage faculty costs by hiring remotely or sharing faculty with institutions around the world? Can the campus reduce capital Deconstructing the Oracle: A Review of and concomitant overhead by offering virtual classes rather than building a new facility? Or, can an institution compete Disrupting Class internationally without having to open physical campuses in Clayton M. Christensen et. al. (New York: McGraw- other communities or countries? Hill, 2008). Even at the primary and secondary institutions, the issue seems to be similar but has been reframed as an educational Reviewed by Tom P. Abeles issue much as the airline pilot unions exhibited concerns Sagacity, Inc. about safety with the removal of a flight engineer from the cockpit of modern aircraft, or the railroad unions objected to The principle thesis of Disrupting Class is that the rise of e- the removal of the caboose and brakeman from freight trains learning changes education in much the same manner as the as roller bearings and other safety features rendered the job Apple computer did for mini/main frames in the business and obsolete. Unlike many disruptive technologies which lead home PC market we have today. Actually, the model being used to new competition, e-learning in its current embodiment of in the book is supposed to follow Christensen’s early writings on mapping bricks into clicks represents transformation within the innovation and disruption in the business sector.1-3 But, while his existing educational system regardless of provider. earlier writing maps post secondary education into his model,4 Clayton Christensen has written extensively about this book focuses on the K-12 educational arena. disruptive enterprises within the private sector and has Christensen defines two methods for the entrance of a developed his model which maps how these innovations have disruptive technology. Type 1 enters the market by meeting entered established markets, often becoming contributing the needs of an under or non-served population. Type 2 enters factors in the demise of major, established players. Disrupting the marketplace by providing a lower cost alternative which, Class is an effort to map the primary/secondary education initially, might prove to be an inferior but acceptable product system in the United States into his model. It is also his first or service. Because K-12 is compulsory, the marketplace is effort to be undertaken with co-authors who are involved distorted. The K-12 market has a host of political pressures, such in both speculating about and advocating for change in the as teachers’ unions and a variety of private pressure groups, education system. Christensen and colleagues identify the and thus the system is highly constrained. Even so, there are introduction of computers into education as the innovative currently a variety of options to the traditional public school change, a technology that is believed to be one that will follow system ranging from home-schools and charters to virtual his disruptive model and establish a dominant or dominating courses within schools or packaged as entire programs. presence in education’s future. Post secondary institutions are feeling fiscal pressure, while The core of the book’s argument rests on the standard the primary/secondary institutions have multiple problems sigmoid adoption curve, which the authors re-plot on a log of ranging from finances to a high percentage of students failing the ratio of converted courses/standard courses versus time to meet academic expectations. Additionally, there are external to determine that in about a decade about 50 percent of all pressures as private/for-profit institutions are starting to compete courses will be taught virtually in the K-12 school systems. by opening both brick-based and “click,” or virtual, campuses. While the authors carefully make the argument as to why virtual Thus there seems to be other motivations, not really stated conversion is compelling in order to reform the educational clearly, as to why Christensen felt compelled to try to fit the system, they fail to convince that this transformation will occur K-12 system into his innovation/disruption model, even if the within that time frame, that it is disruptive and/or transformative. focus is on the rise of e-courses and e-schools. This does not say that individuals are not convinced of the merit Education has always had competition between the public of on-line learning, or that there will be growth in the education and private sectors, both non- and for-profit institutions. The marketplace. Rather, the arguments put forth in Christensen’s expansion of all sectors into e-learning, different from traditional earlier works that look at the business sector do not seem to distance education, represents the equivalent of competitive map into the educational arena as the book attempts to present enterprises entering new or expanding old markets. Some have the scenario. seen the current e-learning technologies as being disruptive,

— 37 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Part of this concern rests on the definition of “disruptive,” demonstration that cyber-space activities, including simple e- which has within its definition a time-sensitive sensibility. To a mails, have consequences in the “outer world.” Already these Mayfly a decade may be an infinitely long time, whereas to a virtual worlds have created cultures so that travel between Galapagos tortoise such a period is but an eye blink. The book, all worlds, including this “outer world” and multiple virtual while suggesting some of the benefits for conversion, stayed worlds is more like traveling between countries with their own at the meta-level, choosing to define how such conversions customs. What happens to education when life-styles and habits would cause transformational changes, and drive the change. learned in cyber space, including collaborative learning and There is no clear and compelling model that would back the sharing, start to appear as disruptive in traditional classroom naïve prediction based on a linear, continuous change from settings? brick-to-click based education. Virtual worlds, unlike the “disruption” postulated in Christensen, in this and earlier works, suggests that these Christensen’s model, where bricks map to clicks, become disruptive technologies address a market with unmet needs the true disruption as cultures come in conflict much like the at a price/performance point that comes under the current, models described in Christensen’s earlier work. As was pointed dominant market in price. In education, with its great concern out above, the transition postulated by Christensen is based on a for cost, there is a reach for any cost cutting opportunity, pseudo-trend. And, as suggested, the virtual worlds have caused including e-learning, where e-delivery, today, is basically a what is labeled in complexity theory as a bifurcation, or what mapping of bricks into clicks and thus competes with current Christensen labels a disruption. This implies that the authors infrastructure rather than with the concept of a competitive need to take a second look at the entire domain of