Facebook's mood study: How you became the guinea pig

That controversial research into how posts affect users' emotions is just latest in a long line of privacy flaps -- and apologies -- for the social networking giant.

Seth Rosenblatt Former Senior Writer / News
Senior writer Seth Rosenblatt covered Google and security for CNET News, with occasional forays into tech and pop culture. Formerly a CNET Reviews senior editor for software, he has written about nearly every category of software and app available.
Ian Sherr Contributor and Former Editor at Large / News
Ian Sherr (he/him/his) grew up in the San Francisco Bay Area, so he's always had a connection to the tech world. As an editor at large at CNET, he wrote about Apple, Microsoft, VR, video games and internet troubles. Aside from writing, he tinkers with tech at home, is a longtime fencer -- the kind with swords -- and began woodworking during the pandemic.
Seth Rosenblatt
Ian Sherr
6 min read


When news spread over the weekend that Facebook had manipulated its news feed to study how social media posts affect people's emotions, the real surprise was that anyone was that surprised.

The study (PDF), published in the Proceedings of the National Academy of Sciences and conducted by Facebook researcher Adam Kramer, Jeffrey Hancock of Cornell University, and Jamie Guillory of the University of California at San Francisco, found that people tended to reflect the emotional state of their friends' posts.

So if your friends wrote happy posts, your posts in turn tended to be happier. Friends posting about a bad day at work would tend to bring you down.

The disclosure triggered a sharp backlash and elicited an attempt by Facebook to seek forgiveness -- one in a long line of mea culpas the company has issued over the years. Yet anyone paying close attention to the boilerplate disclaimers that tech companies regularly publish might have seized upon a couple of seemingly innocent-sounding phrases tucked away in the company's data use policy that spoke volumes.

Among other things, Facebook says quite clearly in the published document that it might use the information it gathers from people on its social network "for internal operations, including troubleshooting, data analysis, testing, research and service improvement."

The tandem phrases "data analysis" and "research" appear to be unique to Facebook's user legalese. They do not appear in Google's terms of service and privacy policy, while "research" does appear in Yahoo's privacy policy but not its terms of service. LinkedIn is open in its privacy policy about the research it conducts on its users.

Google and Yahoo did not respond to a request for comment on whether they perform similar research on their users.

Line crossed?

However, privacy advocates warn that use agreements are carefully worded by design.

"There is no word in any privacy policy that is not there for a reason. If something is missing, then it's missing for a reason," said Brian Pascal, an attorney and privacy researcher at the University of California Hastings College of Law in San Francisco. He added that while there may or may not be a "practical impact" by Facebook's specific policy phrasing, "it's certainly interesting."

"It's one thing for Facebook to A/B test some advertising structure," he said, referring to internal tests that websites frequently conduct to determine what resonates with visitors -- some people see one set of ads, some see another. "It's another to tweak their News Feed to manipulate [users'] emotional state."

That goes to the heart of the latest argument about whether Facebook crossed a red line. It's one thing to test whether Facebook users search more, thus generating more revenue, when presented with more or fewer links. For many people, though, it's quite something else when the company tests whether users' emotional state can be altered artificially.

Carey Sebera, an attorney who's worked on Facebook privacy cases, said that while the research may not have violated the law or even the company's own policies, Facebook ought to have been more ethical. She voiced concerns raised by other critics when she noted that legal documents don't necessarily equate with morality and that something that's legal isn't necessarily ethical.

Responding to the blowback, Kramer posted a brief response defending the project as one of many attempts by Facebook "to provide a better service" and "never to upset anyone."

"Nobody's posts were 'hidden,' they just didn't show up on some loads of Feed," he wrote on Sunday.

"I can understand why some people have concerns about it, and my coauthors and I are very sorry for the way the paper described the research and any anxiety it caused," Kramer subsequently wrote. "In hindsight, the research benefits of the paper may not have justified all of this anxiety."

That response did little to mollify critics after they learned that of the 689,003 people experimented on in the study, which is less than 0.06 percent of Facebook's 1.2 billion users, none was aware of their participation. While that's considered standard operating procedure in business, Princeton University professor and privacy expert Ed Felten noted "the large gap" that exists between the ethical standards of industry practice, versus the research community's ethical standards for studies involving human subjects.

"Industry practice allows pretty much any use of data within a company, and infers consent from a brief mention of 'research' in a huge terms-of-use document whose existence is known to the user," according to Felten. But if people voluntarily give Facebook their data and the company operates within its own terms of service and privacy policy, the upshot is that Facebook can do with that information what it likes.

This is not the first time Facebook has run experiments on its users. In 2010, the company created an "I Voted" button, not unlike a "like" button, that displayed who among a user's friends had indeed voted. Facebook said in 2012 that it believed more than 300,000 voters turned out at the polls as a result of the study.

Facebook's actions are also bound to raise more questions about the power that large Internet companies wield and what they're doing with user data. It's part of a larger struggle within Silicon Valley over consumer privacy: What's the proper balance between enticing customers to use a service and then packaging that information for advertisers to help them to better target ads at their users?

Act first, apologize later

It's been a bumpy path. For instance, when Google in 2010 launched one of its first social-networking efforts, called Buzz, the company came under fire for weak privacy settings. Chief among them was a default setting that published a list of names that Gmail users most frequently emailed or chatted with. Most recently, a European court has forced Google to allow users to request that certain information about them be removed from the company's databases, including search results.

In 2012, Facebook settled with the Federal Trade Commission over charges that the company deceived customers by telling them they could keep information on the network private, but then allowing it to be made public. Facebook promised to offer users a "clear and prominent notice" when their information is shared beyond standard privacy settings. Whether or not Facebook violated that agreement with this research experiment remains unclear.

There have been other instances when Facebook inadvertently fanned privacy fears among its many users.

In 2006, the introduction of the news feed without sufficient privacy controls brought a public apology from CEO Mark Zuckerberg, who acknowledged "we really messed this one up." Zuckerberg had to again apologize a year later after the debut of a controversial product called Beacon that let Facebook friends see what you were doing on partner websites. Zuckerberg said Beacon was conceived as a way "to let people share information across sites with their friends," but he acknowledged that Facebook "simply did a bad job with this release, and I apologize for it." (Facebook subsequently offered a way to opt out of Beacon before dumping the project entirely.)

That congenital tone-deafness has Felton urging more stringent ethical rules on scientific research that would require specific, fully informed consent from the subject, including the right to opt out of a study at any point without penalty.

Considering how many people now use Facebook, the company sees clear benefits in more research about their likes and dislikes. The company has often acted first and then apologized and amended its policies after public outcry and Pascal thinks this situation will be no different.

"A definite first step is [for Facebook] to become much more transparent about how they decide to conduct research on their users," he said. "The truth is that Facebook has access to data that nobody else has. The answer can't be that Facebook must give up on its research. What we want is some degree of accountability and transparency when they do undertake research."