Facebook Added ‘Research’ To User Agreement 4 Months After Emotion Manipulation Study
Unless you've spent the last couple of days in a Faraday pouch under a rock, you've heard about Facebook's controversial ‘emotion manipulation’ study. Facebook data scientist Adam Kramer ran an experiment …
Context & Ripple Effects
Facebook's week of damage control continues. On June 29, Forbes reported that the company dismissed the public outcry over its emotion manipulation study, saying it did not understand the fuss, while data scientist Adam Kramer issued his own additional statement addressing the controversy. The experiment itself — manipulating what users saw in their News Feeds to test emotional contagion — was published earlier this month in an academic journal, which is what turned an internal test into a public event.
Today's disclosure sharpens that story: Facebook added a 'Research' clause to its user agreement roughly four months after the experiment ran, meaning users were enrolled under terms that did not yet name research explicitly. The pickup has been unusually broad for a policy footnote — Business Insider, the New York Times, Gigaom, ReadWrite, PandoDaily, New Statesman, Valleywag and TODAY all carried it on or about June 30 — which signals how sensitive consent-by-default has become as a flashpoint.
First-order effects
- Facebook's roughly 1.3 billion users now operate under a data use policy whose 'Research' wording postdates the very study that provoked the backlash, giving critics a concrete legal-text exhibit rather than just an ethics complaint.
- Kramer and Facebook's Data Science group are personally exposed: with the company publicly unapologetic, the researcher's follow-up statement becomes the de facto response, and scrutiny shifts onto how hundreds of similar feed experiments were consented to.
Second-order effects
- Academic collaborators who publish with platform data scientists face pressure to justify their institutional review processes, since the journal publication tied universities to an experiment run without explicit user consent.
- Advertisers and partners evaluating Facebook's data products now have to weigh whether experimental findings were produced under contested consent terms, putting a reputational discount on the Data Science group's output until governance catches up.
Third-order effects
- If the pattern holds, large-scale A/B testing at consumer platforms will be forced toward formal human-subjects-style review — consent language written into terms of service before studies run, not retrofitted after backlash.
- The episode establishes a template other social platforms will have to answer for: once one network's experimentation practices become a named scandal, every competitor's undisclosed feed tests become a standing regulatory and press question.
The trend: Platform experimentation is shifting from an opaque internal data-science practice to one measured against human-subjects consent standards, with terms-of-service wording becoming the battleground.