Hyperauthorship:APostmodernPerversionorEvidence ofaStructuralShiftinScholarlyCommunication Practices?

BlaiseCronin SchoolofLibraryandInformationScience,IndianaUniversity,Bloomington,IN47405-1901.E-mail: [email protected]

Classicalassumptionsaboutthenatureandethicalen- uel,1997).Curiously,theperceivedseriousnessoftheprob- tailmentsofauthorship(thestandardmodel)arebeing lemdoesnotfindechoinotherscientificfields.Thisarticle challengedbydevelopmentsinscientificcollaboration andmultipleauthorship.Inthebiomedicalresearch (a)beginswithabrief,historicaloverviewofscholarly community,multipleauthorshiphasincreasedtosuch publishing,focusingontheroleoftheauthorandthe anextentthatthetrustworthinessofthescientificcom- constitutionoftrustinscientificcommunication;(b)offers municationsystemhasbeencalledintoquestion.Doc- animpressionisticsurveyandanalysisofrecentdevelop- umentedabuses,suchashonorificauthorship,havese- mentsinthebiomedicalliterature;(c)explorestheextentto riousimplicationsintermsoftheacknowledgmentof authority,allocationofcredit,andassigningofaccount- whichdeviantpublishingpracticesinbiomedicalpublishing ability.Withinthebiomedicalworldithasbeenproposed areafunctionofsociocognitiveandstructuralcharacteris- thatauthorsbereplacedbylistsofcontributors(the ticsofthedisciplinebycomparingbiomedicinewithhigh radicalmodel),whosespecificinputstoagivenstudy energyphysics,theonlyotherfieldwhichappearstoexhibit wouldberecordedunambiguously.Thewiderimplica- comparablehyperauthorshiptendencies;and(d)assesses tionsofthe‘hyperauthorship’phenomenonforscholarly publicationareconsidered. theextenttowhichcurrenttrendsinbiomedicalcommuni- cationmaybeaharbingerofdevelopmentsinotherdisci- plines. Introduction

Inrecentyears,literallyscoresofsurveys,studies,and ABriefHistoryofAuthorship opinionpieceshavebeenpublishedintheexpansivebio- medicalliteratureonthesubjectofmultipleauthorship AccordingtoManguel(1997,pp.182–183),theearliest (e.g.,King,2000).Whilecoauthorshipandcollaboration namedauthorinhistorywastheMesopotamianPrincess, studieshavebeencarriedoutinmanydisciplinesandfields Enheduanna,who,morethan4,000yearsago,signedher (seeBordons&Gomez,2000forarecentreview),thescale nameattheendoftheclaytablets,onwhichwereetched ofthephenomenonandassociatedethicalabuseshave songsinhonorofInanna,goddessofloveandwar.Overthe provedtobesingularlyproblematicinthebiomedicaldo- centuries,however,certaingenresoftext(e.g.,epicpoems, main(e.g.,Houston&Moher,1996).Aswillbecomeclear, sagas)havenotalwaysrequiredauthors:“Theiranonymity thecentralissueisnotjustoneofmultipleauthorship,but wasignoredbecausetheirrealorsupposedagewasa tointroduceaneologismofhyperauthorship—massiveco- sufficientguaranteeoftheirauthenticity”(Foucault,1977, authorshiplevels.Someconsequencesofthesetrendshave p.125).ForBarthes(1977,pp.142–143),though,theauthor beenextensiveeditorialcommentaryandcorrespondencein reallyis“amodernfigure...emergingfromtheMiddle theleadingbiomedicaljournals,revisededitorialguidelines AgeswithEnglishempiricism,Frenchrationalismandthe forauthorsandcollaboratorssubmittingtoreputablejour- personalfaithoftheReformation.”Whilecriticaltheorists nals,and,mostradically,aproposaltoreplaceauthors mayquestionthe“prestigeofauthorship”and“allmanifes- entirelywithlistsofcontributors(Rennie,Yank,&Eman- tationsofauthor-ity”(Birkerts,1994,pp.158–159)— whichhelpsexplainthepredilectionforpostmodernistand eschatologicaltitlessuchasTheDeathoftheAuthor(Bar- ReceivedJuly12,2000;acceptedDecember20,2000 thes,1977),WhatisanAuthor?(Foucault,1977),andThe DeathofLiterature(Kernan,1990)—thereislittledoubt ©2001JohnWiley&Sons,Inc. ● Publishedonline13March2001 thatboththesymbolicandmaterialconsequencesofauthor-

JOURNALOFTHEAMERICANSOCIETYFORINFORMATIONSCIENCEANDTECHNOLOGY,52(7):558–569,2001 ship today are rather more far-reaching than in Enheduan- Authorial rights were not always asserted with the kind na’s time. To state the obvious, public affirmation of au- of forcefulness we take for granted today—think of the thorship is absolutely central to the operation of the aca- often heated disputes which routinely flare up over the demic reward system, whether one is a classicist, ordering, inclusion, and elision of names attached to mul- sociologist, or experimental physicist. Authorship (and the tiauthored papers (Wilcox, 1998). As Rennie and Flanagin recognition that flows therefrom) is the undisputed coin of (1994) remind us, there is no standard method for determin- the realm in academia: it embodies the enterprise of schol- ing order, nor any universalistic criteria for conferring au- arship (Bourdieu, 1991; Cronin, 1984, 2000; Franck, 1999). thorship status: practices vary greatly, within and across But while the traditional model of authorship persists, most disciplines, ranging from alphabetization through weighted noticeably in the humanities, it is no longer the sole or listing to reverse seniority (e.g., Spiegel & Keith-Spiegel, dominant model in certain scientific specialties. 1970; Riesenberg & Lundberg, 1990). Certainly, much has Before the precursors of today’s scholarly journals es- changed since the 17th century, when there was no estab- tablished themselves in the second half of the 17th century, lished convention for the naming of authors of scientific scientists communicated via letters. Letter writing was the communications. As Katzen (1980, p. 191) notes in her principal means for exchanging ideas and experimental re- analysis of early volumes of the Philosophical Transac- sults (Kronick, 2001). Gentleman scholars, including such tions: prodigious correspondents as Samuel Hartlib, intelligencer extraordinaire, Henry Oldenberg, secretary of the Royal . . . no attempt is made to give prominence to the author of Society in London, or Friar Mersenne in France, were the the article . . . there is generally no reference at all to the scientific community’s prototypical gatekeepers, orchestrat- author in the heading that signals a new communication. If ing the flow and documentation of ideas on behalf of their the author is referred to in the title, it is likely to be in an coeval peers across Europe (Knight, 1976; Rayward, 1992). oblique form . . . we are at the threshold between anony- The establishment, in France, in 1665, of Le Journal des mous and eponymous authorship. Sc¸avans and, some months later, in London, of the Philo- sophical Transactions of the Royal Society, or, as it was Foucault (1977, p. 126), in fact, talks of scientific texts originally called, the Philosophical Transactions: Giving “accepted on their own merits and positioned within an some Accompt of the present Undertakings, Studies and anonymous and coherent conceptual system of established labours of the Ingenious in many considerable parts of the truths and methods of verification.” This, as Shapin (1995, World, constituted the beginnings of the journal-based p. 178) notes in his brilliant study of trust in 17th-century scholarly communication system as we know it today. English science, was in keeping with prevailing notions of Nonetheless, letter writing continued as a medium for the civility and gentlemanly conduct in society-at-large: “reluc- informal exchange of information and for requesting fellow tant authorship . . . was a standard trope of early modern scientists to replicate experiments (Manten, 1980, p. 8). culture,” albeit one which withered swiftly in the face of Despite the acknowledged importance of these two con- plagiaristic practices. Inconceivable though the idea of temporaneous publishing developments, much has changed anonymous authorship may be in today’s highly competi- in terms of the ways, both instrumental and stylistic, in tive publishing marketplace, where reputation, career suc- which scientists communicate the results of their research to cess, and, ultimately, remuneration are tightly coupled with their peer communities. However, in the intervening 300- publication salience and citation (Cronin, 1996), the impor- plus years, certain symbolic and rhetorical practices, nota- tance of claim-staking and priority determination were re- bly the assertion and defense of authorship, and all the ally only beginning to be perceived as critical issues in the presumptive rights associated therewith, have remained formative world of 17th-century scientific communication. center stage. In the 17th century, the business of authorship, The standard model of scholarly publishing assumes a as the business of science itself, was much less complicated work written by an author. There is typically a single author and contentious than today—which is not to say that priority who receives full credit for the opus in question. By the disputes were unheard of, that egos were never bruised, or same token, the named author is held accountable for all that “the bauble fame” did not come into play in earlier claims made in the text, excluding those attributed to others times. The following excerpt from one of Charles Darwin’s via citations. The appropriation of credit and allocation of (admittedly much later) letters makes touchingly clear the responsibility thus go hand-in-hand, which makes for fairly recurrent tension between intrinsic and extrinsic motivation, straightforward social accounting. The ethically informed, which he, and countless other scientists, both then and now, lone scholar has long been a popular figure, in both fact and have experienced in the course of their careers: “I am got scholarly mythology. Historically, authorship has been most deeply interested in my subject; though I wish I could viewed as a solitary profession, such that “when we picture set less value on the bauble fame, either present or posthu- writing we see a solitary writer” (Brodkey, 1987, p. 55). But mous, than I do, but not to any extreme degree; yet, if I that model, as Price (1963) recognized almost three decades know myself, I would work just as hard, though with less ago, is anachronistic as far as the great majority of contem- gusto, if I knew that my Book wd be published for ever porary scientific, and much social scientific and humanistic, anonymously.” (Darwin, 1919, p. 452) publishing is concerned. This view is echoed by Poster