e-education educational model. As suggested above, K-12 has not been in light of virtual worlds designed exclusively for education but averse to adopting technology provided it offered a pedagogical founded on the early game-based virtual worlds. advantage. Thus, e-learning competes with alternative delivery Disrupting Class selectively sets forth the standard litany of systems from textbooks to classrooms and faces barriers problems in the K-12 system but recasts them in a manner to couched in educational terms only when it affects the current show that e-learning, particularly when asynchronous, could economic structure, such as teacher job security. accommodate students as individuals rather than age-defined Many in the education field see e-learning as another tool cohorts, allowing more opportunity for personalized learning— that complements current models; and the technology itself is home schooling on steroids. Unfortunately, the current models seen as an alternative to computer delivered systems such as for e-learning, whether synchronous or asynchronous, are still Power Point, word processing, and didactic material delivery treated by Christensen as class-based. This is clearly seen when much as overheads, e-books, white boards, video cameras, and one recognizes that the projected rate of adoption is measured two way audio/video delivery such as closed circuit TV. They using the ratio of classes converted to conventional classes. do see that it has advantages and problems. The entire “disruption” idea is modeled in a non-disruptive If Christensen and colleagues had gotten down to the manner either through technology conversion or introduction ground level to better define the disruptive computer-based of competitors into the current education system. “Virtual technology, they might have skipped over the dominant brick- Schools” are not seen as the emergent “Virtual Worlds” but to-click systems such as Blackboard or Adobe Connect and rather as packaged “click space” courses or courses delivered similar systems and gone straight to the emerging domain of electronically instead of presented synchronously in a traditional virtual worlds. Like the Apple computer that Christensen has classroom. Florida Virtual, K12.com, Advanced Academics, and pointed out in many of his writings, the early example of these others provide packaged courses or even 24X7 on-line tutoring worlds has been in the game and entertainment arena as we which fit within the traditional academic pattern. have seen in the Massive Multiplayer On-line Role Playing The rise of true virtual worlds with their emergent cultures Games (MMORPG) such as World of Warcraft with its millions that impact the traditional schools by their very presence and of global players. From this genre has arisen the user created the fact that students do not leave their avatars when they travel environments such as Kids/Tweens (KT) worlds, for example, back to the physical world represents the disruptive technology Whyville, Club Penguin, and Habbo Hotel, or both the adult and which awaits the second edition and rewrite of this volume. But teen worlds exemplified by Second Life (SL). Current estimates then Christensen, coming from a business school, need also are that registered users of virtual worlds, internationally, are include the impact of these worlds on the larger economy and about 300 million with about 10 percent as active users. culture globally, including the expanding arena of corporate The modern Internet is only about fifteen years old and education and, in particular, corporate universities. This is a virtual worlds only about seven years in age. “Steve Prentice, major shift that breaks the barrier between education and the fellow of the Garner Group, estimates that 80% of active Internet enterprise/cultural worlds outside of the protected boundaries users will be in non-gaming virtual worlds...by the end of 2011.”5 of the education system, K to 20. This seems to imply that as more people become Internet users, Endnotes more will become part of the application defined as virtual 1. Clayton Christensen. The Innovator’s Dilemma (Boston, MA: worlds, probably the specific “disruptive” e-technology. Since Harvard Business School Press, 1997). education is a lagging indicator, it will follow close behind. 2. Clayton Christensen and Michael Raynor. The Innovator’s As suggested above, this seems to imply that the competitive Solution (Boston, MA: Harvard Business School Press, disruption lies more with the providers of space and place, the 2003). construction industry which has provided, from the beginning, 3. Christensen, Clayton, et. al. Seeing What’s Next (Boston, MA: the evolving “little red school house.” Harvard Business School Press, 2004). The disruption is the virtual world, not just the Internet 4. Christensen, Clayton, et. al. Disrupting Education. In The and not just in education. Already several social scientists Forum on The Internet and the University, edited by Maureen have issued major studies of the cultures emerging in these Devilin et al. (Boulder, Colorado: Educause, 2001). worlds. This includes economists and anthropologists of various 5. Anders Gronstedt. Be First in Second Life. e>training, specialties.6-8 September 2008, page 29. Dizzywood, a KT social networking world of about 500,000 6. Tom Boellstorff. Coming of Age in Second Life (Princeton: registered users, recently teamed with a public school as a Princeton University Press, 2008).

— 38 — — Philosophy and Computers —

7. Gronstedt, see note 5. In contrast, the FOSS model assigns value not to the product, 8. Edward Castronova. Synthetic Worlds (Chicago: University of but to services. Chopra and Dexter cite Open Source founder Chicago Press, 2006). Eric Raymond’s remark that over 90 percent of developed software is never productized at all—it is specialized code produced in a services model (21). Programmers render their Decoding Liberation: The Promise of Free and services (for a fee) to produce the software. Value, in this model, Open Source Software is located in expertise. This value ascription is more conducive to the FOSS practice of distributing the source code along with Samir Chopra and Scott Dexter (Routledge, 2007) 232 the program. The releasing of the product as FOSS software pages. ISBN: 978-0-415-97893-4. has little economic impact when considered in the light of the service model. The scarcity of programmers still guarantees, Reviewed by Matt Butcher it is argued, the economic viability of software production. Loyola University Chicago But freed from the requirement of protecting software as proprietary products, the FOSS model can accommodate a It has been twenty-five years since Richard Stallman first broad range of organizational principles. This is readily observed announced his intention of creating a new Unix-like operating in the numerous ways in which FOSS projects self-organize: system and of giving away the source code. The announcement as dictatorships, as meritocracies, as direct democracies, and came in the form of the GNU Manifesto, which asserted even as anarchies. that software must be free for moral reasons.1 With this