Breaking the Algorithmic Contract (ssir.org)

Jose Marichal’s new 146‑page book, You Must Become an Algorithmic Problem (Bristol University Press, 2025), reframes our relationship with platforms as an “algorithmic contract”: an unspoken social bargain where people trade curational autonomy and the relief from the anxiety of choice for comfort, predictability, and voice. Marichal argues that platforms don’t just model human behavior—they monetize it, turning every swipe, like, eye movement or post into nodes and edges for predictive models. Because predictability equals profit, market incentives push platforms to optimize for users who are easier to classify and to encourage habitual, immediate “voice” over reflective expression, effectively training people to be more algorithmically legible. For the AI/ML community this is both a technical and ethical alarm bell. Marichal highlights the “outlier problem”: where classical model parsimony met limits, modern ML answers by scaling parameters and data, blurring the line between model and lived reality and incentivizing the production of more predictable data. That creates feedback loops—datafied behavior becomes training material that further narrows behavior—while obscuring issues like sampling bias, overfitting, and the political economy of attention. The book calls for a renegotiated socio‑technical contract—changes in design, evaluation metrics, and governance that protect curational autonomy and resist incentives that convert human complexity into predictable commodities.