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 559 (1990, p. 114), albeit from a somewhat different perspec- ship, is directly linked to the professionalization of science. tive: “computer writing subverts the author as centered Beaver and Rosen (1978) have shown how the differential subject in yet another way, by introducing new possibilities rates of scientific institutionalization in France, England, of collective authorship.” In general terms, the lone author and Germany are mirrored in the relative output of coau- stereotype ignores the fact that a great deal of the scholarly thored papers. At the end of the 18th and beginning of the literature is the product of a “socio-technical production and 19th centuries, French science was much more profession- communications network” (Kling, McKim, Fortuna, & alized and institutionalized than was the case in either of the King, 1999), which brings together a mix of actors, re- other European powers. Specifically, they found that more sources, and rules. To some extent, authorship has become than half of all the coauthored scientific articles in their a collective activity, with numerous coauthors competing historic sample had been produced by French scientists. for the byline, some of whom may not have written, strictly Their study is a useful corrective to the assumption that speaking, a single word of the associated work (McDonald, collaboration and coauthorship were unheard of prior to the 1995; Kassirer & Angell, 1991). Incidentally, parallel, emergence of ‘big science’. In a sample of 2,101 scientific though not quite so dramatic, growth has been observed in papers published between 1665 and 1800, Beaver and the number of individuals being formally acknowledged in Rosen found that 2.2% described collaborative work. No- scholarly journals for their multifarious contributions: what table was the degree of joint authorship in astronomy, has become known as subauthorship collaboration (Patel, especially in situations where scientists were dependent 1973; Heffner, 1979, 1981; Cronin, 1995, Cronin, upon observational data. 2001)—an important, if underappreciated, indicator of in- Since the early days of professional science, there has formal scientific collaboration. been, in extreme cases, an order of magnitude increase in coauthorship levels, with the most dramatic growth occur- ring in the last couple of decades. The trend, though most Collaboration and the Conditions of Work visible in science, has been documented in many fields and If the 17th century was a watershed in the history of disciplines, which is not to say that sole authorship, both of scholarly publishing, the same may be said of the mid-20th journal articles and monographs, is not still common prac- century. After World War II, collaboration became a defin- tice in fields such as philosophy or women’s studies (Cro- ing feature of ‘big science’ (Bordons & Gomez, 2000; nin, Davenport, & Martinson, 1997). By way of example, Cronin, 1995, pp. 4–13; Katz & Martin, 1997). Major Endersby (1996) has analyzed trends in, and reasons for, scientific challenges (splitting the atom, putting a man on collaboration and multiple authorship in the social sciences. the moon, mapping the human genome) typically require Patel (1973) has described the growth of coauthorship in enormous levels of funding to support the costs of industri- sociological journals for the period 1895 to 1965. Bird al-scale equipment and instrumentation, as well as complex (1997) has found evidence of coauthorship growth in the teams of researchers drawn from multiple disciplines and literature of marine mammal science (1985–1993), while institutions. In some domains, path-breaking work is nec- Koehler et al. (1999) found that the average number of essarily the outcome of collaborative activity rather than authors per article in the Journal of the American Society for individualistic scholarship, a fact reflected in the modest Information Science (previously American Documentation) proportion of federal research funds which is allocated to rose from approximately 1.2 in the 1950s to 1.8 in the individual investigators rather than teams. Collaborations 1990s. are a necessary feature of much, though by no means all, Data produced by the Institute for Scientific Information contemporary scientific research. (ISI) show, inter alia, that the number of papers with 100 The nature and conduct of scientific research have authors—a leading indicator of hyperauthorship?—in- changed enormously since the 17th century. One well- creased from 1 in 1981 to 182 in 1994 (McDonald, 1995) documented corollary of interinstitutional, intersectoral, and and that the average number of authors per paper in the transnational collaboration has been the striking increase in Science Citation Index (SCI) increased from 1.83 in 1955 to rates of coauthorship, though the latter is only a partial 3.9 in 1999 (personal communication with Helen Atkins, indicator of the former: coauthorship and collaboration are Director of Database Development, Institute for Scientific not coextensive (Katz & Martin, 1997, p. 1). This trend is Information, Philadelphia, 2000). To use a couple of ran- most noticeable in experimental high-energy physics dom examples: a 1997 article in Nature (cited almost 600 (HEP), with its often very large teams and highly sophisti- times since then) on the genome sequence of a bacterium cated collaborations (Kling & McKim, 2000). A similar has 151 coauthors, drawn from dozens of research labora- trend, dating from the 1990s, can be seen in the biomedical tories scattered across twelve countries (Kunst et al., 1997). research literature, particularly with regard to publications A recent (Daily et al., 2000). two-page article in Science on arising from large, multi-institutional clinical trials (Rennie, the economic value of ecosystems has no fewer than 17 Yank, & Emanuel, 1997; Horton, 1998). However, it would authors and five acknowledgees. Illustrative of the trend in be erroneous to conclude that collaboration and coauthor- biomedicine is the appreciable change exhibited by the New ship are exclusively late 20th-century phenomena. The England Journal of Medicine (NEJM). A hundred years growth of scientific collaboration, as reflected in coauthor- ago, 98% of the articles published in the (precursor of the)