Perceived in this way, software development becomes announcement, the Free Software movement was born. a pragmatic venture benefiting a community. The goals A decade and a half later, the movement split—forked, as of the community become the roadmap for the software. a programmer would say—as those who shared the same Unlike the proprietary model, which must defend its territory practical vision, but disagreed with the ideology, formed the against threats of others developing competing products, the Open Source movement. Their practical similarities have FOSS model allows a community of like-minded people to resulted in the grouping of these two movements together as cooperatively build tools that fit their needs. Free and Open Source Software (FOSS). The second essay, “The Ethics of Free Software,” is an In Decoding Liberation, Samir Chopra and Scott Dexter attempt to further establish the ethical claims of the Free provide a philosophical defense of the Free Software movement, Software movement. En route to proclaiming Free Software the as well as the FOSS development model. However, the book ethically superior choice, Chopra and Dexter condemn not only is not intended to be a systematic defense. Instead, each of proprietary software, but also Open Source, marking the latter the five essays in the book explores a different philosophical as the subtle co-opting of Free Software by large corporations facet of FOSS, examining it under the rubrics of political eager to profit from the works of others (35). economy, ethics, aesthetics, philosophy of science, and political I think this approach is misleading, though. Granted that philosophy. such co-opting is possible, it is not the desired end of Open The first chapter, “Free Software and Political Economy,” Source development. Rather, advocates would claim that begins with a sprint through the history of computing. Chopra the Open Source model is more amenable to current trends and Dexter cover the rise of the hardware/software distinction, in the software industry, and is therefore more likely to gain the maturation of programming languages, and the genesis widespread acceptance. The Free Software movement attempts of network-based computing. Such technical background is an ethical paradigm shift, complete with an incommensurability requisite to understand the subject matter of the book. But it hypothesis stating that the motivations behind proprietary is the social milieu as much as the technologies themselves software are irreconcilably contrary to Free Software ethics. that becomes the driving force behind the narrative. Calvin In contrast, Open Source advocates favor an evolutionary Mooers’s dispute with Bernie Galler over whether a computer approach in which software developers will gradually “see language can be trademarked; Bill Gates’s infamous letter to the light” and change their practices for practical reasons. the Homebrew Computing Club; the Lisp Machine conflict at The successfulness of FOSS projects when compared to their MIT; the PDP-1 and “hacker culture”—these events become proprietary counterparts will reveal the superiority of the FOSS touchstones for an analysis of the current state of software. model. Better software can be had, FOSS advocates argue, Government-sponsored research agencies, institutions of by changing some of the social (and economic) practices of higher learning, and corporations have all contributed to the software development. Does the possibility that corporations rapid development of computing technology. It should come may gain from work done by FOSS developers make this as no surprise that ideological conflicts have arisen along fault position ethically reprehensible? Does the lack of guarantee lines. Both FOSS and the proprietary software find their roots that Open Source will come out of the process unaltered mean in this history. this approach is wrong? How does FOSS fare when analyzed as a political economy? I would also suggest that Chopra and Dexter’s defense of This is the question addressed in the first essay. “Understanding Free Software has unmet challenges. Primarily, the fundamental the political economy of a system requires an inquiry into ethical claims of Free Software are not defended. Stallman’s both its creation of value and its organizational principles. ethical position is typically couched in language like this: …How is it governed, and how are conflicts resolved?” (17) Chopra and Dexter begin with an examination of proprietary Non-free software is ethically poison. It’s a scheme software, whose value is derived from the combination of legal to keep people divided and helpless. It’s a form of protection (copyright and patent) and trade secrets (keeping the colonization. And that’s wrong. So when a person says software’s source code hidden). Through such mechanisms, to me “I’m going to make my program proprietary an individual software manufacturer can retain “commercial so that I can get money, so that I can work full time advantage” simply by making the code difficult, legally and developing the program” I say to him “That’s like practically, to duplicate. The organizational principles behind saying you’re going to rob people so that you can get the proprietary software model are those that best reinforce money, so that you can spend full time robing [sic] 2 the idea of software as a commodity. people.” It’s all wrong. And you shouldn’t do it. — 39 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Proprietary software is “ethically poison”? The person provides guidance to automated vehicles may be good for, say, writing proprietary software is robbing people? What exactly is providing transportation for the visually impaired. But is there the principle acting behind the rhetoric? Removing the colorful an ethical obligation to open that source when we know it can prose, we are left with a central principle looking something be re-tooled for maleficent purposes—self-guided weapons, like this: for instance?4 Writing non-free software is always morally wrong. There is a second issue that damages the ethical argument for Free Software. Chopra and Dexter invoke the rights of the The key to understanding this principle is found in the Free community over the rights of the individual without specifying Software advocates’ use of the term free. The free in Free why this community is entitled to trump the specific rights of Software is intended to denote freedom or liberty, not price. the individual. And the position is that the users of the software have certain Might the answer be found in the well-being of the “freedoms,” as enumerated in the Free Software Definition. community? In this case, the following proposition would have • The freedom to run the program for any purpose. to be defended: proprietary software damages the community. • The freedom to study how the program works, and But what is the purported harm done by not releasing all adapt it to your needs. software as Free Software?5 According to the phrasing of the • The freedom to redistribute copies so you can help Free Software Definition, the users of the software are the ones your neighbor. who have certain rights