How do recommender systems learn political opinions? A semi-synthetic step-by-step experiment

Tim Faverjon, Jean-Philippe Cointet, Pedro Ramaciotti · 2026

Read How do recommender systems learn political opinions? A semi-synthetic step-by-step experiment by Tim Faverjon, Jean-Philippe Cointet, Pedro Ramaciotti (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13210316.txt.

Description

Recommendation engines quietly learn users' politics, and a 2026 experiment headed by Tim Faverjon shows the learning happen step by step. The research article trains a model on real sharing behavior from a panel of nearly forty thousand X users, then generates synthetic recommendations to trace how ideology gets encoded.

Elementary recommendation rules, fed only on who shared what, built an internal left-right map of users inside the system. That map formed independently of age and gender signals in the authors' checks on the model. Tweaking the ideological representation shifted the political leaning, diversity, and relevance of what users were offered, exposing a trade-off platforms rarely discuss openly.

Catalog record

CREATORTim Faverjon, Jean-Philippe Cointet, Pedro Ramaciotti
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13210316

Browse the full library · About the Tower of Babel