Twitter transformed science communication during the pandemic. Will it last? | Science

issue cover image

A version of this story appeared in Science, Vol 375, Issue 6587.

In January 2020, some 2 months before the World Health Organization declared the coronavirus pandemic a global emergency, a tweet appeared on virologist Benhur Lee’s smartphone. It linked to a website, virological.org, where scientists had just posted the genetic sequence of SARS-CoV-2. Lee, at the Icahn School of Medicine at Mount Sinai, quickly shared the tweet with his followers, along with the words “Here we go” and an animation of planes taking off. Within days, the pharmaceutical firm Moderna and the U.S. National Institutes of Health had announced plans to develop what just 10 months later proved to be an effective vaccine, based on the sequence that codes for the virus’ spike protein.

In an earlier age, it might have taken days or longer for such useful DNA data to reach interested scientists via a Table of Contents alert from a journal. But the rise of Twitter and other social media platforms enabled users like Lee to spread the word about the SARS-CoV-2 sequence within hours, sparking global conversations and accelerating efforts to develop vaccines and treatments.

It was an early sign of how the pandemic prompted many scientists—and the public—to turn to social media to share and learn about hot new findings. COVID-19 “changed the game” because the threat “immediately connects with the public, [so] there’s a much bigger natural audience” for information about pandemic science than for most areas of research, says Michael Thelwall, a data scientist at the University of Wolverhampton, City Campus, who studies social media. In particular, Twitter has become a go-to resource for anyone trying to make sense of the torrent of pandemic studies—and for those intent on quickly pushing back against misinformation.

“I like that there’s a low bar to entry [on Twitter]—I can put something out and see how other scientists are thinking of a problem, people who have a different skill set than mine,” says biostatistician Natalie Dean of Emory University, whose Twitter account has some 138,000 followers.

But the pandemic has also helped demonstrate the limitations of social media. It can be difficult, for example, for scientists to be heard over the cacophony of messages on Twitter—some 500 million each day. And although some scientists have used the platform to elevate their online presence, that has rarely translated into concrete professional rewards. Eventually the sizable Twitter followings some have built during the pandemic may fade. And in the meantime, some have suffered from their digital fame, attracting harsh personal attacks and threats of violence. Despite such challenges, many researchers believe that—like it or not—the pandemic has forever altered how certain scientists communicate with each other and the public.

The marriage of Twitter and science came later than researchers who study scholarly communication expected. A decade ago, many predicted that scientists would flock to social media as a complement to traditional channels such as email alerts, Google searches, and scientific meetings. But at first many researchers expressed apathy or disdain for Twitter, which debuted in 2006. Some saw little value in the platform’s large, nonscientific audience. Others bridled at the initially tight limits on message length—just 140 characters per tweet (now 280). As a result, studies before the pandemic suggested as few as 2% of published scientists, and no more than one in five researchers in the United States and Europe, had Twitter accounts. (The disparate results reflect differences in the studies’ methods, the authors say.) Researchers working in the health and social sciences were more likely to have an account. And scientists who did tweet typically drew little engagement.

Rapid responses

Conversations about new research papers on Twitter happen quickly—then soon tail off, like mentions in news stories. Citations in other journal articles, the more traditional channel for scholarly attention, are spread out over many more months.

(Graphic) K. Franklin/Science; (Data) L. Waltman et al., Scholarly Communication in Times of Crisis, Research on Research Institute, DOI: 10.6084/m9.figshare.17125394 (2021)

Researchers have been even more skeptical of the social media giant Facebook, which has a greater reach than Twitter among the general public. Instead, they gravitated toward online platforms geared to scientists, such as the citations manager Mendeley, which gives users details about how often a paper is bookmarked by other users.

But as the pandemic exploded and researchers sought to pump out information to each other and an eager public, many saw advantages to Twitter. Its vast reach became a draw: more than 200 million active daily users, including an estimated one-quarter of U.S. adults, according to the Pew Research Center. This allows scientists to use a single platform to share research findings with both peers and the public and to foster open discussions. Twitter’s interactivity and viral features added to its appeal—for example, users can amplify others’ tweets by “liking” them or retweeting them to their own followers.