regarding the software, and the creator of the software has many obligations to the software’s users. • The freedom to improve the program, and release Chopra and Dexter take this claim even farther, invoking the your improvements to the public, so that the whole Harms Principle on behalf of the users: a user’s freedoms, with community benefits.3 regard to the software he uses, can only be restricted if such What is unclear, and what Chopra and Dexter do not uses of the software cause harm to others (60). The user, in this adequately address, is why a user has these privileges (claimed argument, is treated as an individual whose privileges regarding as freedoms), and why it is required of the programmer to the use of the works of others can be restricted only when the act as the guarantor of these. When I get ahold of a book, am user’s privileges cause harm to others. I entitled to change the text and redistribute it—even if my It is difficult for me to understand how the individual who alterations are an improvement? If a piece of music strikes labors over a product suddenly finds herself in the position of my fancy, am I acting ethically to rewrite a few notes and being ethically obligated to elect either to keep everything to redistribute it? Is the composer ethically obligated to let me herself (non-distribution), or to give away not only the product do that? When an architect builds a house, is she obligated of her labor, but also the means of creating more of the same. to release the blueprints along with it? And is she obligated to When did the user suddenly have the inalienable privilege of release these blueprints in such a way that others may modify using the product of another’s labor? And how is this granted as or redistribute them? What about the industrial designs created a “freedom”? It seems to me that such a perspective represents by large corporations? Ought every Ford, Chevy, and Saturn an inversion of the typical relationship between the creator, come with complete schematics and a license to modify and the work, and the consumer. While I would not claim that the duplicate at will? Even in a clearly ethical context, the matter inversion is itself conclusive evidence against the Free Software of pharmaceutical research and development, many still balk position, again it seems that some argument in defense of this at the suggestion that a pharmaceutical company is obligated perspective is necessary. to share its drug formulas even if doing so would result in the saving of lives. On the other hand, the principle Chopra and Dexter rely upon can be used to make the opposite point. The harms From art to technology, we readily grant these other principle, when employed in the context Mill intended, achieves endeavors the right not to share. Why is software different? the inverse effect. After all, in Mill’s formulation, the producer’s What characteristic of software justifies the demand that liberties are to be protected.6 Unless it can be shown that the for every binary program released, the source code—the creator is, by keeping her code, doing harm, she is not obligated blueprints—ought also be released as well (and under terms to give up her freedom to do what she wants with her code. that allow modification and redistribution)? Refusing to give her software’s source code to others—even if it In most of the cases sketched above, the protections of cost her nothing to give it to them—is not ipso facto sufficiently patent, copyright, and trade secret laws are invoked to protect harmful to warrant ethical or legislative injunction.7 If she wants the investment required to produce such works. And the guiding to release it as FOSS, that is certainly commendable, as she may principle is intuitive to us: people ought to be able to derive in fact have chosen the most beneficial course. But if she does benefits from their work. Such a position is readily defended not, there are no grounds for appeal to the harms principle. from a utilitarian position. Society draws many benefits when While I have reservations about the successfulness of the the artist, architect, researcher, and engineer have economic argument presented in the second essay, it brings to light many incentive to produce. Is the same not true for software? of the ethical dimensions of the FOSS models. Ethical matters In contrast to the Free Software movement, which seems lie at the heart of the FOSS debate. However, as subsequent to remain silent in regard to the argument above, Open Source essays in this book show, there are other areas of FOSS that are advocates espouse a classic pragmatist response: if what we of philosophical interest. are after is something useful, something good, something When describing code, software developers use terms like stable and bug free, we are more likely to get there if we work elegant, beautiful, and clean. They talk about style, signatures, cooperatively. Roughly put, making software available under and themes. Programmers self-describe their coding styles as an Open Source model is good when it helps us achieve our going through periods characterized by particular constructions desired ends. or design patterns. Programming is an art, they say. A popular T- Furthermore, there are cases in which the Open Source shirt once sold by ThinkGeek.com even advertised its wearer as model seems to have the ethical advantage over Free Software. a Code Poet. In the third essay, “Free Software and Aesthetics,” We can imagine situations in which software should not be Chopra and Dexter examine various claims about the role of released under a FOSS license. A computer program that aesthetics in software development.

— 40 — — Philosophy and Computers —

Drawing from Aristotle, Chopra and Dexter suggest one away in order to give a commercial entity a short term market possible guiding principle, which may elucidate how coding advantage. Stagnation, splintering, and the failure to create an may be construed as an art form: environment of critical inquiry may very well lead, as Chopra and Dexter suggest, to the failure of computer science as a Much as an artist extracts form from objects of science. experience, such as the human body, and imposes that form on matter, such as canvas, the programmer Does FOSS provide a solution? FOSS code is open to imposes the form of an abstract algorithm in, and on, inspection, modification, peer review, and experimentation. a particular programming language. The programmer Novel algorithms can be shared, improved, and employed in uses the medium of a programming language to equally novel situations. As an educational tool, FOSS provides render the algorithm in the material substrate of digital not only the material to study, but also the groundwork for circuitry. (77) experimentation, as students can begin with existing tools and learn by attempting to modify the tools. The tenets of FOSS Code may not provoke an immediate visceral experience, are more compatible with the aims of computer science, qua but this is not detrimental to the aesthetic claim. Just as the science, than the proprietary model. “A computer science appreciation of literature benefits