Bowker’s (2005) Memory Practices in the Sciences is significant for its elaboration of sci-entific memory in terms of technologies for information processing and formal systems that structure the scientific record. Bowker provides an understanding of scientific memory as a technologically mediated practice, against which we can consider the case of software de-velopment, generally, and simulation development in particular. Straightforwardly, memory practices concern useful descriptions of the past. They are activities, processes, strategies, and techniques through which such descriptions are produced, made use of, and carried into the future. The subtleties of memory practices emerge by considering the full scope and spectrum of ways that a useful past has been conceptualized and materially constituted. As this suggests, memory practices are historically specific and they are intimately related to technical conditions.
To illustrate, consider the different kinds of ‘records’ associated with laboratory note-books, on the one hand, and published accounts of laboratory experiments on, on the other.
As Shankar (2007) notes through a study of scientists’ personal record-keeping practices in the lab, the scientific record emerges through the production of documents that are both personally meaningful to the individual scientist and broadly reliable for the purposes of the scientific community. The end result is “a document that is paradoxically wholly per-sonal and yet intrinsically professional.” Published accounts, on the other hand, are widely recognized as idealized versions of what happened during a scientific investigation. This ide-alization has a purpose. As Bowker notes: “Scientific [publications] are written not to record what actually happened in the laboratory, but to tell the story of an ideal past in which the protocols were duly followed: the past that is presented should be impregnable [. . . ]. It takes a great deal of hard work to erect a past beyond suspicion” (Bowker 2005, 7). The point of the comparison between lab notebooks and publications is not to uphold the latter as a white-washed fraud and the former as the messy reality. Rather, the idealized report and the internally-held institutional memory of the lab notebook correspond to distinctive memory practices. They constitute the past as a useful resource in different ways, however, their real
value emerges through their articulation. Memory practices are not isolated collections of activities. Collectively they form “memory regimes” to create a shared—and to a certain degree unitary—conception of a “continuous, useful past.”
Common technologies and formal tools for recording, structuring, and processing infor-mation are conditions of possibility for relating memory practices to one another and for forming memory regimes. As Shankar notes, laboratory notes must be personally mean-ingful to the scientist but they must also lend themselves to becoming “enmeshed in larger meanings that dovetail with expectations of the broader scientific community” (Shankar 2007). The “poetics and pragmatics” of laboratory recordkeeping resides in the personalized strategies for negotiating these twin demands. Bowker further expands the scale of the com-mensurability afforded by common formal systems to the relationship between disciplines.
Interdisciplinarity, Bowker suggests, is possible to the extent that disparate fields share a common knowledge infrastructures within which research objects are constituted. “We do not as a society have a series of separate and separable discourses about the past [. . . ].
Rather, I would argue, the boundaries between disciplines are porous precisely because the same kind of information-processing technology is being used in each case” (Bowker 2005, 136). (This is not a deterministic argument, Bowker argues, because the history of informa-tion processing cannot be isolated from the ‘social’ and the ‘cultural.’)
The central “material substrate” of contemporary memory, and thus the ultimate ma-terial condition of commensurability in the present, is the digital computer. The primary form of computational information processing that Bowker considers is the database. (The emphasis on data storage and data organization is understandable considering the fields that Bowker focuses on—biological sciences, bioinformatics, and biodiversity—fields heav-ily oriented toward the representation of the past and present of life on Earth). Software and software development receive relatively little attention in this elaboration of scientific memory. The computer program is mainly discussed as a piece of “ideal machinery”—as a metaphor for understanding reality (e.g., the universe is a simulation) that structures our conception of the past.
scious recovery of the past (i.e., in the mind of an individual). Bowker associates forms of non-conscious memory with rules and constraints that limit interpretation and action. Rules or constraints are forms of externalized memory. Technical forms—particularly classification systems and standards—are central features of memory practices in this respect. Classifica-tion systems constitute descriptive matrices, limited ontological spaces that constrain what can and cannot be named and remembered. The basic thrust of memory practice, as a way of thinking about science, is that technologies for recording and processing information have a constitutive power derived from the constraints they impose on scientists’ objects of study.
This power is ‘onto-epistemic’ in nature, as the distinction between the object produced in a technologically produced space of representation is not distinguishable from the object of knowledge. I return to this point below in my discussion of Rheinberger’s understanding of the materiality of experimental systems as “graphematic articulation” (Rheinberger 1997, 106). The primary object that Bowker is concerned with is ‘the past’. Importantly this is both the past as an object of study (e.g., the Earth’s past) and the past of sciences’ efforts to elaborate that object.
Of the historical memory regimes that Bokwer considers, the case of cybernetics warrants special considerations for several reasons. First, histories of computing and cybernetics are intimately related, so it is worth looking to the latter to understand something of the former.
Cybernetic systems are also closely related to simulations—they are both closed systems in which complex behaviors emerge through feedback. As a result, the way cyberneticists conceptualized memory and its role in the study of complex systems serves as a useful point of comparison for thinking about contemporary simulation practice. As we will see, however, the memory practices of cybernetics are historically distinct. They have largely been interpreted as an example of knowledge production in a performative idiom (Pickering 2010).