Invitation
By Yuwei Lin
Diverse choices and practices shape innovation patterns in software development
My research is associated with computer hacking and trying to understand the impact that hacking has on computing practice and development. This paper examines the overlap between the hacker community and the open source community, and issues about innovation, practices and identities in terms of the cultural dynamics of computer software innovation.
The Driving Force of F/OSS Innovation
Free/Open Source Software (F/OSS) has been gaining momentum in the software industry. Not only academics but also practitioners want to know why the F/OSS phenomenon succeeds in our capitalist world, where intellectual property rights and control over patents are considered to be key to successful innovation. First of all, one should consider the origin of F/OSS. The scientific method might be regarded as similar to open source practice, sharing information to facilitate scientific discovery and replication; the former is to enable other scientists to go forward where one cannot, and the latter is to enable justification under specific test conditions (DiBona, Ockman & Stone 1999). "Where scientists talk of replication, free/open source programmers talk of debugging; where scientists talk of discovering, free/open source programmers talk of creating." (ibid.) Open source is not a new way of doing things -- rather it might be regarded as the original computer way of doing things (Rosenberg, D.K., 2000).
It is generally agreed that OSS itself originated from the hacker culture of the 1960s and '70s. The open source world is hacker-driven, and hackers define themselves as creative and expert software programmers who push the limits of the "doable" (ibid). Through reference to the General Public License (GPL), a license introduced by Richard Stallman (RMS), the founder of the Free Software Foundation (FSF), one gains a better understanding of how the F/OSS world works.
GPL allows free use, modification and distribution of software and any changes to it, restricted only by the stipulation that those who received the software pass it with identical freedoms to obtain source code, modify it and redistribute it. The products of a developer's mind must remain free for all developers to use and reuse. Free does not mean no money changes hands, but it does mean that no authority or non-disclosure agreement will prevent developers from sharing code. Because of the freedoms granted by the Free Software licensing method, and because the legal means of its enforcement is the copyrighting of the software itself, the use of the GNU GPL is often called "copylefting" (Rosenberg, 2000).
The openness of OSS innovation fosters multiplicity and diversity of programming practices within the system. The extensive creativity that has accompanied this broad active participation has made a single OSS project a more diverse socio-technical phenomenon. "Much like a fork in the road, a code base can at times diverge into two separate, incompatible roads and never the twain shall meet. This is not an idle problem; look, for instance, at the multiple forks that the BSD-based operating systems have taken, leading to Net BSD, OpenBSD, FreeBSD and many others." (DiBona, Ockman & Stone 1999) Each sub-project could become an independent project if there is enough interest engaged (i.e. KDE GNOME; both are derived from Linux's object-oriented desktop interface project) (ibid). While a project might have failed in the commercial sector, it survives or even develops through the attention of open source practitioners who are interested in it (i.e. Lucid Emacs, now Xemacs). "As long as people find a use for the software, they will maintain it to work with new systems and architectures" (ibid).
Among many F/OSS observers, Eric Raymond (1999, 2001), a self-appointed participant anthropologist to the OSS community, gives one answer to the question of why do people develop software only to give it away? Raymond believes that the motivation for open source is not just altruism. There is rather a social mechanism at work that helps to maintain/sustain the gift culture found among OS users. Raymond suggests that the gift culture actually has its competitive dimensions that are rewarded in the OS world. Apart from various licenses in the OSS world that state that all source codes should be distributed together with their binary codes, and that anyone who receives a program has the right to modify the source code and to further distribute the program, changed or unchanged, Raymond shows that there is a set of strict rules of ownership in the hacker community that form part of the informal culture of hacking.
Hackers who make contributions to a program get credit. The system of ownership and credit ensures the producers and users of OSS conform to these rules that produce pride, personal identity and a degree of mutual acknowledgment of effort and skills as some kind of social reward. Raymond argues that hackers compete to get higher social status by making greater contributions in OSS programs. Raymond goes on to explain the drive for competition however not as an appeal to a sociological account of normative adherence to these sharing rules, but as something approaching a social-biological position. He argues that hackers compete because they, as human beings, have an innate drive to do so.