from background knowledge, based on open code, protocols, and standards promises so the appreciation of code is contingent upon the knowledge a reconfiguration of research, industry, and pedagogy, of the reader. The “Bottles of Beer” Perl script (79), composed ameliorating the effects of the current entanglement of the of code formatted to look like bottles, may elicit a passing academic and corporate sectors on the practice of science” smile from the casual observer. But to an experienced Perl (142). programmer, the code is stunning, for she brings with her In the final chapter, Chopra and Dexter turn to political the knowledge of the nature of the insight that, coupled with philosophy. Drawing on an image popular in the public mastery of the language, begot the work. imagination, they use the term cyborg as a description of the What is the relationship between FOSS and the aesthetics combination of the human with the machines upon which he of code? Chopra and Dexter present FOSS communities as relies. From telephones to laptops, automobiles to toaster ovens, practitioners of a form of communal art. They begin by inverting we surround ourselves with devices that help us extend our the normal analysis of the artist and object distinction: it is not abilities and augment our environment for our own benefit. In good coders that produce good code, but “good coders are extending ourselves in this way, have we begun the process of produced by good code” (92). In a community of programmers, voluntary enslavement? This is the tantalizing question driving where all are actively working on the same software, it is “Free Software and the Political Philosophy of the Cyborg the requirements of this massive collaboration that drive the World.” aesthetics of the system. The code must be readable; it must Legal scholar Lawrence Lessig opened his book Code be extensible; it must be clearly and logically organized. with the now-famous sentence, “Code is law.”8 In a world A more radical aspect to this collaborative model is the where humans regularly employ computers as prosthetics, a decentralization of the role of the author in the production of new form of legislation determines the limits of our actions: the work. The community as a whole takes on the productive the hard-coded rules of software. The software that controls role; indeed, even the interaction of creative community with these devices places limits on how we use these artifacts and the “audience,” the users and critics, exemplifies the degree to what they can do for us. which authorship is distributed. Those who file bug reports or The ones who create the code define the limitations. What request features for FOSS software are likely to find themselves recourse can be had if the code for the software is inaccessible? invited to participate in the continued development of the We must simply abide by the laws set by the programmer. And software. The audience then become fellow artists. when software becomes institutionalized, the hard and fast The fourth essay shifts to philosophy of science. “Free rules of the code are effectively combined with other law. From Software and the Scientific Practice of Computer Science” electronic voting to financial software, code defines limits and begins with a defense of computer science as a proper dictates choices. science. But midway through the essay, the tenor changes. If our existing freedoms are jeopardized by the ability of From defending computer science, Chopra and Dexter turn to proprietary code to function as law, then indeed the freeing the challenges posed when research and economic interests of source code is an ethical issue. As Chopra and Dexter become entangled. The themes of political economy (chapter bring the final essay to a close, they extol what they call an 1) and ethics (chapter 2) are woven back into the discussion. Anarchist Ideal that espouses the destruction of asymmetric The political economy of the proprietary software realm, with power relations. If the very laws that create our social fabric in its interests in producing and protecting a product, benefits this cyborg world can be freely altered, then perhaps a radical from its relationship with the sciences. On the pedagogical system of self-governance can arise. “Because the cyborg world side, curriculums biased toward proprietary software packages brings to life new legal and political structures, as code merges help the producers of this software entrench their product, with law, we can see in its fundamentals the glimmerings of a increasing market share by effectively training students to be new, transparent society, one that by making participation in it users. Furthermore, the insinuation of the proprietary model into voluntary attains the true meaning of a compact, on achieved the educational system gives the impression that the proprietary without coercion” (172). model is the dominant, correct, or even the only model. As the first sustained philosophical treatment of the Free But the proprietary model is antithetical to the open attitude and Open Source Software movements, Decoding Liberation of modern science. Inquisition is blocked when the source code achieves one thing for certain. By covering broad areas of the is unobtainable. Experimentation is hindered when the tools philosophical landscape—political philosophy, aesthetics, are artificially constrained and unmodifiable. Reproducibility is ethics, and philosophy of science—Chopra and Dexter have even challenged when information cannot be shared because shown the range of FOSS issues that can be informed by the methods and results of research become the protected philosophy. In opting to paint with a broad brush, they have, I property of a for-profit organization. Instead of an openness that believe, opened many spaces for critical dialog. leads to widespread innovation, new algorithms are secreted

— 41 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Endnotes deeply entrenched in the world and that the body, mind, and 1. Richard Stallman. “The GNU Manifesto.” The GNU Project. world work together in interesting (and often surprising) ways. http://www.gnu.org/gnu/manifesto.html. Much of the empirical support Clark discusses in this section 2. Richard Stallman. “Free Software.” The GNU Project: 2004. can also be found in Clark’s previous works.4 In Chapter http://www.gnu.org/philosophy/nit-india.html. One, “The Active Body,” and Chapter Two, “The Negotiable 3. Richard Stallman. “The Free Software Definition.” Free Body,” Clark presents evidence to support the conclusion that Software Foundation. http://www.gnu.org/philosophy/free- humans are profoundly embodied creatures. Clark argues sw.html. that humans and their environments are best understood as 4. The fact that the Free Software advocate has to answer this coupled systems when performing certain tasks. Clark claims question in the affirmative—yes, the source code must be that our interactions with the world involve more than mere released—makes the Freedom Zero problem, discussed at tool use. Humans and external objects can create unique some length by Chopra and Dexter, all the more troubling. networks