Skip to main content

Some of the Best Data on Human Behavior Is Privately Held by Social Media Companies

How can universities convince them to share that information?
The Facebook icon on a smartphone screen.

The Facebook icon on a smartphone screen.

In 2007, Duncan Watts quit his job as a sociology professor at Columbia University to join Yahoo!. To many of his colleagues and friends, the move was an odd one. "People thought, even back then, that Yahoo! was sort of a fading giant," he said on Monday, speaking on a panel organized by the Social Science Research Council in New York City.

But Yahoo!'s research lab turned out to be a perfect fit. At the time of his job switch, Watts, who had grown increasingly reliant on big data in his own research, felt like Columbia didn't understand the overlaps between social and computational science. (There's now a whole field called computational social science.) At Yahoo! Research, it seemed as if the computer scientists appreciated his methodology and objectives from the onset.

"They had understood, even back then, that the mountains of data they were collecting were essentially social data," he said. They recruited social scientists like Watts to work with them. (Watts now works at Microsoft Research.)

While Watts speaks glowingly of his career transition, his enthusiasm also belies what's become a troubling trend in the social sciences: The migration of researchers away from the ivory tower and over to the office park.

For generations, the social sciences—fields such as psychology and economics, which illuminate how people behave, spend money, live healthfully, and hurt or help each other—were primarily conducted and funded through universities and government agencies. That meant the sum of America's knowledge about human behavior was available to most researchers and to the government. But, according to a new report from the Social Science Research Council, as more and more social scientists join tech companies, any accumulated data about human behavior is used not to solve societal problems—say, encouraging vaccinations, or protecting citizens from falling for conspiracy theories—but rather to pump up companies' profits.

"We'll miss a big opportunity, if we don't find ways to have that knowledge be directed toward both scientific progress and the public good," says Ron Kassimir, executive program director at the non-profit Social Science Research Council. The council's panel on Monday featured experts, including Watts, who were unaffiliated with the report.

While Monday's panel lacked any specific solutions to the data problem, many speakers called broadly for "more collaboration" between universities, industry, the government, and research-funding foundations. Such initiatives would, however, surely run into obstacles, Watts reminded, as company lawyers are often reluctant to release company data to outsiders. That said, the Social Science Research Council is already working on one big cooperative project with industry. It's running an innovative program to let university researchers conduct studies using proprietary Facebook data. The studies are supposed to examine how Facebook affects elections and democracies. In a bid to keep the work independent from Facebook influence, researchers don't have to let the company see the results before they're published, a condition that Facebook had never until this point agreed to. (That program was announced shortly after news broke that a political firm hired by then-presidential candidate Donald Trump's campaign, Cambridge Analytica, gained access to private data about millions of Facebook users.)

Watts thinks now is the time to encourage companies to share data they might otherwise be reluctant to make public. "The world, society, the political sphere, are increasingly looking at technology companies with skepticism," he said. "This provides, I think, one point of leverage to say, 'We, the social science community, can help with this credibility, this trust crisis.'"