x
A controversial policy change threatens to upend large social-media studies.
Twitter announced on 2 February that it would end free access to its API.Credit: Tayfun Coskun/Anadolu Agency via Getty
Akin Ünver has been using Twitter data for years. He investigates some of the biggest issues in social science, including political polarization, fake news and online extremism. But earlier this month, he had to set aside time to focus on a pressing emergency: helping relief efforts in Turkey and Syria after the devastating earthquake on 6 February.
Aid workers in the region have been racing to rescue people trapped by debris and to provide health care and supplies to those displaced by the tragedy. Twitter has been invaluable for collecting real-time data and generating crucial maps to direct the response, says Ünver, a computational social scientist at Özyeğin University in Istanbul.
So when he heard that Twitter was about to end its policy of providing free access to its application programming interface (API) — a pivotal set of rules that allows people to extract and process large amounts of data from the platform — he was dismayed. “Couldn’t come at a worse time,” he tweeted. “Most analysts and programmers that are building apps and functions for Turkey earthquake aid and relief, and are literally saving lives, are reliant on Twitter API.”
Twitter has long offered academics free access to its API, an unusual approach that has been instrumental in the rise of computational approaches to studying social media. So when the company announced on 2 February that it would end that free access in a matter of days, it sent the field into a tailspin. “Thousands of research projects running over more than a decade would not be possible if the API wasn’t free,” says Patty Kostkova, who specializes in digital health studies at University College London.
How Facebook, Twitter and other data troves are revolutionizing social science
Since then, researchers have been anxiously awaiting further details of Twitter’s plans. The date that free access is due to end has been pushed back twice: first on 8 February, and then in an announcement on 13 February that said the new platform would be unveiled in “a few more days”. Furthermore, key information about a proposed new payment plan is missing, says David Lazer, a political scientist at Northeastern University in Boston, Massachusetts. The company says that it will offer tiered pricing, with the cheapest version costing US$100 a month for “low-level” access. But it hasn’t said what that will entail. “People can have different orders of magnitude in their imagination as to what ‘low-level’ means,” says Lazer. “If they’re optimizing the pricing for corporate customers, there’s no way academics are going to be able to afford it.”
The rapid change in policy will cause difficulties for some European Union-funded projects, whose proposals were written with the assumption that API access would remain free, says Ünver. And prices could quickly mount if each researcher on a project had to pay the monthly fee.
Even if well-funded researchers find a way to get the data they need, the policy change will exacerbate inequalities for students or researchers from low-income countries, for whom the monthly fees could be too high to sustain long-term studies. “Twitter is a global platform, and this decision has global ramifications,” says Renee DiResta, research manager of the Stanford Internet Observatory at Stanford University in California. “Discontinuing free access will break free tools developed to democratize research.”
The move also runs counter to the spirit of the EU’s Digital Services Act, which came into force last November, says Claes de Vreese, who studies political communication at the University of Amsterdam. As a very large social-media platform, Twitter could eventually be compelled under the legislation to ensure data transparency and access for researchers.
When the API announcement came, researchers were already feeling unmoored as a result of changes at Twitter after billionaire Elon Musk bought the company last October. Ünver says that since then, his data requests using the API have yielded some worrisome glitches. For example, a program that uses earthquake-related keywords in Turkish doesn’t always return all of the tweets that he sees on his own Twitter feed — even if he searches with the exact words of the tweet. And a few times, he has spotted duplicate tweets in his data. “Both problems are new,” he says.
Restricting access could create further data difficulties. De Vreese says that his colleagues have been frantically collecting data since word of the API change broke. Over time, he says, limiting access to new information could render the field over-reliant on previously downloaded data sets.
How Musk’s takeover might change Twitter: what researchers think
Before the Twitter announcement, Lazer’s lab had gathered enough data to crank out a few academic papers, including one that looks at the impact of Twitter’s interventions, such as banning certain users, after a mob attacked the US Capitol on 6 January 2021. “In the longer run, none of that work can be replicated,” he says. And if another insurrection occurs, researchers will not be able to study it in the same way.
Some projects might simply grind to a halt, because other social-media platforms do not offer the same degree of access as Twitter, or the same quality of data. Kostkova and her colleagues have been searching for ways to keep working on their epidemic-intelligence research — in which they have used Twitter for more than a decade to study subjects including swine influenza, vaccine hesitancy and COVID-19. “At the moment there is nothing,” she says.
The situation also highlights how the availability of Twitter’s data has led the field to depend on the platform, says de Vreese. “This is a moment in time when we have to rethink,” he says. “It’s reigniting a conversation that we know we should be having: that there is too much Twitter-reliance in research.”