In response to Raymond's position on open source (1999, 2001), Risan challenges Raymond's claims about hackers' competing for reputation from a sociological view: "Competition for social status surely takes place in the hacker community, between [software project] leaders as well as between followers, but we need something else than competition to explain the dynamics of the many leader-follower relationships among hackers."
I agree with Risan's remarks that the social dynamics of the hacker community are more complicated and diverse than the picture drawn by Raymond. There are a plurality of factors that perhaps Raymond, Risan and I cannot analyze. I also share the opinion with Risan that along with competition among hackers, cooperation is one of the other crucial elements of the success of OSS. But even then, the picture of the hacker community cannot be fully captured by these two mechanisms.
What is arguable is that Risan follows Raymond's way of categorizing the OSS practitioners into leaders and followers and continues to explain the origin of open source motivation through analogy with a Darwinian model of competition. Here, Risan's analysis is rather unpromising. He regards Darwin's sexual selection as an analogy to his complementary identity game: the two agents, on an evolutionary scale, would produce each other's environment, just like if males were also choosy and females were thereby to produce ever more elaborate traits. Risan presumes that the hacker community is composed mostly of young men in their 20s who develop and show impressive hacking as a means to attract possible sexual mates.
Multiplicity, Heterogeneity and Contingency
Risan is right in noting that there exists a myriad of literatures and practices among hackers or F/OSS practitioners and that none should be ignored. But his arguments based on sexual selection are not adequate for analyzing the hacker community's social world precisely because they reduce that world to paired but opposite forms of social actor.
In light of my observations and available documents, the hacker community is a social world with a blurred boundary and fluid membership. Diverse social groups make different claims and narratives to demonstrate their links to the hacker community (and deploy names such as hacker, cracker, phreaker, tinker, etc.). After all, there is no single definition of the word "hacker".
In the hacking social world, members compete and cooperate across various social groups (cross-boundary). Members' identities change over time and space (no matter if it's virtual or real) rather than staying as leaders or followers in the way that Risan suggests. Instead of assuming a quintessentially fixed subculture, the membership of a hacker community is quite loose and the boundary is soft. Because hacking practices can be seen through diverse social actors, I suggest a "pan-hacker phenomenon," to show that hacking is not a tight subculture; rather, it is a collection of activities with diverse actors who are good at programming. Hackers are those people sharing diverse interests in hacking, with dual or multiple memberships across the hacking boundary. The concept of subculture is essentially flawed due to its attempt to impose a hermetic seal around the relationship between hacking practices and members.
To sum up the argument above, non-sociological attempts to explain the dynamics of hacking have resorted to rather reductionist notions of the hacker, derived from a crude form of social behaviourism. They are also inadequate because they fail to see that there is evidence that software practices supposedly found only in hacking are found elsewhere too, notably in mainstream, commercial innovation. Hence any claim that there is something peculiar that has to be explained in regard to an essential OSS culture is mistaken.
Practices and Social Worlds
So far, one should understand that it is impossible to say who is competing and who collaborating with whom. It is also hard to tell whether or not hackers do these things to survive (natural selection) or to attract the opposite sex (sexual selection). They might just do it for fun (Torvalds 2002). However, aside from various claims, one can see that common practices have emerged on the free/open source software projects.
In some circumstances, hacking activities coalesce and create visible social worlds that are occupied by a range of social actors. This is exemplified in the hacker conferences, but those participants shouldn't all be assumed to engage in all the practices listed. Under certain circumstances, this social world might form a stronger hacker cultural network, especially, as in the process of "deviancy amplification," hacking practices become the focal point for computer users as a response to regulatory authorities' attempts to restrict these very practices. But if such circumstances or conditions change, the network may dissolve or be re-defined, such as, I want to argue, occurred over the last decade with the emergence of Linux as a form of institutionalized "hacking," inasmuch as it is based on a number of the practices. Here are some examples of my data collection via email interview.
