Meta Is Getting Rid of CrowdTangle — and Its Replacement Isn’t As Transparent or Accessible

Meta Is Getting Rid of CrowdTangle — and Its Replacement Isn’t As Transparent or Accessible
Tayfun Coskun/Anadolu Agency via Getty Images

The U.S. elections are less than five months away, and the public has less visibility than ever before into political messaging on Facebook and Instagram. 

In August, Meta is shutting down CrowdTangle, a popular social media monitoring tool used to track misinformation on Facebook and Instagram. The company says its replacement, the Meta Content Library (MCL), is a better tool for researchers. 

This article is copublished with:

Tow Center for Digital Journalism/Columbia Journalism School

But a joint investigation by Proof News, the Tow Center for Digital Journalism, and Algorithmic Transparency Institute found that Meta’s replacement tool is less transparent and accessible than CrowdTangle. On 11 key topics, Meta’s new tool has fewer features than CrowdTangle. And researchers say the process of getting access to the tool is cumbersome and slow — and some journalists can’t access it at all.

That means that journalists and researchers will have less ability to trace online information about the elections, of which over 60 are set to take place worldwide in 2024.

“The decision to deprecate the tool will have a direct effect on election integrity experts and their ability to do their work,”  said Claire Pershan, EU advocacy lead at the Mozilla Foundation. 

For more than a decade, journalists, researchers, and election observers have used CrowdTangle to monitor trends on Facebook and Instagram. The tool has helped journalists shed light on the popularity of right-wing figures on Facebook, monitored influencers during elections, and tracked health misinformation during the pandemic. Now, the rise of generative AI has raised further concerns about the potential proliferation of election misinformation and political candidate deepfakes on social media platforms. 

While the MCL offers some new features, like the ability to see how many people have viewed a public post and access to comments and short-form videos (an increasingly significant form of social content), it also lacks many of the features that make CrowdTangle a valuable research tool. These limitations include restrictions on exporting data, the inability to track post performance over time, the lack of direct links to public posts, and fewer features for collaboration. Nor does Meta’s MCL enable users to track how channels in a custom list are performing — a key feature on CrowdTangle that has been useful to researchers.

But data transparency isn’t the only issue, researchers say. Who gets access is just as important. Unlike CrowdTangle, which was open for anyone to use, the MCL is restricted to academic and nonprofit researchers, which excludes most news outlets. Given that academic research can be notoriously slow, the public could significantly lose out on the rapid-response stories enabled by journalists using CrowdTangle data.

CrowdTangle was founded by Brandon Silverman and Matt Garmur in 2011 and started attracting big media clients like BuzzFeed, CNN, Vox, USA Today, and NBC. Meta acquired it in 2016 and decided to make the tool free and expand access to more media partners. But in March of this year, Meta announced CrowdTangle will be discontinued after Aug. 14. 

In an open letter to Meta, dozens of nonprofits and newsrooms around the world called on the company to keep CrowdTangle functioning until January 2025 and to onboard to the MCL all current CrowdTangle organizations that are focused on election integrity. The letter has been signed by more than 170 individuals and organizations, including CrowdTangle’s former CEO, Silverman. 

But despite outside pressure to delay CrowdTangle’s closure, Meta spokesperson Erin McPike said the company is sticking to the original deadline. 

When asked at an MIT Technology Review conference why CrowdTangle is being discontinued during a big election year, Meta’s president of global affairs, Nick Clegg, claimed it’s because CrowdTangle is a “degrading tool.” CrowdTangle's inability to measure reach, according to Clegg, hampers its ability to show what’s actually going viral. “It only measures a narrow cake slice of a cake slice,” he said. Despite this, Meta has actively used and advocated for CrowdTangle for years — even giving election boards in all U.S. states access during the 2020 election to help quickly identify misinformation and voter interference and suppression. 

Meta says that its decision to discontinue CrowdTangle is an effort to meet data access requirements set by the European Union’s Digital Service Act (DSA). European law requires that large social media companies like Meta put in place effective real-time election-monitoring tools. 

But the European Commission has indicated that the MCL does not meet its requirements. In April, the commission initiated formal proceedings against Meta into potential violations of the DSA.  

One concern: timing. With several European elections on the horizon, the commission stated that deprecating a tool like CrowdTangle could damage civic discourse and electoral processes. There is currently no “adequate replacement” for CrowdTangle, the commission stated in a public statement

In response, Meta added country-specific dashboards in CrowdTangle for each EU country but said it still planned to stop supporting CrowdTangle in August. 

“We have a well-established process for identifying and mitigating risks on our platforms. We look forward to continuing our cooperation with the European Commission and providing them with further details of this work,” Meta said in a statement, as reported by the BBC.