560 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 NJEM were sole authored; today the figure is less than 5%, proto-authors working under the direction of the master with the mean number of authors per NEJM article being six craftsman, or super PI (principal investigator). This is (Constantian, 1999). A study of all clinical, anatomic, and scarcely novel: Shapin’s (1995, p. 367) detailed account of laboratory investigations reported in the Journal of Neuro- Robert Boyle’s scientific milieu includes a compelling im- surgery and Neurosurgery between 1945 and 1995 found age of “the seventeenth-century laboratory as collective that the average number of authors per scientific article rose workshop.” As a result, the author’s persona, insofar as the from 1.8 to 4.6 (King, 2000). impersonal and highly stylized format of the scientific arti- In many scientific fields, multiple rather than single cle permits the extrusion of an authorial self, has been authorship is now the norm (Harsanyi, 1993), which brings dislodged by the depersonalized voice of the ensemble. to mind Newman’s (1996, p. 1) apt phrase, “writing together Such practices, and associated concerns about “unearned separately.” What is not entirely clear is the extent to which, authority” (Kitcher, 1993, p. 315) and dilution of responsi- if at all, publication genre affects rates of coauthorship. In bility, have provoked radical proposals, notably in the bio- other words, will the picture vary depending on whether the medical research community—where multiple authorship unit of analysis is clinical reports, scientific articles, or inflates to become hyperauthorship—to restore authorial review papers? I return to this issue later. In short, collab- accountability and rationalize the allocation of academic orative research and publishing practices in the late 20th credits. Some medical commentators (Rennie, Yank, & century have created challenges for established notions of Emanuel, 1997, p. 582) have advocated abandoning the authorship. Ineluctably, the recognizable voice of the indi- concept of author altogether in favor of ‘contributors’ and vidual author is being replaced by the sometimes pasteur- ‘guarantors,’ thereby freeing us “from the historical and ized prose of the collaboration, reinforcing the well-estab- emotional connotations of authorship.” The potential sub- lished “conventions of impersonality” in scientific writing version of the historical conception of authorship, with, at (Hyland, 1999, p. 355) and resulting in the erasure of style. the risk of over-simplifying, its assumption of singularity, Long before the modern practice of anonymous (double- by contemporary scientific practice has created a number of blind) peer review became an established component of the ethical and procedural difficulties for those, such as univer- post-war scientific bureaucracy (Chubin & Hackett, 1990, sity promotion and tenure committees, responsible for eval- pp. 19–24), experimental claims were subjected to personal uating the nature of individual contributions and also for witness and supported by sworn testimony (Gross, Harmon, those concerned with quality assurance in complex, multi- & Reidy, 2000) either by members of scientific societies or institutional research projects. In biomedicine, the standard others deemed to be of appropriate social standing—a pro- model (one text, one author) faces a determined challenge cess which Shapin (1995, p. xxi) refers to as “the gentle- from the radical model (lists of contributors and their spe- manly constitution of scientific truth.” Bazerman (1988, p. cific inputs), championed most vigorously in recent years by 140), in describing the process of making witness in the Drummond Rennie, a deputy editor of the Journal of the early days of the Royal Society, notes that one rhetorical American Medical Association (JAMA). strategy is “to establish ethos,” that is, convey to a critical audience, that the author or observer of the experiment Assuming Authorship being described is a credible witness. Just as Princess En- heduanna “enabled the reader to read a text in a given Given the standard model of scholarly publishing, in voice” by signing her name to her songs (Manguel, 1997, p. which it is assumed that a work is written by an author, the 182), so might Fellows of the Royal Society seek to create central issue of author attribution can be framed in terms of a culturally acceptable authorial persona, or ethos, to engage Rennie’s and Flanagin’s (1994, p. 469) beguilingly simple and persuade their audience. But circumstances have question: “. . . how many people can wield one pen?” Stat- changed such that today the individual author’s voice is ing this does not mean that the question of multiple author- almost imperceptible. It is difficult to attribute a persona to ship has not received some serious attention. For example, an ensemble of putative authors/collaborators, or even library cataloging, in the form of the Anglo-American Cat- imagine how such a confederacy might establish an ethos, aloging Rules (1998), has provisions for two distinct kinds though such lengthy lists of authors, doubtless, add to the of situations: one in which the wording or typography of a perceived credibility and authority of the text. title page indicates that two or more authors among a larger In some contexts, individualistic notions of authorship number are chiefly responsible for the content of a work (“the fiction, the voice of a single person” [Barthes, 1977, p. (rule 21.6B), and a second in which two or more authors are 143]) have given way to the idea of the author as a collec- listed as responsible for a work with none indicated as being tivity, whose members are often geographically distributed, chiefly responsible for the content (rule 21.6C). In either and, not infrequently, drawn from quite different profes- case, however, when the number of authors considered sions and specialties. Duncan (1999), in fact, uses the anal- chiefly responsible (case 1) or the total number of equally ogy of a partnership with contractual ties to capture the responsible authors (case 2) reaches four or more, the code complexities of the contemporary research and publication reverts to a default judgment about responsibility—that is, system. Another analogy, from the worlds of painting and that authorship is too diffuse to assign chief responsibility. architecture, might be the (virtual) atelier, with a team of In the latter situation, the work’s main listing in the catalog

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 561 is under its title with nominal access given to it under work to which I affix my name, and, by extension, my whoever is the first author listed on the title page. professional reputation. The foregoing library cataloging provisions apply chiefly Unquestionably, the practice of promiscuous coauthor- to books, of course. In the realm of periodical publications, ship puts considerable stress on this tried and tested model. the sovereignty of the standard model is being most hotly In the case of an article with, say, two or three coauthors, contested in biomedical research fields, where intense levels one of whom is the designated corresponding author (to of professional collaboration and coauthorship are common- whom editors and readers channel their queries, comments, place (Croll, 1984; Rennie & Flanagin, 1994; Rennie, Yank, and concerns), this should not be a serious problem, as the & Emanuel, 1997; Rennie & Yank, 1998; King, 2000). specific contributions of each individual presumably can be Proposals for reform, which seek to retire the concept of clearly described to the satisfaction of all parties. This is authorship and replace it with a scheme for the allocation of similar to the recognition in library cataloging and referred specific, task- or job-related credits (e.g., Squires, 1996; to earlier of ‘principal responsibility’ being indicated on a Smith, 1997) are not only being debated by editors and book title page in terms of special wording or typography, others, but are being adopted by leading scientific journals. and to the recognition of different contributions to a work. To understand why such radical thinking would attract However, if the number of listed coworkers is counted in serious attention in the medical establishment, we need to dozens, scores, or even hundreds, as is increasingly com- consider how multiple authorship, in extremis—what I have mon, for instance, in the case of large-scale clinical trials, chosen to term ‘hyperauthorship’—undermines commonly who, then, is to be held accountable for both the whole and held assumptions about the nature and ethical entailments of its constituent parts, should questions about the integrity of authorship, and how, in exceptional cases, it can lead to the work arise? And who, remote though the possibility may fundamental questions about the integrity of the research be, will act as spokesperson in the face of a liability suit community as a whole. Unfortunately, little effort is made brought against the multidisciplinary collaboration? Like- in the biomedical literature to distinguish systematically wise, how should credit for crafting a single publication be between what might be termed acceptable levels of multiple allocated across teams comprising dozens or scores of co- authorship and unacceptable levels of hyperauthorship. workers, whose membership may have altered over the Studies of coauthorship trends, as will become clear, are course of a multiyear project? Where, ultimately, do author- inconsistent in defining and operationalizing the phenomena ity, credit, and accountability reside? In the words of Rennie under consideration, such that discussion of multiple au- and Yank (1998, p. 829), when “the number of collaborators thorship bleeds into condemnation of hyperauthorship. grows arithmetically, it becomes exponentially harder to Under the standard model, the rights and responsibilities affix responsibility.” At this point, the notion of authorship, of authorship are clearly apprehended by all parties: authors, literally interpreted, is effectively rendered meaningless. editors, referees, and readers. In appending my name to this As in biomedicine, these issues have also attracted at- article I am nailing my colors to mast; if the article attracts tention in the high-energy physics research community, critical approval, is discussed, quoted, and, in due course, where hyperauthorship is commonplace. However, multiple cited in the scholarly literature, I shall be happy to bank the authorship and hyperauthorship are not problematized by symbolic capital which accrues to me as author and origi- physicists as they are by the biomedical community. Com- nator. If the paper is challenged because of exiguity of mon to both domains has been discussion of how best to theoretical, historical, or empirical heft, I shall simply have apportion credit when there are many coauthors (Mc- to face the music: there are no coauthors to help deflect Donald, 1995). How, for instance, should a promotion and criticism. Likewise, if I am challenged for drawing too tenure committee view the contribution of the 99th listed sparingly, selectively, or generously on the ideas and work author on a particle physics paper or the 36th author on a of others, I understand the possible consequences. However, genome sequencing study? What may seem to constitute a I have chosen not to hide behind the cloak of anonymity, or miniscule portion of a single journal article may, in fact, bypass the rigors of peer review by posting a version of this have consumed a significant amount of that individual’s paper on my Web site; rather, I want to publicize my ideas professional time and energy. Two other issues, honorific among my peers, and the best way to do that, and signify my authorship and data integrity, seem to be of especial concern trustworthiness, is to pursue publication in an accredited to the biomedical community, given widespread media cov- forum. As a serial author, I am fully cognizant of the rights erage of, and speculation about, fraudulent practice, the and responsibilities of authorship. I understand the norms of effects of which, in both career and personal terms, can be scholarly publishing, and I am aware of the sanctions that devastating (e.g., Kevles, 1998). What this seems to suggest may be invoked if infractions occur. Should the arguments is that size and scale of collaboration alone do not neces- in this paper prove flawed, no one but myself is to blame, sarily engender concerns about the veracity and reliability and that includes those whom I have named in the acknowl- of the reported research. It is likely that the sociocognitive edgments section. If the paper attracts attention, I shall be structures of these two research cultures, and their relative happy to bask in the glow. Just as a lawyer, doctor, or transparency, differ in important respects. As a result, issues dentist would, I implicitly accept full responsibility for the of social trust are manifestly more problematic in the dis-

