Critical Code Studies often relies on the textual representation of code in order to derive extra-textual significance, with less focus on how code performs and in what contexts. In this paper we analyze three case studies in which a literal reading of each program’s code is effectively nonsense. In their performance, however, the programs generate meaning. To discern this meaning, we use the framework of nonmaterial performance (NMP), which is based on four tenets: code abstracts, code performs, code acts within a network, and code is vibrant. We begin with what is to our knowledge the oldest example of nonsense code: a program (now lost) from the 1950s that caused a Univac 1 computer to hum Happy Birthday. Second, we critique Firestarter, a processor stress test from the Technical University of Dresden. Finally, we analyze one of the family of processor power side-channel attacks known collectively as Platypus. In each case, the text of the code is a wholly unreliable guide to its extra-textual significance. This paper builds on work in Critical Code Studies by bringing in methodologies from actor-network theory and political science, examining code from a performance-studies perspective and with expertise from computer science. Code can certainly be read as literature, but ultimately it is text written to be performed. Imagining and observing the performance forces the critic to engage with the code in its own network. The three examples we have chosen to critique here are outliers---very little code in the world is purposed to manipulate the physical machine. Nonsense shows us the opportunity that nonmaterial performance creates: to decenter text from privileged position and to recenter code as a performance.
Rountree et al. (Thu,) studied this question.