25 March 2014, Proceedings of the National Academy of Sciences: Kramer, Facebook Core Data Science Team, et al: Experimental evidence of massive-scale emotional contagion through social networks (PDF)
Furor Erupts Over Facebook's Experiment on Users
Almost 700,000 Unwitting Subjects Had Their Feeds Altered to Gauge Effect on Emotion
By Reed Albergotti
June 30, 2014
A social-network furor has erupted over news that Facebook Inc., in 2012, conducted a massive psychological experiment on nearly 700,000 unwitting users.
To determine whether it could alter the emotional state of its users and prompt them to post either more positive or negative content,  the site's data scientists enabled an algorithm, for one week, to automatically omit content that contained words associated with either positive or negative emotions from the central news feeds  of 689,003 users.
The research, published in the March issue of the Proceedings of the National Academy of Sciences, sparked a different emotion--outrage--among some people who say Facebook toyed with its users emotions and uses members as guinea pigs.
"What many of us feared is already a reality: Facebook is using us as lab rats, and not just to figure out which ads we'll respond to but actually change our emotions," wrote Animalnewyork.com, a blog post that drew attention to the study Friday morning.
Facebook has long run social experiments. Its Data Science Team is tasked with turning the reams of information created by the more than 800 million people who log on every day into usable scientific research.
On Sunday, the Facebook data scientist who led the study in question, Adam Kramer, said he was having second thoughts about this particular project. "In hindsight, the research benefits of the paper may not have justified all of this anxiety," he wrote on his Facebook page.
"While we've always considered what research we do carefully," he wrote, Facebook's internal review process has improved since the 2012 study was conducted. "We have come a long way since then."
The impetus for the study was an age-old complaint of some Facebook users: That going on Facebook and seeing all the great and wonderful things other people are doing makes people feel bad about their own lives.
The study, Mr. Kramer wrote, was an attempt to either confirm or debunk that notion. Mr. Kramer said it was debunked.
According to an abstract of the study, "for people who had positive content reduced in their News Feed, a larger percentage of words in people's status updates were negative and a smaller percentage were positive. When negativity was reduced, the opposite pattern occurred."
The controversy over the project  highlights the delicate line in the social media industry between the privacy of users and the ambitions--both business and intellectual--of the corporations that control their data.
Companies like Facebook, Google Inc. and Twitter Inc. rely almost solely on data-driven advertising dollars. As a result, the companies collect and store massive amounts of personal information. Not all of that information can be used for advertising--at least not yet. In the case of Facebook, there is an abundance of information practically overflowing from its servers. What Facebook does with all its extra personal information--the data isn't currently allocated to the advertising product--is largely unknown to the public.
Facebook's Data Science team occasionally uses the information to highlight current events. Recently, it employed it to determine how many people were visiting Brazil for the World Cup. In February, The Wall Street Journal published a story on the best places to be single in the U.S., based on data gathered by the company's Data Science Team.
Those studies have raised few eyebrows. The attempt to manipulate users' emotions, however, struck a nerve.
"It's completely unacceptable for the terms of service to force everybody on Facebook to participate in experiments," said Kate Crawford, visiting professor at MIT's Center for Civic Media and principal researcher at Microsoft Research.
Ms. Crawford said it points to broader problem in the data science industry. Ethics are not "a major part of the education of data scientists and it clearly needs to be," she said.
Asked a Forbes.com blogger: "Is it okay for Facebook to play mind games with us for science? It's a cool finding, but manipulating unknowing users' emotional states to get there puts Facebook's big toe on that creepy line."
Slate.com called the experiment "unethical" and said "Facebook intentionally made thousands upon thousands of people sad."
Mr. Kramer defended the ethics of the project. He apologized for wording in the published study that he said might have made the experiment seem sinister. "And at the end of the day, the actual impact on people in the experiment was the minimal amount to statistically detect it," he wrote on Facebook.
Facebook also said the study was conducted anonymously, so researchers could not learn the names of the research subjects.
Mr. Kramer said that the content--both positive and negative--that was removed from some users' news feeds might have reappeared later.
The emotional changes in the research subjects was small. For instance, people who saw fewer positive posts only reduced the number of their own positive posts by a tenth of a percent.
Comments from Facebook users poured in Sunday evening on Mr. Kramer's Facebook page. The comments were wide-ranging, from people who had no problem with the content, to those who thought Facebook should respond by donating money to help people who struggle with mental health issues.
"I appreciate the statement," one user wrote. "But emotional manipulation is emotional manipulation, no matter how small of a sample it affected."
Facebook users agree to terms of service that give the company wide leeway in how it can treat them.
Write to Reed Albergotti at firstname.lastname@example.org