Image: Grace Gay for Research Professional News
Drastic changes at Twitter have “demolished” online social research, and academics are fleeing the platform
Two years ago, the world’s richest person, Elon Musk, was mostly known for Tesla electric cars and founding SpaceX with a view to colonising Mars.
A year ago, he bought Twitter—now called X—for roughly $44 billion (€41bn), after changing his mind about wanting to own the platform and then changing it back again.
Since then, Musk’s leadership of the company has mirrored the chaotic nature of that acquisition. A wave of measures— including changing its name, shedding about 80 per cent of its staff, restoring thousands of banned accounts and making users pay for verification symbols, also known as blue ticks—has left the social media giant looking almost like a different entity.
For many, Musk’s changes have become a reason to flee the site. Anecdotal evidence suggests researchers are no exception. But this leaves the question of how academics can wean themselves off the platform, which had become a crucial way for them to network and engage the public. For many, Twitter had become a big part of working life—including as an ‘altmetric’, used as a proxy for research impact.
Research ‘demolished’
In the academic community, scientists researching online behaviour have been particularly affected. When Twitter launched in 2006, it made its application programming interface—which allows third-party developers to gather data—free and publicly available. This gave researchers a powerful tool to monitor conversations on the social media platform. Academics have used the site to study the spread of misinformation online, and even to help respond to natural disasters (see box on P10).
But Musk put a price on the API service. X now offers a “basic” API service for $100 per month, a “pro” one for $5,000 per month and an “enterprise” version at $42,000 a month.
“The new API changes have really demolished the way we do social research online,” says Akin Unver, a computational social scientist at Ozyegin University in Turkey. “It has really messed up entire master’s theses, PhD dissertations, research programmes.”
He tells Research Europe that the problem is not only that researchers now have to pay, but that even if they do, the API service is “spotty”, with chunks of data missing.
Twitter previously had more than 100 in-house engineers working on a specific academic track for the API, Unver says. But Musk’s sudden staff layoffs decimated the team. There were similar cuts in API management and data product service management. Unver says so many people were let go “that the product does not work properly anymore”.
Where do we go from here?
The changes have put an end to some people’s research, with those who had been awarded grants on the basis they would use the API hit particularly hard.
“Let’s say a researcher applied for a big grant in the EU and it is a three-year project. They promised to use a certain amount of API [data] and all of a sudden, that API is gone,” Unver says.
“They don’t have the budget [to adapt]because they never predicted this change would happen and they cannot ask for the extra budget. What happens is that stream of your research gets destroyed and ends right there.”
Unver, who studies conflict, political violence and diplomatic escalation, says the API changes have affected two of his research programmes, one financed by the EU’s funding scheme Horizon Europe and the other backed with a grant from the Turkish government.
For Unver, no other social media platform’s data are as good for conducting research. The changes at X have forced him to switch from the quantitative approach he was planning on using with the API service to a qualitative approach known as digital ethnography.
The new method is more time consuming and more detailed, he says, and will lead to different results, as it will reflect a smaller pool of the population.
Silver linings
Some researchers, though, see a potential, if accidental, upside in making Twitter data less accessible to researchers.
Sacha Yesilaltay, a researcher working on misinformation and mistrust at the Digital Democracy Lab at the University of Zurich, Switzerland, argues that the site is not representative of the wider population and that too much research was focused on the platform.
For example, even in France, Germany and Spain—the EU countries with most users, according to January 2023 figures from data gathering site Statista—only around a fifth of people are active on the site.
“There were too many generalisations made based solely on work done on Twitter,” says Yesilaltay. “In a way, I am happy that people are moving away from Twitter and I don’t think it is a totally bad thing.”
A departure from X could have a positive effect if it motivates researchers to develop tools to better study other platforms, he says.
His lab is working on tools to use TikTok data for research. As it is a video-content platform, it is less straightforward to carry out large studies on TikTok compared with the text-based Twitter, he explains.
