It’s the end of an era for social media research. Meta has shut down CrowdTangle, the analytics tool that for years helped tens of thousands of researchers, journalists and civil society groups understand how information was spreading on Facebook and Instagram.
For a company that’s never been known for being transparent about its inner workings, CrowdTangle was an “invaluable” resource for those hoping to study Meta’s platform, says Brandi Geurkink, the executive director for the Coalition for Independent Technology Research. “It was one of the only windows that anybody had into how these platforms work,” Geurkink tells Engadget. “The fact that CrowdTangle was available for free and to such a wide variety of people working on public interest journalism and research means that it was just an invaluable tool.”
Over the years, CrowdTangle has powered a staggering amount of research and reporting on public health, misinformation, elections and media. Its data has been cited in thousands of journal articles, according to Google Scholar. News outlets have used the tool to track elections and changes in the publishing industry. It’s also provided unparalleled insight into Facebook itself. For years, CrowdTangle data has been used by journalists to track the origins of viral misinformation, hoaxes and conspiracy theories on the social network. Engadget relied on CrowdTangle to uncover the overwhelming amount of spam on Facebook Gaming.
Meta wasn’t always quite as averse to transparency as it is now. The company acquired CrowdTangle in 2016, and for years encouraged journalists, researchers and other civil society groups to use its data. Facebook provided training to academics and newsrooms, and it regularly highlighted research projects that relied on its insights.
But the narrative began to shift in 2020. That’s when a New York Times reporter created an automated Twitter bot called “Facebook Top Ten.” It used CrowdTangle data to share the top Facebook pages based on engagement. At the time, right-wing figures and news outlets like Dan Bongino, Fox News and Ben Shapiro regularly dominated the lists. The Twitter account, which racked up tens of thousands of followers, was often cited in the long-simmering debate about whether Facebook’s algorithms exacerbated political polarization in the United States.
Meta repeatedly pushed back on those claims. Its executives argued that engagement — the number of times a post is liked, shared or commented on — is not an accurate representation of its total reach on the social network. In 2021, Meta began publishing its own reports on the most “widely viewed” content on its platform. Those reports suggested that spam is often more prevalent than political content, though researchers have raised significant questions about how those conclusions were reached.
More recently, Meta executives have suggested that CrowdTangle was never intended for research. “It was built for a wholly different purpose,” Meta’s President of Global Affairs, Nick Clegg, said earlier this year. “It just simply doesn’t tell you remotely what is going on on Facebook at any time.” CrowdTangle founder Brandon Silverman, who has criticized Meta’s decision to shut down the service ahead of global elections, told Fast Company it was originally meant to be a community organizing tool, but quickly morphed into a service “to help publishers understand the flow of information across Facebook and social media more broadly.”
Clegg’s explanation is a “retcon,” according to Alice Marwick, principal researcher at the Center for Information Technology and Public Life at University of North Carolina. “We were trained on CrowdTangle by people who worked at Facebook,” Marwick tells Engadget. “They were very enthusiastic about academics using it.”
In place of CrowdTangle, Meta has offered up a new set of tools for researchers called the Meta Content Library. It allows researchers to access data about public posts on Facebook and Instagram. It’s also much more tightly controlled than CrowdTangle. Researchers must apply and go through a vetting process in order to access the data. And while tens of thousands of people had access to CrowdTangle, only “several hundred” researchers have reportedly been let into the Meta Content Library. Journalists are ineligible to even apply unless they are part of a nonprofit newsroom or partnered with a research institution.
Advocates for the research community, including CrowdTangle’s former CEO, have also raised questions about whether Meta Content Library is powerful enough to replicate CrowdTangle’s functionality. “I’ve had researchers anecdotally tell me [that] for searches that used to generate hundreds of results on CrowdTangle, there are fewer than 50 on Meta Content Library,” Geurkink says. “There’s been a question about what data source Meta Content Library is actually pulling from.”
The fact that Meta chose to shut down CrowdTangle less than three months before the US presidential election, despite pressure from election groups and a letter from lawmakers requesting a delay, is particularly telling. Ahead of the 2020 election, CrowdTangle created a dedicated hub for monitoring election-related content and provided its tools to state election officials.
But Marwick notes there has been a broader backlash against research into social media platforms. X no longer has a free API, and has made its data prohibitively expensive for all but the most well-funded research institutions. The company’s owner has also sued two small nonprofits that conducted research he disagreed with.
“There is no upside to most of these platforms to letting researchers muck around in their data, because we often find things that aren’t PR-friendly, that don’t fit the image of the platform that they want us to believe.”
While CrowdTangle never offered a complete picture of what was happening on Facebook, it provided an important window into a social network used by billions of people around the world. That window has now been slammed shut. And while researchers and advocates are worried about the immediate impact that will have on this election cycle, the consequences are much bigger and more far reaching. “The impact is far greater than just this year or just work related to elections,” Geurkink says. “When you think about a platform that large, with that much significance in terms of where people get their sources of information on a wide array of topics, the idea that nobody except for the company has insight into that, is crazy.”