that amount to new systematic wholes. In Chapter 5. Of course, there are cases where harm may be done by not Three, “Material Symbols,” Clark examines the role language releasing some software’s source code. Chapter five provides plays in human cognition. The final chapter of the first section, examples. But at issue is whether harm is necessarily done “World, Incorporated,” focuses on ways that the environment when software is released as non-free software, where that shapes cognition and cognition shapes the environment. In this harm is clearly a result of the software’s non-free status. chapter, Clark also briefly outlines the arguments made in the 6. John Stuart Mill. On Liberty (NY: Bobbs-Merrill, 1956), 114- 1998 paper, “The Extended Mind.” 15. The second section of Supersizing the Mind, “Boundary 7. In a similar context, Mill lists “fraud or treachery, and force” as harms warranting intervention. He explicitly rules out Disputes,” attempts to move beyond the thesis that the competition, scarcity, wasted exertion, and disappointment. mind and body are embedded in the world and argue that Mill, 115. the physical machinery of the mind extends past corporal 8. Lawrence Lessig. Code: Version 2.0 (NY: Basic, 2006), 1. boundaries. Rather than objects in the world functioning as simply tools, some objects become an intricate part of a system of cognition. In this section Clark also examines criticisms of Supersizing the Mind: Embodiment, Action, the extended view and formulates responses to critical worries. In Chapter Five, “Mind Re-bound,” Clark discusses current and Cognitive Extension arguments against the extended perspective and offers forceful Andy Clark (New York: Oxford University Press, 2008). counterarguments. In Chapter Six, “The Cure for Cognitive Hiccups,” Clark compares and contrasts the extended thesis 286 pp. with a closely related view. This related view is The Hypothesis of Embedded Cognition (HEMC). HEMC has been defended, Reviewed by Amy E. White 5 Ohio University–Zanesville most notably, by Robert Rupert. In this chapter Clark examines the arguments Rupert offers against the extended view. Clark In Supersizing the Mind: Embodiment, Action and Cognitive argues that Rupert’s criticisms of the extended position are Extension, Andy Clark persuasively argues that the human mind unwarranted. Clark concludes that the extended thesis is is extended beyond mere physical boundaries. Clark believes superior to HEMC in capturing the full extent of the embedded that the body and world create coupled, extended cognitive nature of humans. Chapter Seven, “Rediscovering the Brain,” circuits. These circuits extend and supersize the machinery stresses the fundamental role of the biological brain in human of the mind. Clark contrasts his vision of the mind, which he cognition. Clark maintains that our biological brain is essential 6 labels EXTENDED, with a more traditional view of the mind as to the problem-solving success of humans. Clark believes BRAINBOUND. A brainbound view of the mind is one where the that the extended thesis is compatible with the idea that the brain and mind are basically identical. According to this view brain is essentially a thinking thing and that the locus of human the body is simply a sensory system and the world is the arena cognition resides in the biological brain. where the brain and body act. Clark claims his extended view The third section of Supersizing the Mind describes of the mind is “maximally opposed” to the brainbound vision.1 limitations of the extended vision. Chapter Eight, “Painting, Clark asserts that human cognitive activity sometimes includes Planning, and Perceiving,” investigates the Strong Sensorimotor “inextricable tangles of feedback, feedforward, and feed-around Model. Clark presents intriguing empirical evidence in support loops: loops that promiscuously cris-cross the boundaries of of his conclusion that the Strong Sensorimotor Model goes brain, body and world.”2 According to Clark’s extended view, too far. Clark argues that the model is correct to stress the the mind is not only in the head, “it leaks out into the world.”3 importance of embodiment but ignores many complexities of Supersizing the Mind is divided into three sections and human information processing. Chapter Nine, “Disentangling contains a total of ten chapters. Also included in the book is Embodiment,” is a forward-looking chapter where Clark makes a useful appendix where the 1998 foundational paper, “The headway towards developing a science of the embodied mind. Extended Mind,” co-authored by Clark and David Chalmers is The final Chapter, “Conclusions: Mind as Mashup,” provides a reprinted. Readers who are not familiar with the basic tenets short conclusion to the book. of the extended perspective will benefit from reading the Supersizing the Mind is a remarkable and important appendix before proceeding to the main text. David Chalmers work; however, I believe there are a few areas of concern that also provides an engaging foreword where he introduces should be addressed. While ten years have passed since the the iPhone as a possible candidate for becoming part of the paper “The Extended Mind” appeared in Analysis, the central machinery of cognition. arguments in favor of Clark’s extended view have not changed. The first section of Supersizing the Mind, “From Embodiment Most of the empirical evidence presented in the first section to Cognitive Extension,” contains four chapters. Each chapter of the text can be found in Clark’s previous work. Thus, the offers illustrations of the ways the human body is embedded first eighty-two pages of Supersizing the Mind presents little in the world. In this section Clark presents a great deal of material that is novel and is simply review to anyone already empirical evidence that supports the notion that humans are familiar with arguments for the extended view. Thankfully, the

— 42 — — Philosophy and Computers — second and third sections do provide new material. Especially of with the empirical evidence offered by Clark in the first section interest in the later sections of the book are Clark’s forceful and of Supersizing the Mind. Rupert’s embedded view would generally compelling responses to criticisms of the extended keep the distinction between object and world intact while perspective. stressing the role the environment plays in shaping a subject. Another aspect of Supersizing the Mind that is repetitious In the sixth chapter of Supersizing the Mind Clark argues for is the often-used example concerning Otto’s notebook. This the superiority of the extended perspective over alternative example appeared in the 1998 paper “The Extended Mind” views; however, his arguments remain unconvincing. Clark and concerns Otto, a man whose memory is impaired. Otto even seems to acknowledge this underdetermination when he carries a notebook in which he writes helpful reminders. The advises theorists to flip between perspectives when addressing information in the notebook supplements Otto’s failing memory. questions.9 Clark argues that Otto’s notebook becomes an extended part My final concern involves a potential contradiction in of