562 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 course of the biomedical community. I return to the matter Journal of Medicine, and American Journal of Cardiology, of field-dependence and generalizability later. American Journal of Medicine, American Journal of Ob- stetrics and Gynecology, respectively), Flanagin et al. (1998) developed a multivariate logistic regression model to Multiple Authorship in Biomedicine test the hypothesis that coauthored articles (operationalized Many studies have documented the recent, dramatic in- as papers with six or more authors) were increasing at a rate crease in multiple authorship in medical and related fields, greater than would be expected when confounding vari- in turn generating a mass of editorial commentary and ables, such as the number of centers, were taken into ac- correspondence in the letters pages of major journals (e.g., count. They found that 19% of original research reports had Constantian, 1999; Rennie, Yank, & Emanuel, 1997). How- honorific authors, individuals who were garnering phantom ever, as noted earlier, it is not always easy to establish what fodder for their curricula vitae. They also discovered that constitutes unacceptable levels of multiple authorship in the 11% of articles had ghost authors, which means that quite a minds of the many concerned commentators, be they au- few individuals were not receiving due credit for their thors or editors. Individual research and sole author publi- creative or material contributions to the research process— cation are almost inconceivable in certain biomedical spe- “the ghostly inferred hosts of unnamed actors who shifted cialties, where complex (national and international) net- instruments about and exerted their muscular labor’ works of resources, facilities, technical expertise, and (Shapin, 1995, p. 379). Their findings, based on surveys of subjects are the sine qua non of serious, funded research. corresponding authors, are in keeping with other estimates Collaboration and coauthorship are inescapable conse- of honorific authorship in the biomedical literature. quences of increasing specialization and technical sophisti- Slone (1996), in a survey of “major research” articles cation. In short, structural interdependence is a fact of life in published in the American Journal of Roentgenology, found the age of industrialized medical research, and the phenom- that undeserved coauthorship rose from 9% on papers with enon is powerfully revealed in the lengthy lists of authors three coauthors to 30% on papers with more than six coau- and acknowledgees that adorn the majority of published thors. Drenth (1998) found that the mean number of authors articles and reports. As far as biomedical research is con- per article in the British Medical Journal rose from 3.2 in cerned, it seems that the lone scholar/independent re- 1975 to 4.7 in 1995. A number of publication genres feature searcher is now on the endangered species list. in these and other studies. For example, King (2000) exam- To be sure, coauthorship in itself is not inherently prob- ined research articles; Kahn, Nwosu, Kahn, Dwarakanath, lematic, but some of the hidden social practices which it has & Chien (1999) examined two kinds of articles, randomized occasioned, such as receiving credit under false pretenses, controlled trials and controlled observational studies; Slone are cause for grave concern (Anderson, 1991). Even a (1996) restricted his focus to major papers; Drenth (1998) cursory examination of the mainstream biomedical litera- examined original articles; and Flanagin et al. (1998) inves- ture over the last decade or two makes this fact abundantly tigated research articles, review articles, and editorials. It is clear. In particular, the abuses implied by the phrase ‘hon- not clear from the various analyses whether, or to what orific authorship’ are damaging the perceived credibility extent, genre and coauthorship levels are linked. and integrity of the medical communication system as a Further confirmation that inflationary trends in author- whole, a development which has been exacerbated by the ship are not explicable purely in terms of intercenter col- number of high-profile cases of actual or alleged scientific laboration, research funding and related factors has been fraud in recent years (e.g., Relman, 1983; Cohen, 1999; provided recently by Kahn, Nwosu, Kahn, Dwarakanath, & Eysenck, 1999). Richard Horton (1998, p. 688) editor of Chien (1999). Additionally, Ducor (2000) investigated a The Lancet, speaks, in fact, of “the shattered system of small set of patents in molecular biology and their concom- academic reward and its symptom, broken rules of author- itant publications in the scientific literature. Of the 40 ship,” a view which seems neither extreme nor marginal patent-article pairs examined, all but two listed more au- judging by the tenor of the debate being conducted in the thors than inventors, which raises interesting questions pages of the global biomedical literature (Klein & Moser- about the relative stringency of the criteria employed for Veillon, 1999). conferring authorship and inventorship. Although the increase in coauthorship is a function of Such studies reveal the apparent ease with which autho- scale and specialization effects in medical and related re- rial spurs can be earned. However, as Slone (1996, p. 576) search, the picture is more complicated still. There is com- notes, there is what might, at first blush, appear to be a fairly pelling evidence that many individuals receive unwarranted simple quid pro quo in operation—bylines for bodies: “The coauthor status (variously referred to as ‘guest’, ‘gift,’ or number of coauthors in multiinstitutional clinical trials is ‘surprise’ authorship) while others are denied legitimately understandable, and it is tempting in complex studies to earned author status (‘ghost’ authorship). As Slone (1996, p. offer authorship to referring physicians and ancillary med- 578) notes, authorship “cannot be conferred but must be ical specialists to obtain the needed cooperation.” These and earned.” In a survey of three large- and three smaller- kindred practices may well help explain the startlingly high circulation medical journals (Annals of Internal Medicine, and sustained rate of scholarly production demonstrated by Journal of the American Medical Association, New England some biomedical researchers (Anderson, 1992). Whether

