ABSTRACT
Algorithmic calculations currently play a central role in organizing digital information, and in making it visible. Faced with the deluge of disordered and disparate data collected on the web, algorithms form the basis of all the tools used to guide the attention of Internet users (Citton 2014). In turn, rankings, social media buttons, counters, recommendations, maps, and clouds of keywords impose their order on the mass of digital information. For many observers, algorithms have replaced various human editors (journalists, librarians, critics, experts, etc.) to prioritize content that deserves to be highlighted and brought to public attention. Algorithms have thus come to serve as the new “gatekeepers” of public digital space (Zittrain 2006). It is therefore common that criticisms of algorithms reproduce, in a new context, the accusations often leveled at mass media in general: that they reflect the economic interests of the owners, distort markets, ignore the margins, are sensational, conformist, vulgar, etc. It is as if the calculation techniques of the web reflect only the interests of those who program them. But this simple manner of critiquing the power of algorithms neglects the strictly technical dimension of these new gatekeepers, as they make transparent the economic forces that extend throughout the new economy of the web. In this chapter,1 we argue that we cannot view the new computational techniques of the web as merely reflections of the interests of their owners. Extending the philosophical approach of Gilbert Simondon, we want to explore the technical and statistical properties of these computational tools, focusing particularly on the ways in which they require us to think differently about the production of power and hegemony on the web, and the ways it shapes and orients information online. The various calculation techniques implemented on the web exhibit great differences that are often effaced by the unifying effect of algorithms. Indeed, there exists a huge variety of ordering and classifying procedures: the search rankings of Google, the reputation metrics of social media, techniques of collaborative filtering, the ‘programmatic’ advertising of ‘real-time bidding’ (RTB), and the multiple ‘machine learning’ techniques that are becoming increasingly widespread in the calculations used by ‘big data.’ We would also like to clarify the different web calculation techniques in order to describe the digital worlds they give rise to, each according to their own individual logic. Designers delegate
values and goals to computer objects that make “cognitive artifacts” (Norman 1991) responsible for operating processes and choices, as well as for authorizing and preventing, for classifying and orienting (Winner 1980; Introna and Nissenbaum 2000). Their development has progressively integrated technical solutions that can address a wide range of problems related to statistics, usage, laws, markets, etc. that appear at various stages. In addition, we would like to investigate the connection between the “mode of existence” of the technical object (Simondon 1989) and the regimes of engagement that determine and promote certain modes of action, hierarchies, and forms of representation in the social world (Introna 2011).