Otto’s mind. Clark also maintains that Otto’s notebook has Clark’s view. Clark is clear that the brain remains the locus and sufficient functional similarity to biological stored memory control center of human cognition. However, if the other tenets and should be considered an extended set of stored beliefs. A of the extended view are taken seriously, there is no reason to decade after this famous example was formulated, Clark still hold that the brain need have this special place. Crucial to the frequently appeals to the case in his arguments. Numerous extended perspective is the Parity Principle. The Parity Principle criticisms have been levied against this example. Even David claims that if a process works in such a way that we would Chalmers, in the foreword to Supersizing the Mind, suggests that call it cognitive if it were in the head, it should also be counted there may be problems concerning this example.7 In the second as a cognitive process even if it occurs outside the brain. section of Supersizing the Mind, Clark eloquently answers many According to this principle, we should ignore the traditional of the criticisms given of this example; however, it remains boundaries of skull and skin.10 Concerning the Parity Principle, troubling. Given the problems Otto’s notebook encounters, it Clark is careful to clear up what he believes is a persistent is unclear why Clark doesn’t abandon this example for another misreading. According to Clark, the Parity Principle has often heuristic devise. been criticized by invoking differences between the function The example of Otto’s notebook is simply not intuitive. of the outer world and the inner mind. However, Clark believes We can rightly ask if stored memory in a notebook really has a that the principle implies that there is no need for any “deep high degree of functional similarity to stored biological beliefs. similarity” to exist between inner and outer processes.11 Clark Clark claims the differences between biological beliefs and writes of the Parity Principle, “The Parity Principle was meant Otto’s notebook are merely superficial. When Otto calls upon to engage our rough sense of what we might intuitively judge his notebook, Clark believes, the information serves the same to belong to the domain of cognition—rather than, say, that of purpose as biologically stored beliefs. While some similarity digestion—but to do so without the pervasive distractions of may exist there are vast differences to be found. Clark claims skin and skull.”12 This implies that we may be able to count that information retrieved from Otto’s notebook can “guide things as performing cognitive acts that are very different from Otto’s reasoning and behavior in the same way as information the biological brain. On the other hand, Clark claims that the retrieved from biological memory.”8 This is not necessarily extended view is compatible with a persistent biological core true. My biological memories and beliefs often appear when I of cognition. In fact, Clark asserts that the biological brain is do not explicitly call upon them. They weave themselves into in the “drivers seat” in all cognitive functioning.13 Clark writes, my thoughts in often-unexpected ways. They interact with my “Individual Cognizing, then, is organism centered even if it is other stored beliefs and often make interesting connections not organism bound.”14 Clark also believes that supporters that sometimes appear without me seeking them out. Through of profound embodiment have often downplayed the vast years of conditioning it is even possible that my biologically contributions of the biological brain to cognition and that the stored beliefs operate at an unconscious level and influence biological brain is necessary for human cognitive complexity. my actions and behaviors in ways that I am not fully aware. Given the Parity Principle and Clark’s insistence that the brain Otto’s notebook lacks this functional similarity. Another obvious is the locus of cognition, a potential contradiction is created. problem with the example of Otto concerns his endorsement If the boundaries of skin and skull are arbitrary to cognition, of the notebook’s contents. If Otto’s memory is truly impaired, there is no reason to claim that the brain is necessarily the the question can reasonably be asked as to how he remembers locus of cognitive control. While there may be nothing that what the notebook contains. He can’t claim to believe what currently functions like the brain outside of skin and skull, there street The Museum of Modern Art is on if he can’t remember is certainly a logical possibility that something could function if his notebook contains the information. Also, if he does find in a similar way (perhaps an advanced computer). If this is the information in the notebook, how can he be said to endorse the case, the brain need not, following the Parity Principle, be an material inside if he can’t remember the reasons the material indispensible locus of control. exists? It seems much more intuitive and likely to claim that The possibility of an extended mind is especially fascinating Otto is simply using the notebook as a resource. when considering consciousness. In the foreword to Supersizing The fact that Otto’s notebook can easily be thought of the Mind David Chalmers claims that the extended mind thesis as a tool brings up another point. There can be no doubt that may be extended. Thus, we may be able to have extended the human mind and world interact in many ways and that emotions, perception, imagination, and desires.15 Chalmers we are deeply embodied creatures; however, to take this even claims that consciousness could possibly be extended. A assertion as far as Clark wishes may be unwarranted. There deep problem with the idea of extended consciousness is that are other reasonable views and it is possible that the extended we have very little agreement on the nature of consciousness. mind thesis is underdetermined. An alternative view, such as Given this lack of agreement it is nearly impossible to definitively that proposed by Robert Rupert, offers an equally plausible tell if something is conscious and performing cognitive functions. option. According to Rupert’s view, humans have deeply If we lack a biological benchmark to compare a system to, we embedded cognition and cognitive processes depend heavily risk facing something like John Searle’s Chinese Box. Thus, this on environmental influences. However, under Rupert’s thesis is an area where much exciting work could be done. the mind remains unextended. Such a perspective is consistent

— 43 — — APA Newsletter, Spring 2009, Volume 08, Number 2 —

