I have a new paper out in the Colorado Technology Law Journal, The Law and Ethics of Experiments on Social Media Users. It’s the scholarly version of my work from last summer on the Facebook and OkCupid experiments. The basic argument should be familiar: running scientific experiments on users without their consent or institutional oversight raises serious ethical and legal concerns. But, thanks to the CTLJ and Paul Ohm’s December conference at the University of Colorado, When Companies Study Their Customers, I have taken the opportunity to revise and extend my remarks. It’s long for a symposium essay – 23,000 words – and I hope that it can also serve as a reference on last summer’s controversy.
Here’s the abstract:
If you were on Facebook in January 2012, there is a chance that it tried to make you sad. If you were on OkCupid, there is a chance that it tried to match you up with someone incompatible. These were social psychology experiments: Facebook and OkCupid systematically manipulated people’s environments to test their reactions. Academics doing similar experiments in a university setting would typically need to obtain informed consent from participants and approval from an Institutional Review Board (IRB). But Facebook and OkCupid, and the academics working with Facebook, had neither. This, I believe, is a problem.
These experiments offer us a moment for reflection, a chance to discuss the law and ethics of experiments on social media users. In this essay, I will consider social media research through the prism of the Facebook and OkCupid experiments. I will focus on three questions: (1) When do social media experiments constitute research involving people? (2) What does it take to obtain the informed consent of users? (3) What institutions are responsible for reviewing such experiments?
Part I offers an initial review of the Facebook and OkCupid research projects. Part II – the bulk of the essay – takes up these questions under current law. Part III considers the broader question of what the rules for regulating social media research ought to be. The most immediately pressing priority is to prevent the unraveling of the existing ethical framework through IRB laundering, in which a regulated institution outsources enough work to an unregulated one to evade IRB review and informed consent. Looking further ahead, I offer some tentative thoughts on the scope of coverage, informed consent, and oversight for social media experiments. Finally, the conclusion reflects on how we should think about consent in this setting.