One result is that the platform has carried posts about a majority of the total COVID-19 literature—about 51% of journal articles on pandemic research had been mentioned in at least one tweet through May 2021, according to a report by the Research on Research Institute (RoRI). That exceeds the number cited in scholarly articles or mentioned in several other communications venues, including news stories, Facebook, YouTube, Wikipedia, blogs, and policy documents. And it’s well above the level before the pandemic, when studies found that just 10% to 30% of papers on any scientific topic got a mention on Twitter.

One scientist who jumped headlong into tweeting about COVID-19 is Muge Cevik of the University of St. Andrews. Before the pandemic, her tweets about tuberculosis and HIV had attracted only about 2000 followers to her Twitter account. Now, she has more than 170,000, among them members of the U.K. Parliament, scientists, and journalists. “I never thought that I would get this much engagement,” she says. But she realized the “privilege … comes with a lot of responsibility”—for communicating respectfully and “distinguishing opinion versus facts based on data,” she says.

Cevik’s approach to tweeting is to synthesize the findings of several scientific papers in a way that researchers can use and nonscientists can understand. She does this by posting long threads—a linked series of tweets, often 20 or more, which Twitter has allowed since 2017. In 2020, for example, Cevik posted a thread about how poorly ventilated rooms increased the risk of transmitting the pandemic coronavirus, highlighting research that was driving government calls to pause many kinds of gatherings.

Tweet threads, such as this one from Muge Cevik, an infectious diseases researcher at the University of St. Andrews, allow scientists to expand on ideas beyond the 280-character limit for a single tweet.

“I mainly want people to understand that there’s a scientific basis to the [public health] decisions being made,” Cevik says. “There are so many published data, I think no one is able to follow them.” Cevik’s approach has proved popular: Other Twitter users have retweeted her more than 200 posts about COVID-19 research papers at least 12,000 times, placing her Twitter account in the top 50 on that measure, according to unpublished data provided to Science by Zhichao Fang of Leiden University (which he developed for RoRI’s report).

Other researchers are less enamored with Twitter’s threading capability. Dean, for instance, finds the process cumbersome and says it is difficult to embed scientific charts. She’s experimenting with posting longer essays on Substack, a newsletter platform. “It’s more readable and writable,” she says—but the content appears to reach fewer people.

For Cevik, however, Twitter threads have paid off. She expanded her thread on poor ventilation into a paper published in Clinical Infectious Diseases. Her posts were also retweeted by Harvard University’s Marc Lipsitch, who then invited her to collaborate, leading to three co-authored papers about COVID-19. That partnership “would have never happened prior to social media,” says Cevik, who also became a pandemic adviser to Scotland’s chief medical officer.

Beyond that, it’s difficult to quantify the impact of her tweets, Cevik says. But several readers told Cevik the threads helped persuade them to get inoculated. Cevik says she works to build such trust by spelling out what scientists don’t yet know about COVID-19 and answering questions from readers. “We can’t really deal with misinformation just by sharing facts,” she says. “We just need to use facts as the foundation for a strategic conversation.”

Even as some users have turned to Twitter to promote clarity, however, others have flooded the platform with misleading or false information, including tweets deliberately intended to confuse readers. Tweeters, including some scientists, have also helped elevate questionable findings, by linking to studies that were done quickly, had problematic designs, or were first published online with no peer review.

This double-edged sword can be seen in Fang’s list of the top 50 accounts whose posts about COVID-19 papers have drawn the most retweets. Besides Cevik’s account, the list includes others run by scientists who have tried to highlight the best available research, including those of Eric Topol, director of the Scripps Research Translational Institute, and several leading scientific journals, such as Science and the Nature family of journals.

