Will Facebook Backlash Kill Scientists’ Data Dream?

0
107

Will Facebook Backlash Kill Scientists’ Data Dream?

Earth-lights-nasa

Image: NASA

The difference between Facebook users and lab rats? Lab rats can see who’s conducting the experiment.

Facebook’s 2012 experiment on 689,003 unsuspecting users has stoked controversy for breaking ethical norms that are widely recognized by academics and researchers. Facebook has apologized, though this is far from the first study that the social network has performed on its users.

But Facebook isn’t the only one watching. It has become the holy grail for academics studying social sciences, hosting a massive data set of personal interactions with minimal external influence. Studies can now be done on hundreds of thousands of subjects with little to no awareness of anybody observing them.

“There’s basically a new article coming out every day,” said Robert Wilson, a Ph.D. student at Washington University who has studied the use of Facebook data.

Facebook-Study

Image: Robert Wilson

Academics and Facebook are both wrestling with just how to use the massive dataset that the social network provides to effectively and ethically generate a greater understanding of topics like psychology and sociology. They will have to figure it out, said Michal Kosinski, deputy director of the Psychometrics Centre at Cambridge University, because the use of Facebook has already changed the disciplines.

“My opinion is that Facebook and other similar environments will be the future of social sciences. The time when you were getting 20 people into a lab and looking at what they were doing is over,” Kosinski said.

The academics interviewed by Mashable say that thanks to Facebook’s 1.3 billion users, researchers can perform groundbreaking studies on human nature.

As long as Facebook lets them.

Now, as Facebook backpedals on its study on emotion, some are worried that the company may begin to close off its data.

“[Some of] the biggest scientific studies ever run were Facebook studies,” Wilson said. “Now I’m kind of worried that it’s all going to fall apart because there’s this risk.”

The social network effect

In just a few years, Facebook has helped transform how researchers in social sciences conceive and execute studies.

“Suddenly a subset of psychologists realized, ‘Wait a minute. This really allows us an opportunity to study real psychological processes going on in the real world as they occur,'” said Sam Gosling, a professor of social psychology at the University of Texas.

The combination of scale and almost unnoticeable tracing makes Facebook an unbeatable platform for social research, though it isn’t the only place where this kind of data is available. Kosinski said data from Google searches, emails or even services like Spotify has predictive qualities that are proving valuable to researchers.

But none of them have the complete package like Facebook. Not only are classic social topics more easily studied, but Facebook has also helped open the door to areas that are much more difficult to research, although it still has shortcomings.

“Facebook data can provide researchers with information about who people choose to affiliate with, what kinds of disclosures they share with one another, and how relationships and networks form and dissolve over time,” said Nicole Ellison, a professor at the University of Michigan.

“But like all data, Facebook data have biases,” she said. “For instance we don’t choose to disclose some things to some people and we may not affiliate online with people we spend time with offline. So interpretation is important.”

Researchers have used Facebook data to predict postpartum depression in new mothers and assess language learning during study abroad trips. And the same researcher from Facebook’s emotion study even tried to chart “gross national happiness.”

But who watches the watchers?

The Facebook study on “emotional contagion” caused concern not just because users were watched without their knowledge, but also because imperceptible changes to people’s News Feeds could have influenced their thoughts. If Facebook can figure out methods to manipulate its millions of users, the theory goes, the ramifications become terrifying.

Sheryl Sandberg, Facebook’s No. 2 executive, sought to allay fears that the company could use this position to affect its users or pursue company interests.

“I want to be clear — Facebook can’t control emotions and cannot and will not try to control emotions,” she said on Wednesday.

“Facebook would never try to control elections,” she added.

The 2012 emotions study did not make major changes to the experience of the chosen Facebook users. It worked by “very minimally deprioritizing a small percentage of content in News Feed,” according to Adam Kramer, a Facebook researcher who worked on the study. Facebook says its research is done solely to improve its site, and testing changes on users is a common part of how companies optimize products.

Regardless of whether Facebook actually has control over user emotions — its own study seems to indicate that it does, while other studies show that positive News Feeds can generate sadness — the company is clearly in a position of immense power over its users. And it’s not just emotions — Facebook’s News Feed is already well known for its ability to direct people’s media consumption habits.

Facebook’s role as a platform for research is further complicated by the company’s own interests. Michael Corey, a quantitative researcher at Facebook and sociology Ph.D. candidate at the University of Chicago, wrote in a January blog post that, while Facebook pushes academic work, the studies are vetted:

While we are encouraged to publish, everything has to get approved by the communication and legal teams. It is a small price to pay for such an incredible data set; but we don’t do sharable academic data. This is less of an issue in sociology than in fields where open sourcing published data is standard.

The clash between Facebook’s priorities and the independence of studies creates concern about the studies that originate from within the company and its influence on studies that use its data.

Last year, Kosinski figured out that a variety of traits about people could be accurately predicted through Facebook likes, and he wrote a paper on his findings. He provided Facebook with a preview, which elicited legal threats from the social network, warning him not to publish it. He did so anyway and received more calls from Facebook’s lawyers. But Facebook’s own data team also reached out to Kosinski and lauded his work.

The closing book

Facebook has become increasingly cozy with academics (particularly sociologists), building an internal research operation that often works with outside academics.

Facebook also reaches out to academic communities, and researchers whom Mashable spoke with had a generally positive perspective on Facebook’s data operations, noting that data is anonymized to protect user privacy. Facebook is reportedly planning a meeting with researchers ahead of the American Sociological Association’s annual conference in August.

But researchers also said Facebook was in the wrong for how it handled its emotion study.

“They screwed up. You can’t not get consent from people,” Wilson said.

Observing people is one thing, said Kosinski, but experimenting with people without consent, as Facebook did, is over the line.

“I do think that Adam [Kramer, one of the researchers of the Facebook study] went too far with his research,” Kosinski said.

“I think that it’s OK to observe people as long as you observe what they’re doing, don’t try to identify them, don’t try to change anything in the environment,” Kosinski added.

Facebook’s internal research team has grown since the 2012 study to include more oversight, but new reports make clear that the emotion study was not the only one of its kind.

A Facebook spokesperson told Mashable that its internal research is designed to improve its product.

“Our research is designed to understand how people use Facebook and how we can make our services better for them,” the rep said.

The spokesperson, however, declined to comment on Facebook’s future plans for working with academics.

But if Facebook’s public relations problems continue, outside researchers fear they’ll lose considerable opportunities.

“If you look at it from their perspective, what do they have to gain from academics and publishing?” Wilson said.

Gosling noted that Facebook needs academic interaction to keep it in line with the generally recognized ethics of modern social sciences. If Facebook were to do only internal research, it could end up with even less oversight and fewer reasons to limit its social experiments.

“They have some interests but their priorities are not the same as scientific priorities,” Gosling said. “Scientists will want to publish even if it shows that Facebook is very bad for you or bad for society.”

Have something to add to this story? Share it in the comments.