DY0001: "The need to share information stems from the need to resolve problems caused by the multitude of proprietary systems and you need the code-breaking mentality to take on proprietary manufacturers."
JV0001: "The best solution to a problem typically comes from someone who has a direct interest in solving it, not from somebody who is merely being paid for it. The differences in productivity among programmers are so great (factors of 100 are very common!) that "real interest" can typically defeat "full-time labour" as a means of getting to the best solution."
Speaking of the main feature of OSS innovation, interviewees tend to regard it as a problem-solving process.
DY0001: "Software like everything else is driven by a need to solve a particular problem. There are some developers who are innovative for the sake of being creative. I think commercial software is mainly driven by the market and the need to sell a product while open source software is driven mostly by the need of a solution. The list has probably captured the essential elements of software innovation."
JV0001: "A lot of the work is really just thinking about problems, discussing problems, or mapping out potential solutions. Another very important activity is reporting bugs. Small bugs tend to have low priorities in proprietary product development, but in Open Source it's easy for a complete stranger to observe the problem, look at the source code, find the cause, and hand you a solution on a silver platter. This happens all the time in Open Source development."
The socio-technical element in the OSS innovation process is especially pointed out in some OSS practitioners' responses, such as:
JV0001: "The actual process is much more a social and psychological one. Remember the study where people were distracted from work by outside noise, and their productivity levels went back up when they were given a means of shutting the noise off, even if they didn't actually bother to do so? Software problems can be very much like that. In the proprietary model the user feels like a victim, and his productivity suffers while you as a developer aren't even likely to hear about the problem. In the Open Source model, the user knows that if the problem is bad enough for him, he can put his money where his mouth is and fix the problem, or perhaps pay somebody to do it for him, or at least report it to the developer. Previously stifled productivity is released on both sides, plus any other users also benefit when the problem is fixed."
Here is a more specific example of the OSS innovative practice:
JV0001: "Another way of looking at it is to observe the typical activities of someone developing proprietary software (of which I've also done a fair deal). The first concern is typically, screening oneself off from users who encounter problems with the software (make sure somebody else picks up the phone, then prioritize their complaints) or who want to request enhancements (typically poorly thought out because they don't get to see the internal structure of the product). Prioritization of requests then goes through revenue analysis, which typically doesn't make for the best technical solution. Then there's bug-hunting, an immensely costly activity that is consistently underestimated; users can't really help you do this because they aren't allowed to see the source code. When you do get around to solving customers' problems, you find yourself explaining the same workarounds over and over again because you're not allowed to fix the problem ahead of its prioritization."
Note the conclusion this respondent draws about proprietary software! And the list goes on. Looking back, proprietary development is weird and needs studying. There are many problems the model creates for itself, all because scarcity is something that needs to be maintained artificially through great effort, just to enable a licensing model of revenue where actual licenses form only one-eighth of software cost anyway.
In observing software practice, one can see some sense in which there is a degree of closure or at least convergence around a group of practices that are framed as OSS/hacking activity. Hacking practices gradually integrate in the mainstream software engineering along with the development of F/OSS. This more or less indicates that the F/OSS practices have been useful and beneficial for the whole of the computing industry and therefore many companies (IBM, HP etc.) and governments (Germany, France, etc.) have adopted F/OSS and contributed to the community. To study the F/OSS community based on a pan-hacker phenomenon in terms of STS, "practices" are the main elements connecting the two social worlds (i.e. the hacking social world and F/OSS social world).
"Why doesn't scientific culture continually disintegrate as scientific actors develop it in the myriad different ways that are conceivable in principle?" "How is closure -- the achievement of consensus on particular extensions of culture -- to be understood?" Pickering (1992) responds to the two questions that, actors [in the scientific knowledge network] have their own interests that [scientific] instruments can serve well or ill. While formally the proprietary computing sector is critical of the function of open source code for wider programming and debugging use, in reality it does appear to act as an instrument that "serves it well."