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 563 coat-tailing on one’s peers and rent-taking are dismissed as known as the Vancouver Group. But before considering the understandable peccadilloes or seen as a corroding of the ICMJE’s guidelines for authorship, it may be instructive to norms and institutional values which undergird the social reflect on the role of acknowledgments in the primary system of science (Merton, 1973), the practical conse- communication process, and the relationship between au- quences are ramified in cases of fraud. In such instances, thorship and acknowledgment. readers and editors may find that nominal authors adroitly The explosion of coauthorship naturally raises the ques- disclaim responsibility, deny accountability, and proffer a tion why authorial surplus could not be accommodated in variety of excuses to ensure that culpability falls elsewhere. the acknowledgment sections that accompany the great ma- In rare cases, a questionable, published paper may acquire jority of scientific articles. The acknowledgment serves as a “orphan” status (Rennie & Flanagin, 1994, p. 469), as all parking lot for miscellaneous contributions, cognitive, tech- concerned try to wash their hands of it, invoking hyperlabor nical and social, which typically fall short of meeting com- specialization as grounds for exoneration. Such a scenario is monly understood criteria for awarding coauthorship. But inconceivable under the standard model, where authorship we should not assume too much in terms of common un- and accountability are isomorphic. But when authorship/ derstandings: the dividing line between the two classifica- ownership of a study is distributed across multiple contrib- tions, author and acknowledgee, is neither universally ap- utors, many of whom may have zero or weak relation- preciated nor consistently applied. Cronin (1995, pp. 85– ships—whether personal or institutional—with their myriad 86), for instance, has shown that interpretative disputes are coworkers (Katz & Martin, 1997), the practical (i.e., en- not uncommon, and that some researchers feel that they forceable) allocation of accountability may pose intractable have been denied their just deserts by being downgraded problems. from coauthor to acknowledgee. Ambiguities and fuzzy interpretations notwithstanding, the ubiquity and signifi- cance of “sub-authorship collaboration” (Patel, 1973, p. 81), Authorship and Acknowledgment otherwise conceptualized as “trusted assessorship” (Mul- Historically, authorship implied writing and this associ- lins, 1973, p. 32), has been amply documented in recent ation with the act of writing remains the core of the standard years. Bazerman (1984, 1988) has chronicled the evolution model of authorship acknowledgment. However, today, the of the acknowledgment during the 19th and 20th centuries concept of authorship has expanded to accommodate a in the journal literature of experimental physics, showing diverse array of contributions and inputs, some of which how it became, to paraphrase Grafton (1997, p. 233), an may require little or no engagement with the text qua text. integral part of the rhetoric of narration and annotation. This broadening of definition finds its echo in the world of Others have recorded the social significance of acknowledg- cataloging and metadata, where schemes such as the Dublin ment practices in a variety of disciplines, including astron- Core and REACH (Record Export for Art and Cultural omy (Verner, 1993), genetics (McCain, 1991), biology Heritage) prefer the terms ‘creator’ and ‘maker’ to the more (Heffner, 1981), chemistry (Heffner, 1981), psychology traditional ‘author’ (Baca, 1998). The Dublin Core Meta- (Heffner, 1981; Cronin, 1995), information science (Cronin, data Initiative (see: http://purl.org/dc/documents/rec-dces- 1995, 2001), sociology (Patel, 1973; Cronin, 1995), politi- 19990702.htm) includes the element ‘Contributor’ as some- cal science (Heffner, 1981), and philosophy (Cronin, 1995). one who is defined as being “responsible for making con- It is abundantly clear from the findings of these studies that tributions to the content of the resource.” In addition, there acknowledgment is a well-established and valuable feature is the element ‘Other Contributor’ which can accommodate of the apparatus of 20th-century scholarly communication, a “person or organization not specified in a Creator element one which can shed additional light on “institutional depen- who has made significant intellectual contributions to the dencies” (Bazerman, 1984, p. 183) not fully reflected in the resource but whose contribution is secondary . . .” It could article’s byline. be argued, of course, that this change is not intended as a At first glance, the idea of using acknowledgments as a diminution of the idea of writing, but rather has been form of triage leading to “two-tiered authorship” (Saffran, necessitated by the ever-increasing numbers of nontextual 1989, p. 9) would seem to make sense, assuming, for the informational objects found on the Web (e.g., graphics, or sake of argument, that consensus could be reached on cri- graphical representations of 3-D objects) for which the term teria for allocating coworkers a byline as opposed to ‘mere’ ‘author’ is inappropriate. With respect to more traditional acknowledgment status. But things are not quite so simple. textual entities, however, the new inclusiveness defines au- Almost a decade ago, Kassirer and Angell (1991, p. 1511) thorship of a scientific article as requiring “expertise in its of the New England Journal of Medicine were bemoaning content and thorough knowledge of the investigation re- not only “ambiguous authorship” but “lengthy acknowledg- ported” (Gaeta, 1999, p. 297). Yet, under this not unreason- ments,” inflated by the inclusion of “everybody who had able-sounding definition, the author need not write a word. anything to do with the study, including those who were Within the biomedical community, considerable effort has merely carrying out their jobs, such as technicians.” Their been invested in developing guidelines and criteria for au- radical proposal at that time was to limit the amount of thorship, notably those adopted by the International Com- journal space devoted to acknowledgments, with the excess mittee of Medical Journal Editors (ICMJE, 1997), also names being deposited with the National Auxiliary Publi-

564 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 cations Service—scholarly publishing’s equivalent of the usually adhered to by their peers. What these studies suggest salon des refuse´s. Some journals place a limit on the number is that even the best-intentioned guidelines for authors, such of coauthors; for example, the Dutch Journal of Medicine as those specified by the Journal of the American Medical does not publish articles with more than six authors (Hoen Association (see: http://jama.ama-assn.org/info/auinst.html), et al., 1998). The problem with arbitrary capping, whether will be variously and inconsistently interpreted. Absent of authors or acknowledgees of one kind or another, is that unrealistic, and presumably unacceptable, levels of policing some individuals’ potentially important contributions, be by editors and their advisory boards, it seems unlikely that they clinical investigators (Carbone, 1992) or telescope illicit authorship and cronyism will be extirpated as a result operators (Cronin, 1995), may be erased. This could, con- of such guidelines. ceivably, have negative downstream implications in terms The standard model accepts that authorship is linked of remuneration and promotion prospects for those—the inextricably to writing. But writing is no longer a necessary “invisible technicians” (Shapin, 1995, p. 355)—whose ef- condition of coauthorship in certain cases. Thus, an alter- forts have been withheld from the public ledger. It might native to authorship is required to accommodate the many also reduce, in line with theories of reciprocal altruism other contributions that shape the published byproducts of (Nowak & Sigmund, 2000), potential collaborators’ will- collaborative activity, be they research reports, journal ar- ingness to ‘donate’ their services. ticles, conference papers, or technical reports. Comparisons Overpopulation is now a feature of both author lists and have been made (Rennie, Yank, & Emanuel, 1997) with the acknowledgment statements, and, as already noted, the pic- credit sequences at the end of movies, in which all those ture is muddied by the lack of generally accepted guidelines responsible, from key grip through stunt men to director, are for allocating credits. Confusion over the criteria for award- thanked publicly for their services. Cronin (1991) has anal- ing authorship as opposed to acknowledging colleagues— ogized in similar fashion, and entitled an early paper on and acknowledgment, it should be noted, is often subdi- acknowledgments, ‘Let the credits roll . . .’ However, it is vided into intellectual and technical, which further compli- important to distinguish between generic job categories and cates the calculus—has convinced Rennie, Yank, and the specification of tasks performed; the contributorship Emanuel (1997) that the distinction between the two modes model is designed to record each individual’s actual input of credit allocation is inherently artificial. Consequently, (e.g., experimental design, data collection, statistical analy- they have argued for explicit description of all individual sis, final article revision), not job title (e.g., coprincipal contributions as a means of eliminating ambiguity. Such a investigator, technician, systems analyst), since the latter proposal would remove both authorship and acknowledg- may on occasion mask or inflate the former (Stern, 2000). ment from the frame, a really quite significant break with The general case for replacing authors with lists of con- scholarly publishing tradition. This alternative amounts to a tributors and guarantors has been most cogently detailed by radical model of authorship attribution in contrast to the Rennie, Yank, and Emanuel (1997) in the pages of the standard model. Journal of the American Medical Association (JAMA). Their paper systematically addressed the pros and cons of the radical model. In this scheme of things, project cowork- Assessing Authorship ers would discuss and agree upon the distribution of the To date, more than 500 journals have adopted the various responsibilities and tasks at the outset, assess the ICMJE’s (1997) principles of authorship as laid out in the relative merit of their contributions, and determine the im- 5th edition of the Uniform Requirements for Manuscripts plications for listing collaborators’ names, so that first au- Submitted to Biomedical Journals (Klein, 1999; Stern, thorship would be based on earned credit rather than alpha- 2000). According to these concrete guidelines, candidate betic advantage or professional status. Appropriately, the authors must satisfy three conditions. They must make: final paragraph of their JAMA manifesto is itself a detailed “. . . substantial contributions to (a) conception and design, itemizing of the specific contributions of each coauthor, or analysis and interpretation of data; and to (b) drafting the such that the reader is left in no doubt as to who did article or revising it critically for important intellectual precisely what. The great attractions of the scheme are the content; and on (c) final approval of the version to be implications for transparency and equity, but there is also no published.” Laudable though these guidelines are, it is un- doubt that serious-minded application of the contributorship likely that they will solve the problem. A study by Hoen et model would impose additional overhead on medical re- al. (1998) in the Netherlands found that authors and their searchers, and team leaders in particular. This would cer- coauthors did not always agree with one another’s assess- tainly be true if individual contributions in a large multi- ments that the ICMJE criteria had been met. In the UK, center trial were to be calibrated with the degree of mean- Bhopal et al.’s (1997) survey of medical researchers dis- ingful precision described by Ahmed, Maurana, Engle, covered that, although most respondents concurred with the Uddin, & Glaus (1997). three criteria (more than 80% in each case), a majority Their authorship matrix (which they apply reflexively to (62%) did not feel that all three conditions should have to be their own article) combines seven key tasks associated with satisfied to warrant author status. Further, more than half of multiauthored papers (e.g., conception, design, data analy- the respondents believed that the ICMJE’s criteria were not sis), with each investigator’s contribution(s) on each dimen-

