Algorithms, the so-called black box powering the internet and our every interaction there, is no longer an obscure element. While the isolated look at algorithms in particular and software in general depicts them as a non-human element, the understanding around algorithms cannot be any worse than it is right now.
The advancement around artificial intelligence and data science has made algorithms even more pervasive. Increasingly, the most popular field of study for students abroad is also around algorithms. Florian Jaton, Postdoctoral Researcher at the STS Lab at the University of Lausanne, has now shed some light on the human side of algorithms.
Algorithms should not be studied from outside
An algorithm is defined as a set of “rules that must be followed when solving a particular problem.” While this general definition explains how algorithms work within a predefined set of rules, Florian Jaton explores the concept from the inside. He also does not go the popular route of working backwards from a working algorithm and understanding how it came to be.
Instead, Jaton starts from people, desires, documents, and curiosities, and studies how all these entities come together and interact to form algorithms. His book, The Constitution of Algorithms: Ground-Truthing, Programming, Formulating, explores all the small and important details that go into creation of an algorithm.
Through his book, he also shows how humans and algorithms impact each other. The Constitution of Algorithms can help researchers, software designers and programmers align the future of software with that of human values.
“When I started to be interested in the topic of algorithms in 2013, there was already substantial, and generally quite critical, literature on the social effects of algorithms,” Jaton told The Next Web. “These important works studied the ways in which algorithms acted on our lives, while also emphasising algorithms’ opacity,” he adds.
Algorithms underpin a number of software and services that we use everyday. Their depiction, however, could be counterproductive and Jaton says he found studying algorithms from outside to be the cause. This methodology of studying algorithms from outside, he writes, filters out the “fragile scaffolding that had previously contributed to their progressive shaping.”
Training in Science & Technology Studies
Jaton says his training in Science & Technology Studies (STS), a sub-field of social science, was useful in studying algorithms from the inside. He adds that one of the basic postulates of STS is to consider “techno-scientific devices as the products of situated and accountable practices.”
“A possible remedy to the disarming critical discourse on algorithms seemed then to lie in a drastic change of method, privileging the anthropology of science as framed, for example, by the in situ ethnographic works of Bruno Latour, Michael Lynch and Lucy Suchman, over distant document analysis (that yet remains important).”
Jaton then spent two-and-a-half years working as part of a team of research scientists working on a computer vision algorithm. During this time, he took part in and documented discussions, data gathering efforts, programming sessions, code debugging practices, and refinement of theories. Throughout this process, he realised that some of the important work is disregarded while being studied in social settings.
“The invisibility of the practices underlying the development of algorithms can indeed no longer be considered positive: as they are the object of repeated disputes, it is now certainly important, or at least interesting, to document the practical processes that enable them to come into existence,” he writes in The Constitution of Algorithms.
Ground-truthing, Programming and Formulation
The book cleverly explains the concept of Ground-truthing, Programming and Formulation. Before developing an algorithm, a group of computer scientists, researchers, or engineers go through a process of problematization and ground-truthing. The researchers use this phase to precisely define the problem they want to solve and determine the type of data they need to validate their algorithm.
Once the problem is formulated, they must establish the ground truth by collecting the right information that will enable them to verify their algorithms and the models. After this, the algorithm moves to the programming phase, where a set of modules and list of instructions are created to solve the defined problem. It is then verified against the ground truth.
While the phase of programming might seem entirely technical and related to source code, Jaton argues in his book that “there is much to programming than just putting together a list of computer instructions.”
Once an algorithm is implemented and verified against the ground truth, it is formulated into a mathematical object that can be used in other algorithms. An algorithm, Jaton explains in his book, is designed to stand the test of time and prove useful in other scientific and applied work.