Germain Gauthier

Publications

• The Political Effects of X’s Feed Algorithm
  (with Roland Hodler, Philine Widmer & Ekaterina Zhuravskaya)
  Nature, 2026
  appendix, replication files, policy brief

  Abstract

  Feed algorithms are widely suspected to influence political attitudes. However, previous evidence from switching off the algorithm on Meta platforms found no political effects. Here we present results from a 2023 field experiment on Elon Musk’s platform X shedding light on this puzzle. We assigned active US-based users randomly to either an algorithmic or a chronological feed for 7 weeks, measuring political attitudes and online behaviour. Switching from a chronological to an algorithmic feed increased engagement and shifted political opinion towards more conservative positions, particularly regarding policy priorities, perceptions of criminal investigations into Donald Trump and views on the war in Ukraine. In contrast, switching from the algorithmic to the chronological feed had no comparable effects. Neither switching the algorithm on nor switching it off significantly affected affective polarization or self-reported partisanship. To investigate the mechanism, we analysed users’ feed content and behaviour. We found that the algorithm promotes conservative content and demotes posts by traditional media. Exposure to algorithmic content leads users to follow conservative political activist accounts, which they continue to follow even after switching off the algorithm, helping explain the asymmetry in effects. These results suggest that initial exposure to X’s algorithm has persistent effects on users’ current political attitudes and account-following behaviour, even in the absence of a detectable effect on partisanship.

• Relatio: Text Semantics Capture Political and Economic Narratives
  (with Elliott Ash & Philine Widmer)
  Political Analysis, 2023
  appendix, replication files

  Abstract

  Social scientists have become increasingly interested in how narratives—the stories in fiction, politics, and life—shape beliefs, behavior, and government policies. This paper provides an unsupervised method to quantify latent narrative structures in text documents. Our new software package relatio identifies coherent entity groups and maps explicit relations between them in the text. We provide an application to the U.S. Congressional Record to analyze political and economic narratives in recent decades. Our analysis highlights the dynamics, sentiment, polarization, and interconnectedness of narratives in political discourse.

• Les déterminants de la mobilisation des “Gilets Jaunes”
  (with Pierre Boyer, Thomas Delemotte, Vincent Rollet & Benoît Schmutz)
  Revue Economique, Vol. 71 - 1. Jan. 2020
  policy brief, column

  Abstract

  Cet article présente les résultats d’une étude sur les territoires dont sont originaires les « Gilets jaunes », au début de la mobilisation. Dès le premier samedi de mobilisation, le 17 novembre 2018, ce mouvement se démarque par son caractère local et sa couverture nationale. À partir de données inédites de la mobilisation sur Facebook, nous montrons une forte corrélation entre mobilisation online (sur Facebook) et mobilisation offline (blocages des ronds-points). Nous réalisons alors une cartographie fine et contrastée de la contestation. L’étude économétrique met en évidence le rôle de la mobilité pour expliquer les origines du mouvement, au travers notamment du passage des routes à 80 km/h et des distances domicile-travail.

Working Papers

• Automating Inequality? Gender Bias in AI-mediated Labor Markets
  (with Kenza Elass, Debora Nozza & Paola Profeta)
  Accepted at Proceedings of the National Academy of Sciences

• Cycling through Elections: The Political Consequences of the Tour de France
  (with Ala Alrababah, Maelle Delouis-Jost & Adam Polak)
  R&R at American Political Science Review

  Abstract

  Support for far-right parties has grown in recent years, fueled in part by economic decline and neglect in left-behind places. Can a brief but highly salient event reverse these trends? We study the Tour de France, which provides a positive and salient boost combining economic activity with non-material benefits such as national visibility and collective experience. Using variation in the route between 2002 and 2022, we show that exposed communes experience declines in far-right support and increases in mainstream support of 0.03–0.05 standard deviations (SD). The estimates exceed 0.15 SD in recent elections and are largest in communes associated with left-behind places. While we cannot cleanly isolate the mechanism, we provide suggestive evidence consistent with a non-material channel. Local economic effects are small, electoral responses are larger when French riders win stages, and survey evidence collected around the 2025 Tour points to changes in how residents perceive their local communities.

• The Lifecycle of Protests in the Digital Age
  (with Pierre Boyer, Yves Le Yaouanq, Vincent Rollet & Benoît Schmutz)
  R&R at Journal of the European Economic Association

  Abstract

  We propose a theory of the emergence, size, intensity, and duration of modern protest movements. Moderates and radicals are both needed to sustain large coalitions, but when radicals resort to violence, they drive moderates away. Social media can spark a mass movement by revealing the potential for protest, but also precipitate the movement’s demise by enabling radicals to coordinate. We illustrate this phenomenon with the 2018 French Yellow Vests uprising. Online mobilization initially helped organize large, peaceful protests, but those triggered another wave of more radical online activity, causing moderates to leave the movement when exposed to radical content.

• Measuring Crime Reporting and Incidence: Method and Application to #MeToo
  replication files

  Abstract

  A long-standing issue with crime statistics is that they reflect both variations in the propensity to report crimes and actual crime incidence. I propose a two-way fixed effects model that exploits the delay between an incident and its police report to separate these two margins. I then study the Me Too movement’s impact on sex crimes in New York City. While reporting was already rising before October 2017, #MeToo led to higher victim reporting and arrest probabilities. Using non-sexual crimes as a control group, difference-in-differences estimates suggest the movement also deterred sex crimes.

Work in Progress

• Deep Latent Factor Models for the Social Sciences
  (with Philine Widmer & Elliott Ash)
  slides, code

• Neural Ideal Point Models
  (with Hugo Subtil & Philine Widmer)
  code

• Do politicians put their votes where their mouths are?
  (with Caroline Le Pennec & Vincent Pons)
  slides

• On Differences in Social Media Consumption
  (with Roland Hodler, Katharina Kaeppel, Alberto Ornaghi & Philine Widmer)

• Algorithmic Amplification of Politics on Social Media
  (with Philine Widmer & Ekaterina Zhuravskaya)

• Changing recommender algorithms
  (with Philine Widmer)
  in preparation for the Cambridge Handbook of Digital Manipulation