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 565 sion duly ranked on a three-point scale (minimal, some, arena is collaboration so spectacularly institutionalized as in significant). The respective weights (1, 3, 5) are summed to experimental high-energy physics. In HEP, the massive (a) estimate each individual’s overall contribution, which scale and cost of conducting path-breaking research neces- can range from 1–35 (7ϫ5 being the maximum possible sitate resource concentration around a number of major score), and (b) determine the sequence in which coauthors’ laboratories (e.g., CERN, Fermilab, SLAC), at which large names are listed on the resultant publication. The logic (distributed) collaborations are instituted to tackle basic behind the algorithmic approach is impeccable, but its broad research problems, typically requiring multiyear funding practicability remains unproven. Other not entirely unre- (Traweek, 1992). Such highly complex, distributed socio- lated efforts to increase transparency in the scientific com- technical ensembles (Kling & McKim, 2000) can bring munication system have proved unsuccessful. For instance, together literally hundreds of physicists from countries, it was suggested in the 1980s that authors should classify laboratories, and universities worldwide, who compete di- the citations attached to their papers in terms of the func- rectly with collaborations centered elsewhere. These multi- tions they performed, but the idea failed to get off the disciplinary teams routinely produce publications which ground, despite some early interest from the Institute for have numerous coauthors, just as in the case of the biomed- Scientific Information in Philadelphia (Duncan et al., 1981; ical research community (Newman, 2000). Yet, despite Cronin, 1984). In the early 1990s, skepticism also greeted some concerns about proliferating HEP authorship rates, the proposal that an online acknowledgment index, akin to and occasional discussion of hyperauthorship in the higher the Science Citation Index, be developed to record subau- education press (e.g., McDonald, 1995), the kind of soul- thorship collaboration in science (Cronin & Weaver- searching and public debate on authorial integrity that have Wozniak, 1992). characterized the biomedical community in recent years are While listing contributions may clarify the nature of not mirrored in the world of high-energy physics. Why coworkers’ participation and, thus, both reduce the inci- should this be? dence of honorific authorship and ensure more equitable The answer probably has to do with the relative intensity allocation of credit, it does not necessarily address the of socialization and oral communication (Traweek, 1992, thorny issue of ultimate responsibility for the overall integ- pp. 120–123), along with the character of the organizational rity of the study. Consequently, Rennie, Yank, and Emanuel structures and value systems, which define collaborations in (1997) have also proposed that special contributors be des- large-scale, high-energy physics and biomedical research. ignated who would act as owners, or guarantors, of the work Social coordination theory provides a potentially rich ana- as a whole. These individuals would be expected to assure lytic framework for understanding cooperative acts. Inter- the integrity of the work and be able to defend the research’s actants—networks of coworkers, coauthors, contributors, findings in the event of peer challenge or public criticism. In and acknowledgees in the present context—must satisfy a the matrix adumbrated above (Ahmed, Maurana, Engle, number of conditions, such as establishing co-presence and Uddin, & Glaus, 1997, p. 43), the dimension ‘public respon- sibility’ is included to signify that an author participated in, revealing mutual responsiveness, for “a cooperative act to and is accountable for, the work and related results. Despite unfold” (Fine, 1993, p. 72; Couch, 1984). Ethnographic continuing discussion of the merits and shortcomings of the approaches could be used to explore a variety of research radical model (e.g., Should acknowledgments and contrib- and publishing subcultures and document the ways in which utor lists be combined?; Is it reasonable to expect any social relationships and group behaviors mold material individual to be able to account for all aspects of a complex practice in these two domains. Second, within HEP collab- multidisciplinary collaboration?), some editors and their orations, there is a powerful admixture of “agonistic eval- journals (e.g., The Lancet, British Medical Journal) are uation of other physicists” (Traweek, 1992, p. 117) and very supportive of the idea of contributor lists (Horton, 1998). explicit, transparency-inducing procedures for vetting work internally. This point is stressed by Kling, McKim, Fortuna, & King (1999) in their study of interaction networks: A Local Issue? “. . . concern for professional reputation both individually Are the developments in biomedical research and publi- and for the collaboration causes most, if not all, collabora- cation sketched above peculiar to that domain, or do they tions to have developed systems whereby only research find echo in other fields? To put it another way, and to results that have been “blessed” by the collaboration may be paraphrase Mark Twain, are the rumors of the death of shared with the world on the collaboration web site.” The authorship greatly exaggerated? Although increasing rates HEP research community is thus characterized by high of multiple authorship have attracted attention in a number levels of internal scrutiny, mutual trust—witness, for in- of fields, such as psychology (e.g., Bartle, Fink, & Hayes, stance, the institutionalized practice of relying upon, and 2000), only in biomedicine has the research community felt citing, preprints—and peer tracking, such that it is not a need for collective, aggressive action to deal with the susceptible to systematic fraud. Contrary to what might be excesses and documented abuses. Why, then, does the same thought, contemporary scientific collaboration need not nec- not hold true for high-energy physics, the other acknowl- essarily imply impersonality and virtual anonymity, a point edged locus of hyperauthorship? In perhaps no scientific made tellingly by Shapin (1995, pp. 414–415):

566 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 While outsiders—including many philosophers and sociol- Changes in the conditions of scientific work and the asso- ogists—tend happily to refer to vast numbers of practitio- ciated reward structures have had significant impacts on our ners called ‘scientists,’ insiders function within specialist understanding of authorship. The “modern scriptor” (Bar- groups of remarkably small size . . . The group of people thes, 1977, p. 146) is no longer the sole conceiver, fabrica- mutually judged capable of participating in each of these tor, and owner of the published article. Instead, today’s specialized practices—what has been called the core-set— may be very small. biomedical journal article is the progeny of occasionally massive collaborations, the individual members of which may have minimal involvement in the fashioning of the Impressionistically, biomedical collaborations are more literary end-product itself, with the act of writing being heterogeneous and socially diffuse in character and do not delegated to a subgroup or designated spokespersons. In appear to have the same degree of multilayered, internal review as HEP research collaborations. This suggests that such circumstances, the standard model, with its connota- local practices, social and organizational, shape knowledge tions of individualism and intimacy, seems inadequate. Ever production and dissemination behaviors in high-energy more fine-grained division of “cognitive labor” (Kitcher, physics in ways that make it significantly different from 1993, p. 303), and ever-increasing interdependence in areas medical and related disciplines, with which it may have such as biomedicine and high-energy physics, may require surface similarities. An example would be electronic pre- a fundamental rethinking of the concept of author. print archives, which are a well-established and much- In the contexts of biomedical and HEP research, it is discussed phenomenon in the physics community, but, as difficult to reject outright the appeal of the radical model, yet, little imitated in other scientific disciplines (Kling, with its bifurcation of authorship into contributorship and McKim, Fortuna, & King, 1999). For this reason, it is wise guarantorship, since what is implied by a byline in these to avoid generalizations and to concentrate instead on show- cases is typically a very precise, often specialized input to a ing how interactions between coworkers, specifically the complex, multidisciplinary project. The classical idea of orchestration of information exchange and coauthorship, are authorship cannot credibly accommodate the legions of grounded in local culture. That said, “the usual methods of coworkers associated with large-scale collaboration, nor can observation, interviews or questionnaire” may lack the nec- it adequately reflect “the epistemic role of support person- essary sensitivity to capture precisely the often evanescent nel” in the conduct of science (Shapin, 1995, p. 359). In and inherently social nature of collaboration (Katz & Mar- some, admittedly extreme, instances, authorship and writing tin, 1997, p. 2). However, a diverse body of work on the are being decoupled. It is clear from an examination of the socially situated nature of scientific communication already journal literature of biomedicine that some kind of ontolog- exists which points the way. This ranges from Crane’s ical reassessment of authorship is called for to ensure that (1969) pioneering analyses of invisible colleges through authority, credit, and accountability, currently apportioned Latour and Woolgar’s (1979) classic study of laboratory life in confused fashion across authors, acknowledgees and con- at the Salk Institute to Traweek’s (1992) richly textured tributors, are henceforth distributed appropriately, parsimo- ethnography of the HEP community. In addition, the work niously, and unambiguously. It is possible that the ongoing of Schatz and colleagues on the Worm Community System debate in the biomedical literature will spark fresh exami- project, which was designed to capture the full range of nation of publication, and reward, practices in other fields, knowledge, formal and informal, of the community of mo- but much more systematic investigation will be required, lecular biologists who study the nematode worm C. elegans along with some exploration of the extent to which article (see: http://www.canis.uiuc.edu/projects/wcs/index.html) genre and place of publication influence rates of coauthor- can provide useful insights; so, too, research into the mate- ship across disciplines. In the future, it is quite likely that rial practices and social interactions of scientists working in the concept of the “author-function”, to use Foucault’s collaboratories, such as the Upper Atmospheric Research (1977, pp. 124–131) term, will vary from one “epistemic Collaboratory (see: http://intel.si.umich.edu/crew/Research/ community” (David, 1998, p. 139) to the next. In some resrch08.htm) or the Space, Physics & Aeronomy Research domains, the solo scholar may well take a back seat to the Collaboratory (see: http://intel.si.umich.edu/sparc/) at the investigative ensemble in crafting the annals of science, but University of Michigan. it would be grossly premature to herald the death of the author. Summary and Conclusions