Overall, Supersizing the Mind is a fascinating manuscript. SOCIAL IMPACT OF THE NEW MEDIA Clark presents a generally compelling case for the extended Steven Tötösy de Zepetnek — “New Media Technology, thesis and makes a valuable contribution to the literature. Both Interculturalism, and Intermediality” philosophers and cognitive scientists will benefit from a careful Michał Ostrowiski — “Immersive Nature of Art” reading of the work. Włodzimierz Gogołek — “Some Aspects of the Social Impact Endnotes of the Web” 1. Andy Clark. Supersizing the Mind: Embodiment, Action and * * * Cognitive Extension (New York: Oxford University Press, 2008), p. xxviii. Andrzej Zabłudowski, in Memoriam 2. Ibid. Dialogue and Universalism invites articles in all areas of 3. Ibid. philosophy; a special block of articles on the social impact of 4. See especially Andy Clark, Natural-Born Cyborgs: Minds, new media is planned for a Winter issue 2009/2010 Technologies and the Future of Human Intelligence (New More information at: [email protected] http://www.dialogue. York: Oxford University Press, 2003). uw.edu.pl 5. See Robert Rupert, “Challenges to the Hypothesis of Extended Cognition,” Journal of Philosophy 101 (August 2004): 389- 428. The Philosophy of Computer Science 6. Clark, Supersizing the Mind, p. 149. 7. Ibid. p. xii. http://pcs.essex.ac.uk/ecap09/cfp.html 8. Ibid., p. 98. Special issue of Minds and Machines (2010) 9. Ibid., p. 139. We invite submissions concerned with philosophical issues 10. Ibid., p. 77. that arise from reflection upon the nature and practice of the 11. Ibid., p. 114. academic discipline of computer science. In particular we welcome submissions concerned with questions such as the 12. Ibid. following: 13. Ibid., p. 122. 1. What kinds of things are programs? Are they abstract or 14. Ibid., p. 123, emphasis in the original. concrete? (Moor 1978; Colburn 2004) 15. Ibid., p. xiv. 2. What are the differences between programs and algorithms? (Rapaport 2005a) 3. What is a specification? And what is being specified? (Smith 1985; Turner 2005) CALLS FOR PAPERS 4. Are specifications fundamentally different from programs? (Smith 1985) 5. What is an implementation? (Rapaport 2005b) Dialogue and Universalism informs of a 6. What distinguishes hardware from software? Do programs special issue Vol. XIX, No. 1–2/2009 related exist in both physical and symbolic forms? (Moor 1978; Colburn to the focus of this Newsletter 2004) 7. What kinds of things are digital objects? Do we need a new WEB-BASED TECHNOLOGY AND THE NEW UNIVERSITY ontological category to house them? (Allison et al. 2005) Guest Editor — Piotr Bołtuć 8. What are the objectives of the various semantic theories of IMPLICATIONS OF INFORMATION TECHNOLOGY programming languages? (White 2004; Turner 2007) Gaetano Aurelio Lanzarone — “Computational Reflection. 9. How do questions in the philosophy of programming Machines and Minds” languages relate to parallel ones in the philosophy of language? Luciano Floridi — “Artificial Companions and their Philosophical (White 2004) Challenges” 10. Does the principle of modularity (e.g., Dijkstra 1968) Piotr Bołtuć — “Paradigm Change in Higher Education Due to relate to the conceptual issues of full-abstraction and the World Wide Web” compositionality? TECHNOLOGY, PHILOSOPHY, AND ACADEMIA 11. What are the underlying conceptual differences between Tom P. Abeles — “Does Philosophy have a Future?” the following programming paradigms: structured, functional, logic, and object-oriented programming? Jerzy Mischke — “The Role of e-Learning in Paradigmatic Transformation” 12. What are the roles of types in Computer Science? (Barandregt 1992; Pierce 2002) Boria Sax — “Knowledge and Wisdom in Academia” 13. What is the difference between operational and denotational H. E. Baber — “The Virtuous and Vicious Circles of Academic semantics? (Turner 2007) Publishing” 14. What does it mean for a program to be correct? What is Viorel Guliciuc — “How Do We Need Universities in a the epistemological status of correctness proofs? Are they Technological World?” fundamentally different from proofs in mathematics? (DeMillo ONLINE LEARNING (in collaborations with E-Mentor) et al. 1979; Smith 1985) Vlad Wielbud — “The Second Waive—Why Is US Higher 15. What do correctness proofs establish? (Fetzer 1988; Fetzer Education Changing?” 1999; Colburn 2004) Andrzej Wodecki, Rafał Moczadło — “University in Second 16. What is abstraction in computer science? How is it related Life—the Experiment’s Results” to abstraction in mathematics? (Colburn & Shute 2007; Fine 2008; Hale and Wright 2001)

— 44 — — Philosophy and Computers —

17. What are formal methods? What is formal about formal Minds & Machines special issue methods? What is the difference between a formal method Important dates (tentative): and informal one? (Bowen & Hinchey 2005; Bowen & Hinchey • Submission deadline: 1 Dec. 2009 1995) • Notification: 1 May 2010 18. What kind of discipline is computer science? What are the roles of mathematical modelling and experimentation? • Appearance: Dec. 2010 (Minsky 1970; Denning 1980; Denning 1981; Denning et al. 1989; Submission guidelines: See Springer’s Instructions to Authors Denning 1985; Denning 1980b; Hartmanis 1994; Hartmanis 1993; Two special editions of Minds and Machines (2007) and the Hartmanis 1981; Colburn 2004; Eden 2007) Journal of Applied Logic (2008) dedicated to the philosophy 19. Should programs be considered as scientific theories? of computer science have already appeared in print. Another (Rapaport 2005a) special edition of Minds and Machines is planned for 2010. 20. How is mathematics used in computer science? Are Raymond Turner, editor, special issue mathematical models used in a descriptive or normative way? School of Computer Science and Electronic Engineering, (White 2004; Turner 2007) University of Essex 21. Does the Church-Turing thesis capture the mathematical Amnon H. Eden, associate editor, Minds and Machines notion of an effective or mechanical method in logic and School of Computer Science and Electronic Engineering, mathematics? Does it capture the computations that can University of Essex be performed by a human? Does its scope apply to physical • The topic will be also discussed at the 7th European machines? (Copeland 2004; Copeland 2007; Hodges 2006) conference on Computing and Philosopy ECAP 2009 22. Can the notion of computational thinking withstand conference 2-4 July 2009 philosophical scrutiny? (Wing 2006) 23. What is the appropriate logic with which to reason about program correctness and termination? (Hoare 1969; Feferman Last Call on the Ontological Status 1992) How is the logic dependent upon the underlying programming language? The APA Newsletter on Philosophy and Computers is still 24. What is information? (Floridi 2004; Floridi 2005) Does this seeking contributions on the topic of “The Ontological Status notion throw light on some of the questions listed here? of Web-Based Objects”—the block of articles pertaining to this topic will appear in our Fall issue (closing May 30, 2009; 25. Why are there so many programming languages and late contributions may go to the following issue). Authors are programming paradigms? (Krishnamurthi 2003) encouraged to tackle this relatively new and exciting topic. We 26. Do programming languages (and paradigms) have the hope that some of the contributions would reach beyond the nature of scientific theories? What causes a programming horizon of current work. paradigm shift? (Kuhn 1970) Contributions, preferably of up to 3,000 words, should be 27. Does software engineering raise any philosophical issues? emailed to the editor, to: [email protected] (Eden 2007)

— 45 —