Facebook announces new research guidelines

Facebook announces newThree months after inciting the ire of researchers and users with their stealth emotional manipulation study, Facebook has announced a revised research policy. The new policy addresses four key areas:

  • Clearer guidelines for researchers
  • An additional layer of review
  • Enhanced training for new Facebook employees around research
  • Increased transparency via a page compiling all Facebook research

First, I think it’s great that Facebook took the time to address the controversy in an organized manner. I also really like some of these changes, especially the new webpage where anyone can see the research Facebook has done. And I will only rarely complain about additional research methods and ethics training (when it gets to be redundant or distracting, which does not seem the case here).

However, the new research policy fails to address what I still see as the key problem with the original study: Participants never provided informed consent and were unable to opt out without relinquishing use of the product. As far as I can tell, Facebook intends to continue to conduct research with only their standard user agreements as consent.

My suggestion to Facebook is to create an opt-in research panel for users, where they are walked through a clear, easy-to-understand explanation of what research they might participate in. Users can agree to participate in ongoing research without being alerted to each individual study. Users who decide not to participate initially can update their preference at any time in their settings. Users who do opt in can renew their agreement annually. This arrangement would cover most standard research Facebook might do and would not impose significantly more burden on anyone than already existing Facebook requests.

I’ve been asked why Facebook should be held to different standards than other consumer companies which essentially test on their users all the time to optimize products. This is a hard question. I have a few answers but recognize none of them truly closes the issue:

  • Research done with the intention of contributing to science (versus advancing product sales) should adhere to the norms of science.
  • The Facebook study used a psychological measure (emotions) as its dependent variable. Most product optimization research looks at behaviors (buying) as the dependent variable. That said, at least Facebook inferred emotional state from observable phenomena rather than a more intrusive method.
  • My weakest response but the one I am most emotionally vehement about: Research ethics can  be a slippery slope and so we must hold the line. If there is a case to be made to loosen research protocols for online research, then it should be made to the professional organizations that govern human subjects research. It’s not the responsibility of any one research team or the company that employs them to make that decision.

Facebook has done well to revisit their research policy and to make it more public. I hope they will take the time to consider issues of informed consent within a broader human subjects ethics context, and either adopt a procedure to obtain consent or make the appropriate appeals to the research community for a change.