In biomedicine, authorship has irrevocably shed some of Acknowledgments its craft associations: to be an author is not necessarily to be a writer. But it should not surprise us. Contemporary sci- I am very grateful to Helen Atkins, Elisabeth Davenport, ence is quite different from 17th-century science in terms of Charles Davis, Ron Day, Elin Jacob, Rob Kling, Alice both its social and economic structures. In many areas, Robbin, and Debora Shaw for their feedback on earlier interdependence is an inescapable fact of research life, and drafts of this paper. The highly constructive criticisms and the idea of the lone scholar something of an anachronism. suggestions of both referees are warmly acknowledged.

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 567 References Cronin, B. (in press). Acknowledgment trends in the research literature of information science. Journal of Documentation. Anglo-American cataloging rules (2nd ed.). (1998). Chicago, IL: American Cronin, B., & Weaver-Wozniak, S. (1992). An online acknowledgements Library Association. index: rationale and feasibility. In Proceedings of the 16th International Ahmed, S.M., Maurana, C.A., Engle, J.A., Uddin, D.E., & Glaus, K.D. Online Meeting, London, December 8–10, 1992. Oxford, UK: Learned (1997). A method for assigning authorship in multiauthored publica- Information (pp. 281–290). tions. Family Medicine, 29, 42–44. Cronin, B., Davenport, E., & Martinson, A. (1997). Women’s studies: Anderson, C. (1991). Hunger strike at volcanology institute. Nature, 354, 3. Bibliometric and content analysis of the formative years. Journal of Anderson, C. (1992). Authorship. Writer’s cramp. Nature, 355, 101. Documentation, 53, 121–136. Baca, M. (Ed.). (1998). Introduction to metadata: pathways to digital Daily, G.C. et al. (2000). The nature of value and the value of nature. information. Los Angeles, CA: Getty Information Institute. Science, 289, 395–396. Darwin, C. (1919). Autobiography. In F. Darwin (Ed.), The life and letters Barthes, R. (1977). The death of the author. In R. Barthes, Image, music, of Charles Darwin. New York, NY: Appleton. text. Essays selected and translated by S. Heath (pp. 142–148). New David, P.A. (1998). Communication norms and the collective cognitive York, NY: Hill and Wang. performance of “invisible colleges.” In G. Barba Navaretti, P. Dasgupta, Bartle, S.A., Fink, A.A., & Hayes, B.C. (2000). Psychology of the scientist: K.-G. Maler, & D. Siniscalco (Eds.). Creation and transfer of knowl- LXXX. Attitudes regarding authorship issues in psychological publica- edge: Institutions and incentives (pp. 115–163). Berlin: Springer. tions. Psychological Reports, 2000, 86, 771–788. Drenth, J.P.H. (1998). Multiple authorship: The contribution of senior Bazerman, C. (1984). Modern evolution of the experimental report in authors. Journal of the American Medical Association, 280, 219–221. physics: spectroscopic articles in Physical Review, 1893–1980. Social Ducor, P. (2000). Intellectual property: Coauthorship and coinventorship. Studies of Science, 14, 163–196. Science, 289(5481), 873–875. Bazerman, C. (1988). Shaping written knowledge: the genre and activity of Duncan, A.M. (1999). Authorship, dissemination of research findings, and the experimental article in science. Madison, WI: University of Wiscon- related matters. Applied Nursing Research, 12, 101–106. sin Press. Duncan, E.B. et al. (1981). Qualified citation indexing: Its relevance to Beaver, D. de B. & Rosen, R. (1978). Studies in scientific collaboration. educational technology. In E.B. Duncan & R. McAleese (Eds.), Infor- Part I: The professional origins of scientific co-authorship. Scientomet- mation retrieval in educational technology. Proceedings of the first rics, 1, 65–84. symposium on information retrieval in educational technology, April 1, Bhopal, R., Rankin, J., McColl, E., Thomas, L., Kaner, E., Stacy, R., 1981, Aberdeen (pp. 70–79). Aberdeen, UK: University of Aberdeen. Pearson, P., Vernon, B., & Rogers, H. (1997). The vexed question of Endersby, J.W. (1996). Collaborative research in the social sciences: authorship: views of researchers in a British medical faculty. British Multiple authorship and publication credit. Social Science Quarterly, 77, Medical Journal, 314(7086), 1009–1012. 375–392. Bird, J.E. (1997). Authorship patterns in marine mammal science, 1985– Eysenck, H.J. (1999). Why do scientists cheat? Journal of Information 1995. Scientometrics, 39,19–27. Ethics, 8, 27–35. Birkerts, S. (1994). The Gutenberg elegies: the fate of reading in an Fine, G.A. (1993). The sad demise, mysterious disappearance, and glorious electronic age. New York, NY: Fawcett Columbine. triumph of symbolic interactionism. Annual Review of Sociology, 19, Bordons, M., & Gomez, I. (2000). Collaboration networks in science. In B. 61–87. Cronin & H.B. Atkins (Eds.), The web of knowledge: a festschrift in Flanagin, A., Carey, L., Fontanarosa, P.B., Phillips, S.G., Pace, B.P., honor of Eugene Garfield (ASIS Monograph Series, pp. 197–213). Lundberg, G.D., & Rennie, D. (1998). Prevalence of articles with Medford, NJ: Information Today Inc. honorary authors and ghost authors in peer-reviewed medical journals. Bourdieu, P. (1991). Language & symbolic power. Cambridge, MA: Har- Journal of the American Medical Association, 280, 222–224. vard University Press. Foucault, M. (1977). What is an author? In D.F. Bouchard (Ed.), Language, Brodkey, L. (1987). Academic writing as social practice. Philadelphia, PA: countermemory, practice: Selected essays and interviews. Ithaca, NY: Temple University Press. Cornell University Press, 113–138. Carbone, P.P. (1992). On authorship and acknowledgements. New England Franck, G. (1999). Scientific communication—a vanity fair? Science, 286, Journal of Medicine, 326, 1084. 53–56. Chubin, D.E., & Hackett, E.J. (1990). Peerless science: peer review and Gaeta, T.J. (1999). Authorship: “Law” and order. Academic Emergency U.S. science policy. Albany, NY: State University of New York Press. Medicine, 6, 297–301. Cohen, P. (1999). Your mission is . . .. New Scientist, 2 July, 38–41. Grafton, A. (1997). The footnote: A curious history. Cambridge, MA: Constantian, M.B. (1999). The Gordian knot of multiple authorship. Plastic Harvard University Press. and Reconstructive Surgery, 103, 2064–2066. Gross, A.G., Harmon, J.E., & Reidy, M.S. (2000). Argument and 17th- Couch, C.J. (1984). Symbolic interaction and generic sociological princi- century science: A rhetorical analysis with sociological implications. ples. Symbolic Interaction, 8, 1–13. Social Studies of Science, 30, 371–396. Crane, D. (1969). Social structure in a group of scientists: A test of the Harsayni, M.A. (1993). Multiple authors, multiple problems: Bibliometrics “invisible college” hypothesis. American Sociological Review, 32, 335– and the study of scholarly collaboration. A literature review. Library & 352. Information Science Research, 15, 325–354. Croll, R.P. (1984). The noncontributing author: an issue of credit and Heffner, A.G. (1979). Authorship recognition of subordinates in collabo- responsibility. Perspectives in Biology and Medicine, 27, 401–407. rative research. Social Studies of Science, 9, 377–384. Cronin, B. (1984). The citation process: The role and significance of Heffner, A.G. (1981). Funded research, multiple authorship, and subau- citations in scientific communication. London: Taylor Graham. thorship collaboration in four disciplines. Scientometrics, 3, 5–12. Cronin, B. (1991). Let the credits roll: A preliminary examination of the Hoen, W.P., Walvoort, H.C., & Overbeke, A.J.P.M. (1998). What are the role played by mentors and trusted assessors in disciplinary formation. factors determining authorship and the order of the authors’ names? A Journal of Documentation, 47, 227–239. study among authors of the Nederlands Tijdschrift voor Geneeskunde Cronin, B. (1995). The scholar’s courtesy: The role of acknowledgement in (Dutch Journal of Medicine). Journal of the American Medical Associ- the primary communication process. London: Taylor Graham. ation, 280, 217–218. Cronin, B. (1996). Rates of return to citation. Journal of Documentation, Horton, R. (1998). The unmasked carnival of science. The Lancet, 52, 188–197. 351(9104), 688–689. Cronin, B. (2000). Semiotics and evaluative bibliometrics. Journal of Houston, P., & Moher, D. (1996). Redundancy, disaggregation, and the Documentation, 56, 440–453. integrity of medical research. The Lancet, 347(9007), 1024–1026.

