Understanding Algorithms: A Critical Analysis of Big Data's Societal Impact

Extract

[...] Cardon emphasizes that our dependence on these technologies structures our ways of thinking and interacting with the world, reinforcing a form of cognitive and decision-making delegation to machines. This dynamic, he argues, requires an epistemological critique of the very foundations of these systems: on what data do they rely? What types of knowledge do they produce? And what are the implications for our understanding of reality?" In the face of these challenges, Dominique Cardon advocates for a critical and interdisciplinary method. He proposes approaching algorithms through a "critical radiography", a approach aimed at analyzing their internal mechanisms, external effects, and the social logics they impose. [...]

[...] For Cardon, this approach is not just an academic exercise: it is a political necessity in a world where algorithms increasingly guide our choices, often without our awareness. 2. The four families of numerical calculations Dominique Cardon distinguishes four major families of algorithms, each reflecting a specific logic of information processing and classification. These families translate diverse approaches to organizing the digital world, influencing user perceptions and behaviors : - Popularity metrics: These algorithms, such as those of Médiamétrie, count clicks or views to hierarchize content based on its popularity. They reflect a quantitative logic where importance is attributed to the most frequently consulted content. [...]

[...] - Predictive algorithms: These systems, such as Amazon recommendations, analyze digital footprints to anticipate future user behavior based on past data. These four families illustrate different conceptions of truth and value, each highlighting a particular approach to digital reality. Popularity metrics prioritize the critical mass, reflecting a "democratic" approach where each click has the same weight. In contrast, authority rankings value peer recognition, often biased in favor of already dominant actors. Reputation metrics, on the other hand, focus on social interactions and relational dynamics, while predictive algorithms translate a prospective logic where individual behavior is anticipated from collective actions." These tensions reveal a fundamental epistemological issue: algorithms do not simply represent a pre-existing reality, but produce normative representations that influence choices and behaviors. [...]

[...] Cardon does not just expose the technical stakes; he highlights the urgent need for collective awareness and regulation of digital technologies. Ultimately, "What Algorithms Dream Of" invites us to question the place of these tools in our societies and to challenge the idea of an inevitable machine domination over humanity. [...]

[...] Their functioning relies on massive data capture and processing logics that hierarchize, classify, and orient behaviors." Cardon explains that algorithms are not just simple technical tools. They translate the choices, intentions, and values of their creators, and these choices are far from being neutral. For example, determining which data is relevant or defining evaluation criteria implies underlying human subjectivity, often influenced by cultural, economic, or political factors. In this sense, these systems do not only reflect the world as it is, but actively participate in its construction by selecting, valuing, or excluding certain realities. The "society of calculations" raises deep epistemological issues. [...]