Book of the Week: Decoding Liberation (2): Towards an open computer science

We continue and conclude our presentation of the recommended book, Decoding Liberation, with 2 excerpts.

If you want a copy of the book, drop the authors a line at schopra (or sdexter) AT sci.brooklyn.cuny.edu

The first excerpt, a critique of classic computer science, is contrasted with its alternative, an open computer science, in the second excerpt.

Samir Chopra and Scott Dexter:

1. The Failure of Computer Science

“Science is truth-tropic because of open criticism. In the realm of computing, free software provides and protects openness, not only through community-wide criticism but also through its challenge to the very idea of “intellectual property.” These practices, contrary to the constraints of proprietary software, have the effect of making public the scientific knowledge embedded in software. Contemporary computer science fails to meet most standards for objectivity in the sciences (Longino 1990), yet it carries the potential to become a strongly objective science, the publicly practiced discipline exemplified by the free software community.

The requirement that an objective science must provide recognized avenues of criticism is only partially met by a world in which proprietary code is so prevalent. While the discipline of computer science is equipped with all the requisite institutions for public criticism, such as peer-reviewed conferences and journals, practitioners who trade in closed code only participate in limited fashion. There is no avenue of criticism for closed code other than private code review among a small group of developers. Corporations or university researchers who work with or produce proprietary software may release research reports, but they provide little knowledge of the inner workings of their code: problem-solving techniques implemented in closed code are not subject to the critique of the entire community. Using closed code is akin to the deliberate obfuscation of research so as to avoid critique, emendation, or poaching: there is no scope for the collective improvement afforded by intersubjective criticism.

Scientific objectivity requires shared principles and values to undergird meaningful discipline-wide discourse. Much of computer science is done in corporate settings, which have their own norms — not necessarily congruent with those of the academy — about the practice of science. If the products of one group within a discipline are not available to the rest, this impoverishes any notion of “scientific community”…if so little of the knowledge produced within corporate walls is available to the broader scientific community, it may not be accurate to say commercial computing enterprises truly practice the communitywide discipline of “computer science.” A discipline is not just a subject matter but also the social structures that sustain it…

…..In computer science, the political power to determine the scientific direction of the discipline flows from the technical and economic power wielded by its dominant commercial members; one significant way in which this power is retained is through the mechanisms of closed code. All authority on closed code rests with its writers and anyone else who has agreed not to disclose its contents.….The closed model thus encourages the creation of an intellectual monopoly in computer science, an artificial scarcity of code that may become the basis of an economic monopoly for its owner. Monopolistic owners of proprietary code then exercise power derived from their domination of the market, introducing inertia by setting impossibly high economic barriers for potential competitors to breach, and forcing conformance by users to their technical standards.

….Closed code adversely affects, too, the objective evaluation of rival hypotheses in computer science. Even if such evaluation were entirely objective, if nonscientific values partly determine the candidacy of these hypotheses, they will play a role in determining which hypotheses are adopted (Okruhlik 1998). We would evaluate scientific theories, then, not against all possible competitors, but against a small number of rival possibilities. If bias or prejudice has prevented some possibilities from being considered at all, then an apparently well-confirmed theory may just be the best of a bad lot. If closed code restricts the evaluation of competitors in the ecology of computer science epistemology, then not all theories compete equally against selection pressures. Closed code, if understood as a hypothesis for a possible solution to a problem, is not subject to refutationist or falsificationist pressures (Popper 1962). When the level of detail for “full disclosure,” a standard of “normal” science, is not acceptable, concerns about the decline of objectivity invariably follow.”

2. Toward “Open Computer Science”

“A computer science based on open code, protocols, and standards promises a reconfiguration of research, industry, and pedagogy, ameliorating the effects of the current entanglement of the academic and corporate sectors on the practice of the science. Universities and commercial vendors could become true partners, with the innovative capabilities of both equally available to the other. Universities dream of being technology incubators; these could take the shape of consortia modeled on the Apache Software Foundation, drawing partners from multiple sources. The availability of previously closed code could infuse teaching and research with new knowledge, as huge bodies of code, currently hidden from the curious, became available for inspection and modification. As a previously closed science opened its textbooks, experimental manuals, and lab notes, we would gain access to new facts, practices, theories, paradigms, and epistemologies. The resulting wave of innovation would benefit not only computer science but also scientists in other disciplines, commercial users, and citizens: the creative expression of closed-code programmers would finally be on display.

Open computer science would bring the discipline closer to meeting the requirements for an objective science (Longino 1990). Opportunities for public intersubjective criticism would be much broader as open code is predicated on vigorous critical activity: “the greater the number of different points of view included in a given community, the more likely it is that its scientific practice will be objective” (Longino 1990, 184). The very act of merging open and formerly closed code would require the development of shared technical standards and norms. This sharing of standards and values and, thereby the reworking of the university–industrial relationship, would give meaning to talk of “the community of computer scientists.” The meritocratic structure so familiar to the free software development world would support a more equitable distribution of intellectual authority. The opening of code would bring with it a greater role for the Internet in distributing results and facilitating communication among computer scientists as “Heterogenous computer networks . . . provide the possibility for a kind of peer-review that precedes, and occasionally bypasses . . . the existing, trusted, and well-known peer-review of scientific journals and societies” (Kelty 2001). In this setting, knowledge, no longer under exclusive control to support the extraction of monopoly rent, would gain in value with its increasing availability.

The discipline we describe resembles the fledgling computer science of the 1950s and 1960s, which, because of a dramatic change in its political economy during the 1970s, moved toward a proprietary economy that free software counters. The safeguarding of this potentially objective sphere for the practice of computer science might take inspiration from strong free-software licensing schemes such as the GPL, as its practitioners’ endeavors would be available to all under conditions of reciprocity. Closed software is a fundamental negation of the social character of computer science. Free software reminds us that the political economy of science must not be limited only to the consideration of efficient allocation of scarce resources; it must also address the fundamental principles by which a technologized society should be organized (Kelty 2005). For economic reasons, computer science is currently looking past — indeed, defending — practices that stifle innovation. If economic arguments are used to support bad science and attack potentially reformatory practices, then the science is not easily distinguished from an astrology whose practitioners concoct fantasies for a fee.”

Book: Decoding Liberation: The Promise of Free and Open Source Software. Samir Chopra and Scott Dexter. Routledge, 2007.

Leave A Comment

Your email address will not be published. Required fields are marked *