568 JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 Hyland, K. (1999). Academic attribution: Citation and the construction of Merton, R.K. (1973). The normative structure of science. In R.K. Merton disciplinary knowledge. Applied Linguistics, 20, 341–367. and N.W. Storer (Ed.), The sociology of science: Theoretical and em- International Committee of Medical Journal Editors (ICMJE). (1997). pirical investigations. Chicago, IL: University of Chicago Press. Uniform requirements for manuscripts submitted to biomedical journals. Mullins, N.C. (1973). Theories and theory groups in American sociology. Journal of the American Medical Association, 277, 927–934. New York, NY: Harper & Row. Kahn, K.S., Nwosu, C.R., Kahn, S.F., Dwarakanath, L.S., & Chien, P.F.W. Newman, D. (1996). Writing together separately: Critical discourse and the (1999). A controlled analysis of authorship trends over two decades. problems of cross-ethnic co-authorship. Area, 28, 1–12. American Journal of Obstetrics and Gynecology, 181, 503–507. Newman, M.E.J. (2000). Who is the best connected scientist? A study of Kassirer, J.P., & Angell, M. (1991). On authorship and acknowledgements. scientific coauthorship networks. Available at: http://eprints.lanl.gov/ New England Journal of Medicine, 325, 1510–1512. cgi-bin/w3vdkhgw?qryGCA02ifHs;conm-0011144 Katz, J.S., & Martin, B.R. (1997). What is research collaboration? Re- Nowak, M.A. & Sigmund, K. (2000). Shrewd investments. Science, 288, search Policy, 26, 1–18. May 5, 819–820. Katzen, M.F. (1980). The changing appearance of research journals in Patel, N. (1973). Collaboration in the professional growth of American science and technology: an analysis and a case study. In A.J. Meadows sociology. Social Science Information, 12, 77–92. (Ed.), Development of scientific publication in Europe (pp. 177–214). Poster, M. (1990). The mode of information: Poststructuralism and social Amsterdam: Elsevier. context. Chicago, IL: Chicago University Press. Kernan, A. (1990). The death of literature. New Haven, CT: Yale Univer- Price de Solla, D.J. (1963). Little science, big science. New York, NY: sity Press. Columbia University Press. Kevles, D.J. (1998). The Baltimore case: A trial of politics, science, and Rayward, W.B. (1992). Restructuring and mobilizing information in doc- character. New York, NY: Norton. uments: A historical perspective. In P. Vakkari & Cronin, B. (Eds.), King, J.T. (2000). How many neurosurgeons does it take to write a research Conceptions of library and information science: Historical, empirical article? Authorship proliferation in neurosurgical research. Neurosur- and theoretical perspectives. London: Taylor Graham, 50–68. Relman, A.S. (1983). Lessons from the Darsee affair. New England Journal gery, 47, 435–440. of Medicine, 308, 1415–1417. Kitcher, P. (1993). The advancement of science: Science without legend, Rennie, D., & Flanagin, A. (1994). Authorship! Authorship!: Guests, objectivity without illusions. New York, NY: Oxford University Press. ghosts grafters, and the two-sided coin. Journal of the American Medical Klein, C.J., & Moser-Veillon, P.B. (1999). Authorship: Can you claim a Association, 271, 469–471. byline? Journal of the American Dietetic Association, 99, 77–79. Rennie, D., Yank, V., & Emanuel, L. (1997). When authorship fails: A Kling, R., & McKim, G. (2000). Not just a matter of time: Field differences proposal to make contributors accountable. Journal of the American in the shaping of electronic media in supporting scientific communica- Medical Association, 287, 579–585. tion. Journal of the American Society for Information Science, 51, Rennie, D., & Yank, V. (1998). If authors became contributors, everyone 1306–1320. would gain, especially the reader. American Journal of Public Health, Kling, R., McKim, G., Fortuna, J., & King, A. (1999). A bit more to IT: 88, 828–830. Scientific multiple media communication forums as socio-technical in- Riesenberg, D., & Lundberg, G.D. (1990). The order of authorship: who’s teraction networks. Available at: http://www.slis.indiana.edu/SCIT/ on first? Journal of the American Medical Association, 264, 1857. publications.html Saffran, M. (1989). On multiple authorship: Describe the contribution. The Knight, D. (1976). The nature of science: The history of science in western Scientist, 3, 9. culture since 1600. London: Deutsch. Shapin, S. (1995). A social history of truth. Chicago, IL: University of Koehler, W. et al. (1999). A bibliometric exploration of the demographics Chicago Press. of journal articles: Fifty years of American Documentation and the Slone, R.M. (1996). Coauthors’ contributions to major papers published in Journal of the American Society for Information Science. Available at: the AJR: Frequency of undeserved authorship. American Journal of http://www.ou.edu/cas/slis/courses/Methods/jbib/ Roentgenology, 167, 571–579. Kronick, D.A. (2001). The comerce of letters: Networks and “invisible Smith, R. (1997). Authorship is dying: Long live contributorship. British colleges” in seventeenth and eighteenth-century Europe. Library Quar- Medical Journal, 315(7110), 696. terly, 71, 28–43. Spiegel, D., & Keith-Spiegel, P. (1970). Assignment of publication credits: Kunst, F. et al. (1997). The complete genome sequence of the Gram- ethics and practices of psychologists. American Psychologist, 25, 738– positive bacterium Bacillus Subtilis. Nature, 390(6657), 249–256. 747. Latour, B., & Woolgar, S. (1979). Laboratory life: The social construction Squires, B.P. (1996). Authors: Who contributes what? Canadian Medical of scientific facts. Beverly Hills, CA: Sage. Association Journal, 155, 897–898. McCain, K.W. (1991). Communication, competition, and secrecy: The Stern, E.B. (2000). Authorship criteria: Opening a dialogue. American production and dissemination of research-related information in genet- Journal of Occupational Therapy, 54, 214–217. ics. Science, Technology, & Human Values, 16, 491–516. Traweek, S. (1992). Beamtimes and lifetimes: The world of high energy McDonald, K.A. (1995). Too many co-authors? Chronicle of Higher Ed- physics. Cambridge, MA: Harvard University Press. ucation, April 28, A35–A36. Verner, D.A. (1993). Astronomy acknowledgement index 1992. Messen- Manguel, A. (1997). A history of reading. London: Flamingo. ger, 71, 59. Manten, A.A. (1980). Development of European scientific journal publish- Wilcox, L.J. (1998). Authorship: The coin of the realm, the source of ing before 1850. In A.J. Meadows (Ed.), Development of scientific complaints. Journal of the American Medical Association, 280, 216– publication in Europe. Amsterdam: Elsevier, 1–22. 217.

JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY—May 2001 569