None of this addresses the exclusion of for-profit press organizations from access to the new tool. And researchers who are eligible to apply for access to the MCL API — which allows programmatic queries of the data — say the application process can be time-consuming and cumbersome. 

Researchers must submit their application through Meta’s independent partner ICPSR at the University of Michigan — a process that requires approval from an institutional review board (IRB), institutional signatures, and evidence of technical skills. (These steps are not required for the web version of the MCL application, which allows researchers to explore data through a user-friendly graphical interface but without the same level of control and flexibility as the API.) 

Some researchers’ applications were stuck in limbo for a month — sometimes longer — according to Josephine Lukito, assistant professor at the University of Texas Austin’s School of Journalism and Media. 

“It has been really damaging both to my work, as well as that of graduate students I work with and other researchers I know who are studying the U.S. elections,” Lukito said.  

Meta initially released the MCL and its associated API as a beta version in 2023, allowing a select group of researchers to test it. Lena Frischlich, an associate professor at the Digital Democracy Centre of the University of Southern Denmark, was one of them. According to Frischlich, the technical focus of the MCL application could pose hurdles to some researchers with no programming knowledge.

“Assuming I apply as a Principal Investigator with a less technical, for example, humanistic background, I need to hire a data scientist,” Frischlich said. “But CrowdTangle was accessible even for someone with zero programming skills.” 

While there are differences between the two tools, Frischlich said she is encouraged by the open dialogue between Meta’s developer team and researchers. 

“I do see that some of the things we discussed during that time have resulted in substantial changes in data access,” she said. For instance, Meta added a feature that lets researchers analyze comments on public posts — something that was never available on CrowdTangle. This feature, along with the ability to download specific subsets of data, was later added to the library.

But none of the content of the MCL matters, other researchers say, if no one can access it. 

“It’s not primarily about data points, it’s about access,” said Pershan, the EU advocacy lead from Mozilla. While CrowdTangle was accessible to tens of thousands of people, the MCL might potentially be very limited, she said. 

Fabio Giglietto, an associate professor of internet studies at the University of Urbino, in Italy, was also among the early testers of the Meta’s MCL.

“In my opinion, the Meta Content Library seems to be more modern,” Giglietto said. “But it does have limits. In my specific use case, it prevents me from performing certain tasks that I can currently do with CrowdTangle.” 

One of these limits, Giglietto said, is the restrictions on data exporting. Giglietto is working on Vera AI EU, a disinformation research project that taps into CrowdTangle data to detect coordinated link sharing. The results are automatically uploaded to a Google Sheet three times a day and handed off to fact-checkers and disinformation experts. But this kind of automated export is currently not permitted on the MCL, Giglietto said, which makes it more difficult to share results with outsiders.  

CrowdTangle’s accessibility allowed New York Times reporter Kevin Roose to track the rise of far-right media on Facebook. With Giglietto’s assistance, Roose set up an automated Twitter account in 2020 called @FacebooksTop10 that showed the sources of the most engaged link posts by U.S. pages based on CrowdTangle data. The daily listing was regularly dominated by partisan news outlets and right-wing commentators like Ben Shapiro and Dan Bongino. It became an embarrassment for Facebook executives who, according to Roose, were left to explain “the disparity between what they thought Facebook was — a clean, well-lit public square where civility and tolerance reign — and the image they saw reflected in the Twitter lists.” 

Giglietto pointed to the @FacebooksTop10 bot as an example of something that would no longer be possible with the MCL in its current form. 

Meta spokesperson McPike said the MCL provides more comprehensive data overall than CrowdTangle: “Although downloads for all public content is not supported in the MCL’s User Interface, MCL has more comprehensive data that spans a wider range of accounts and profiles than CrowdTangle.” 

Getting rid of CrowdTangle will produce ripple effects throughout the broader research community. Journalists, civil society researchers, and academics have built infrastructure on top of CrowdTangle that relies on its stream of data to function. The Illuminating project — which provides real-time analysis of the U.S. presidential candidates’ X and Facebook accounts — is just one example. According to Illuminating Research lead Jennifer Stromer-Galley, the corporations who facilitate our public town squares have a civic obligation to enable journalists and researchers to help make the discussions, and the actors who facilitate them, transparent. “The health of our democracy depends on it,” she wrote in an email. 

Another example of a tool that depends on CrowdTangle is Hamilton 68, an online influence tracker that uses network analysis to detect Russian propaganda and disinformation efforts. Hamilton 68 may stop running altogether once CrowdTangle is put to rest, the Columbia Journalism Review reported

Republish This Article