Against Algorithm

• In Event: The Social Life of Algorithms - I

Abstract

Paul Feyerabend’s book, Against Method, opened philosophy to the possibility that there is not just one scientific method but many. For the contemporary abstraction of “algorithm,” we need to follow Feyerabend. The “algorithm” is said to be rooted in the distant past (Euclid is frequently cited), yet, the definition of algorithm used today is due to efforts by Donald Knuth and others in the 1960s. Algorithms are still at the center of computer science research and teaching; and, they are also the gatekeepers of the computer industry: job interviews for Silicon Valley companies frequently feature tests about algorithms. Ironically Knuth et al. made algorithms the pivot of computer science by paring them down to be tantamount to mathematical functions and, thereby, making them useless for the analysis of most everyday uses of computing; e.g., via this definition, interactive systems are not algorithms. Knuth et al. insist that algorithms are mathematics but they also employ a metaphor that opens them to social and cultural interpretation: algorithms are said to be like recipes. This can connect them to Pamela Long’s history of writings from and about the mechanical arts, including Hellenistic engineering books and manuals of political and military praxis. To launch a new study of algorithms we need see the converse of computer science: algorithms are really recipes and only metaphorically mathematics. I began this work in my recent book, The Software Arts (MIT Press, 2019) but would like the opportunity to discuss this further with STS colleagues.

Author

• Warren Sack, University of California, Santa Cruz