Parution du livre de Florian Jaton, chercheur à l'Institut des sciences sociales et STS Lab aux éditions MIT Press
Issu notamment de la thèse de Florian Jaton menée entre le LADHUL et le STS Lab, ce travail ethnographique décrit la manufacture d’algorithmes en traitement du signal.
A laboratory study that investigates how algorithms come into existence.
Algorithms—often associated with the terms big data, machine learning, or artificial intelligence—underlie the technologies we use every day, and disputes over the consequences, actual or potential, of new algorithms arise regularly. In this book, Florian Jaton offers a new way to study computerized methods, providing an account of where algorithms come from and how they are constituted, investigating the practical activities by which algorithms are progressively assembled rather than what they may suggest or require once they are assembled.
Drawing on a four-year ethnographic study of a computer science laboratory that specialized in digital image processing, Jaton illuminates the invisible processes that are behind the development of algorithms. Tracing what he terms a set of intertwining courses of actions sharing common finalities, he describes the practical activity of creating algorithms through the lenses of ground-truthing, programming, and formulating. He first presents the building of ground truths, referential repositories that form the material basis for algorithms. Then, after considering programming's resistance to ethnographic scrutiny, he describes programming courses of action he attended at the laboratory. Finally, he offers an account of courses of action that successfully formulated some of the relationships among the data of a ground-truth database, revealing the links between ground-truthing, programming, and formulating activities—entangled processes that lead to the shaping of algorithms. In practice, ground-truthing, programming, and formulating form a whirlwind process, an emergent and intertwined agency.