“I think [the changes to Twitter] are awful too but it may have good and unintended consequences which is moving away from Twitter data and developing other tools to study other platforms.”
Terminal decline
As well as the academics who use Twitter as a research tool, many more use it as a way of finding things out, networking and shooting the breeze.
Mark Carrigan—an education researcher at the University of Manchester who has written a book on social media for academics—told Research Europe that, pre-Musk, Twitter gave academics a one-stop-shop to network with each other while also connecting with those outside the scientific community.
“One of the things that made Twitter important was the scale of the network, the ease with which interactions would take place and content would travel across institutional boundaries.
“If you are trying to do external networking it was, at its peak, very powerful because your networks would intersect with media networks, policymaker networks and third sector parties.”
Musk’s changes—including a focus on paid visibility, promoting posts by users who pay a monthly subscription—have made the platform less useful and rewarding for many.
Yesilaltay, for example, says the algorithm is behaving in a “weird” way, serving him an inexplicable number of posts about cats and dogs.
Carrigan notes that such shifts mean users see fewer posts from those they are following. Harassment and unpleasantness have also increased, he adds. All these factors have caused many academics to flee the platform. Carrigan also believes that some who still have accounts now log in much less regularly.
Unver agrees that Musk’s takeover has affected academics’ use of X for networking, with an increase in trolls, bots and fake accounts. He puts the rise of disinformation and inauthentic accounts on the site again down to the mass firings, saying Musk “broke the original system”.
“The networking function of X is in terminal decline,” Carrigan says. “Information flows in less productive ways [and academics] are not having spontaneous and easy interactions with the diverse people that they once were.”
Splinter factions
There are other platforms that academics seem to be turning to as X declines. These include the business and employment-focused site LinkedIn, the so-called decentralised social media platform Mastodon, and the invite-only Bluesky, which was set up within Twitter in 2019 before being spun out.
Carrigan is concerned that users will be scattered across these and other sites. “It is just going to be a lot harder for an academic to accumulate that visibility if you are having to post across lots of different sites,” he says.
“Effectively, to get a high public profile an academic would have to become a professional communications practitioner in their spare time. I don’t think that is tenable given academic workloads.”
Even if X users migrate to one site over another, or the old Twitter’s functions return, Carrigan thinks academics’ ability to use social media to have a wider influence will still be hampered.
“On Twitter, the [networking and public engagement] functions merged together for a lot of people in a fairly natural way. In this social media environment, it is going to be much more difficult to have an effective and broad public presence that is capable of making an impact.”
Academics are not alone in being affected by the changes. This summer, the asset manager Fidelity—one of the company’s few remaining external investors—estimated that X’s stock value had plummeted by more than half since Musk’s takeover.
In June, Musk stepped down as chief executive of X. He was replaced by Linda Yaccarino, formerly head of advertising at the US TV network NBC, nicknamed the “Velvet Hammer” for her tough negotiating style. Musk stayed on as chair and chief technology officer.
Time will tell whether these changes at the company make life any easier for academics scrambling to move their research forward with the changed API, or for those wondering how they will make connections with other researchers or the public.
We asked X to a comment on a range of issues in this piece but did not receive a response before publishing.
Life-saving tweets
Besides using Twitter’s API to research public opinion and misinformation online, scientists have also used it to save lives.
In February this year, a series of devastating earthquakes killed over 55,000 people in Turkey and Syria, injuring a further 100,000.
As rescue workers raced to find those trapped underneath rubble, data scientists in Turkey harvested information from social media to help the rescue and relief efforts.
Using the API, data scientists were able to figure out the locations, needs, genders and age groups of many of the group of the missing and create a map for rescuers using an AI-based system, says Akin Unver, who was part of the team of scientists behind the effort.
“Before the API mechanism started to break down slowly, Turkish data scientists used Twitter data to map rescue and relief efforts during the February earthquake in Turkey, which was extremely significant,” he says.
“A lot of the data science efforts ended up saving lives.”
This article also appeared in Research Europe