Pickering introduces the sociological concept of interest to explain the closure of scientific development: "On the one hand, actors can be seen as tentatively seeking to extend culture in ways that might serve their interests rather than in ways that might not; and on the other hand, interests serve as standards against which the products of such extensions, new conceptual nets, can be assessed." (ibid.) Pickering takes "the representation of practice as an open-ended process of modeling structured by interest," which is different from "the representation of scientific culture as a single conceptual network" (ibid). Consequentially and substantively, to understand the social dynamics of F/OSS innovation system, it would make more sense to study the multiplicity of cultural practice appearing on this open-ended process of practices across social worlds.
The coexisting positions in the development of science and technology on the question of the relation between competition and cooperation have been discussed widely. Some evolutionary economists see competition as more important than cooperation, while others attempt to give the two dimensions equal weight. But from a sociological point of view, it does not matter how much weight competition and cooperation occupy respectively in the development and innovation process. What is crucial is the multiple culture and various practices found in the social worlds that are not fixed or ordered, rather, indeterminate, flexible and contestable, that shape the F/OSS innovation. The F/OSS social worlds are much more dynamic than the paradigm given by Raymond or the ones suggested by others.
To analyze a dynamic and complex field, social worlds theory has been strongly advocated in comparison to other STS methodologies (such as actor-network theory). Social worlds are made up of groups with shared commitments to certain activities sharing resources of many kinds to achieve their goals.
In each social world, at least one primary activity (along with related activities) is strikingly evident. There are sites where activities occur; hence space and a shaped landscape are relevant. Technology (inherited or innovative means of carrying out the social world's activities) is always involved. In social worlds at their outset, there may be only a temporary division of labor, but once underway, organizations inevitably evolve to further one aspect or another of the world's activities. (Strauss 1978)
People typically participate in a number of social worlds simultaneously (Unruh, 1979; Clarke 1997); mixed worlds are both possible and common (ibid.). Participation in social worlds usually remains highly fluid -- the structure of social worlds is also highly fluid. After being discussed and debated by various scholars, the theory is used in the sense of Strauss and other scholars that the social worlds and subworlds themselves become the units of analysis in the study of collective action.
I'd especially like to draw on Fujimura's works to understand not only the dynamics but also the stability of a scientific arena. For Fujimura, to employ an arena analysis/social worlds theory is to analyze the heterogeneous constructions of OSS by these various actors. In doing so, a multiplicity of perspectives, heterogeneous voices and contingent actors and actants (programs) in an arena can be considered more fully. Apart from investigating claims made by diverse actors, it is also important for pragmatist and symbolic-interactionist researchers to explore the collective practices among the actors, as their claims and identities are embedded in practices, through which a technological tendency of software engineering can be mapped.
Though it seems to be impossible to identify a single boundary object in OSS arena, ambiguous concepts and standardized tools (or I'd add practices) of theory and methods are powerful tools for insuring fact stabilization (Fujimura 1992: 204). Drawing upon Callon and Latour's actor network approach, Fujimura emphasizes both that such boundary objects are actively "crafted" in a process of "mutual enrollment," and that the production of successful boundary objects reacts back upon the social worlds thus linked and upon the larger whole they make up, reconstituting the very objects of study, as well as the material, conceptual and social practices that surround them.
As Pickering comments, Fujimura offers us an image of the interactive stabilization of a plethora of cultural elements, while enriching our understanding of that process in her focus on the new patterns of intersection and circulation that come into sight when one recognizes the social heterogeneity of practices (Pickering 1992). Therefore, I stand for the concepts of "bandwagon" and "standardized package" proposed by Fujimura. As what she has claimed: "The next step in the social study of science is to examine this malleability and dynamism of truth matters by studying the processes of its production in different situations and from different viewpoints. While we will never be able fully to understand and represent the views of the other, this does not mean that we should not even attempt to 'explain' science" (Fujimura 1991: 218).
