Society for Cultural Anthropology
Oral Presentation Session
The design of digital infrastructures has been a central issue of technical and political debate for decades. Centralized infrastructures in particular have been met with strong opposition from activists, especially if they perceive them to be symbols of massive corporations (Turner 2006). Hackers have sought to support their own ethics and aesthetics of decentralized computing by creating free software tools − programs that anyone can download, modify and redistribute (Kelty 2008 ; Coleman 2012). These online communities can often gather hundreds, sometimes thousands of contributors, reaching a size that few corporate software teams achieve.
Partly because of their impressive size, a significant number of companies have officially adopted free software to run their infrastructures. But engineers soon found out that these tools did not work adequately in a corporate setting, making daily tasks difficult. They labeled the problem as one of "scale": the tools were simply not designed to handle such large infrastructures. The corporate teams reached out to the free software projects to ask them to change their algorithms, implying that decentralized designs were not optimal and starting a heated debate within these communities.
Drawing on this controversy and on my fieldwork following software engineers, I will explore what the concept of "scale" and the process of "scaling" mean when applied to algorithms and why they can be so controversial. I will argue that issues of scale are closely tied to competing imperatives of growth, material and technical cultures as well as institutional embeddings.