One might question why not use Callon and Latour's actor network theory? As Fujimura mentions there are many affinities and compatibilities between Callon and Latour's approach and the pragmatist-interactionist approach. But the differences in degrees, stances and goals still stand (ibid). Both theories are constructionist, relativist and focused on relations among actors. But SWT does not follow the most powerful actor in what Star (1991) calls the "executive approach," as in Latour's injunction to "follow the technoscientists!" Rather arena analysis attempts to view the constructed world metaphorically over the shoulders of all the actors. (Clarke & Montini 1993)
Two observers in the same situation might see, hear and note down entirely different kinds of activities. [Social worlds theory] emphasizes process, plurality, flexibility, novelty, the generation of many theories and concepts, and attention to local situations, and thus prevents us from falling into the rigid structural frameworks of functionalism and Marxism. Strauss argues that phenomena are so rich that we should mine them for more stories, more concepts, more "grounded theories," more ways of "seeing" phenomena rather than limit ourselves to one set of concepts, theory, or way of seeing (ibid.: 226)
Except for analyzing actors individually and collectively presenting articulate and committed to action in the arena, Clarke & Montini stress the crucial but often ignored role of those implicated by actions in that arena (1993). Methodologically, Clarke & Montini highlight multiple visions and means of achieving them by attempting empirically to view the world in the actors' own terms. There is a community of practice where local knowledge and situated knowledge (Mills 1940; P. H. Collins 1990; Haraway 1991) derive from identities and commitments largely developed through prolonged interaction toward shared, yet continually emergent, goals (ibid.).
It is exactly what I am trying to deal with my research -- to be open to the diverse and complex forms of data that can be found in the world of computer software engineering. In the OSS arena, various actors from diverse backgrounds shape progress digitally and in the material world each day. In light of the point that Clarke & Montini have made to compare SWT with ANT, there is really no centered actor in the OSS arena to be followed/traced as each actor plays as important a role as the others do. Moreover, the worlds of OSS actors are a hybrid combination of digital and real ones. This situation makes it difficult to concentrate on merely one actor/actant. SWT provides an overall perspective on actors from different social worlds engaging in a collective arena. This methodology offers each actor or social group an equal chance to express itself and the researcher is able to analyze the complicated interactions between them and the heterogeneous constructions of the technology.
Conclusion: Identity and Innovation
To bring the argument further, social worlds and identity are interactively constructed, and perspectives and aspirations emerge dynamically from this interaction. Since meanings are both culturally created and mediated, all interpretations or perspectives are based in communities or social worlds (Fujimura 1991). This is why one should not ignore the identity of actors when telling the story in terms of social worlds theory. However, this emergence is a complicated affair, since communities are not clearly circumscribed and defined and individuals participate in many different communities simultaneously. Moreover, individuals also participate in the ongoing construction of social worlds. Thus, identities and perspectives are multiple, processual, and dialectical. There is no simple one-to-one mapping between perspective and community membership, a principle that we can call the indeterminancy of perspectives (in line with the pragmatists' notion of the indeterminancy of situation) (ibid.).
Fujimura denotes that the analysis of the tension between indeterminancy and structure, identity and socio-historical location, and political choices and collective memory is the beauty of Strauss's work on identity and social psychology. Rather than focus on indeterminancy of all representations like the postmodernists or on the partial or naive view of the structural determinism of Marxists, Strauss focuses on the multiplicity, fluidity, temporality and processual nature of identity, perspectives and social worlds. He provides no neat categories, no certain way to classify people, ideas, actions.
Actors involved in the innovation process are often given different identities as developers, users, innovators, manufacturers and so on. Von Hippel (1988) proposes that innovators are most often users rather than scientists. He suggests a user-developed innovation and lead users often find the solution for the innovation. Various innovation studies have shown that innovation is a process and diverse actors including users, manufacturers and suppliers --whoever the innovators are -- all play important roles in the process. Von Hippel's idea that users are innovators may be novel but it is too limited as well. He ignores the heterogeneous and contingent factors in the innovation process and their complicated relationships and interactions. Users as localized collectives certainly play important roles in the innovation process, however, the identity and meaning of users changes along with contexts and settings. It is rather inadequate that Von Hippel simplifies users and other associated actors in the innovation process and produces an abridged story.
The innovation pattern I propose is not a linear cause-effect process. Actors in the innovation process are not given fixed a label/identity. In light of social worlds theory, the innovation pattern I suggest is highly fluid, heterogeneous and contingent. In the OSS innovation process, the notion of a problem/requirement varies from one actor to the other; the communication process between developers themselves, users themselves, or developers and users is contingent. Given the evolution of the digital computation industry, "the creative capabilities of a firm to do something new, and to take risks with unknown possibility of success, rests fundamentally on recognition of a problem as an opportunity." (Jackson, Mandeville & Potts 2002).
Sociologically speaking, diverse choices and recognition of problems shape the different innovation patterns. One should note that I am not suggesting that open source is the best practice for software innovation; whether the venture capital in the computing industry will flow to OSS will depend on the communication and negotiation between different actors from various social worlds. But in light of the OSS development, one can understand how the process of software engineering is shaped by various factors.
References
Clarke, A. E. 1991. Social worlds/arenas theory as organizational theory. In Social Organization and Social Processes: Essays in Honour of Andelm L. Strauss, edited by D. Maines. NY: Aldine Gruyter.
Clarke, A. & Montini, T. 1993. The many faces of RU486: Tales of situated knowledges and technological contestations. Science , Technology, & Human Values, Vol. 18 No. 1, p. 42-78.
Clarke, A. E. 1997. A social worlds research adventure: the case of reproductive science. In Grounded Theory in Practice, edited by A. Strauss & J. Corbin. London: Sage publications.
DiBona, C. & Ockman, S. & Stone, M. (eds.) 1999. Open Sources: Voices from the Open Source Revolution. CA: O'Reilly.
Fujimura, J. H. 1991. On methods, ontologies, and representation in the sociology of science: where do we stand? In David Maines (ed.) Social Organization and Social Processes: Essays in Honor of Anselm L. Strauss. NY: Aldine Gruyter.
Fujimura, J. H. 1992. Crafting science: standardized packages, boundary objects, and "translation", in Andrew Pickering (ed.) Science as Practice and Culture," Chicago: the University of Chicago Press.
Jackson, M. & Mandeville, T. & Potts, J. 2002. The evolution of the digital computation industry. Prometheus, Vol. 20 No. 4, p. 323-336.
Pickering, A. 1992. From science as knowledge to science as practice. In Science as Practice and Culture, edited by A. Pickering. Chicago: The University of Chicago Press.
Raymond, E. 2001. The Cathedral and the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary. Revised edition. O'Reilly.
Risan, L., "Hackers produce more than software, they produce hackers," Version 2.1 http://folk.uio.no/lr isan/Linux/Identity_games/
Rogers, E. M. with Showmaker, F.F. 1971. Communication of Innovations: A Cross-Cultural Approach, 2nd ed. NY: Free Press.
Rosenberg, D. K. 2000. Open Source: The Unauthorized White Papers, NY: M&T books.
Von Hippel, E. 1988. The Sources of Innovation, Oxford: Oxford University Press.
Ubiquity, Vol. 4, Issue 4, March 18 - 24, 2003
Forum
[Home] [About Ubiquity] [The Editors]
Ubiquity welcomes the submissions of articles from everyone interested in the future of information technology. Everything published in Ubiquity is copyrighted ©2000 by the ACM and the individual authors. |