But the same list also includes Didier Raoult, the controversial French microbiologist at the Hospital Institute of Marseille Mediterranean Infection who has authored a blizzard of tweets urging the use of hydroxychloroquine to treat COVID-19, despite little evidence that the drug is efficaceous.

Some scientists have repeatedly taken to Twitter to challenge Raoult’s views in a bid to curb the viral spread of misinformation in real time—a tactic some call “Twitter peer review.” These watchdogs say such rapid response is necessary because some manuscripts published only as preprints never get formally reviewed at all, and journals often take months to publish letters critiquing peer-reviewed papers. “Obviously, in a pandemic that’s far, far too long” to wait, says Gideon Meyerowitz-Katz, a Ph.D. candidate in epidemiology at the University of Wollongong. He has used Twitter to rapidly challenge scientific claims he believes are backed by little evidence.

A tweet by Edward Holmes of the University of Sydney at the pandemic’s outset helped kick off a global response by scientists.

In March 2021, for example, Meyerowitz-Katz used his “Health Nerd” account to challenge a paper that, contradicting the weight of scientific evidence, found no proof that lockdowns reduced COVID-19 deaths. Critics of lockdowns had ballyhooed the article, which was published by Scientific Reports. In a long thread published only 3 days after the paper appeared, Meyerowitz-Katz took issue with its mathematical assumptions. Soon after, Scientific Reports invited him to submit the critique for publication—but it didn’t appear until December 2021, 9 months after the original paper. A week later, after the journal published the critique, it retracted the lockdown study, which has been accessed online by nearly 400,000 people.

Meyerowitz-Katz says data from his Twitter account indicate his March thread was available immediately to tens of thousands of readers, vastly more than later accessed his journal letter. “There’s an immediacy to Twitter I don’t see in traditional academic formats,” he says.

Other COVID-19 papers that might have influenced public policy also drew critical peer review on Twitter before they were retracted. One wrongly claimed that the antiparasitic drug ivermectin could treat the disease; another drew the conclusion, contested by many scientists, that cotton face masks did not prevent transmission of the pre-Omicron variants of the virus.

Such instances remain rare. The Retraction Watch blog, for example, lists more than 200 COVID-19 papers and preprints that have been retracted or withdrawn. But Science found that only about a dozen of those had received Twitter peer review. And among 673 scientists who posted preprints about COVID-19, more than twice as many (65%) reported receiving colleagues’ comments about them via emails than via social media posts (30%), according to a 2021 RoRI survey.

Meyerowitz-Katz says such numbers might reflect the lack of professional rewards for scientists who use social media to critique papers, rather than taking a more traditional route by serving as a journal peer reviewer. Many scientists list their journal service, but not their tweeting, on CVs. Those who publicly challenge findings may also run career risks. Journal peer reviewers typically remain anonymous, he notes, but many academic tweeters list their names on their accounts. “You may be calling into question the results of people who make hiring and firing decisions in your field,” he says.

That kind of attention could harm a career. But an emerging body of research about tweeting suggests that, overall, scientists often struggle to be heard on social media. One study, for example, found tweets containing links to scholarly papers typically get little engagement. Of 1.1 million such tweets about papers published before the pandemic, half drew no clicks, and an additional 22% attracted just one or two, according to a 2021 paper in the Journal of the Association for Information Science and Technology.

Other research suggests scientists need to attract a critical mass of Twitter followers before their tweets will reach many nonscientists. A 2018 study in Facets that examined the Twitter accounts of 110 academic ecologists and evolutionary biologists found those with fewer than 1000 followers were read mostly by colleagues. But above that tipping point, policymakers, journalists, and other nonscientists made up a majority of followers. And scientists might have more success reaching nonscientists on Twitter if they tweet about news articles that mention a scientific paper, rather than the paper itself, according to forthcoming research by information scientist Stefanie Haustein of the University of Ottawa and colleagues.

Such studies are in the vanguard of nascent scholarly efforts to examine Twitter’s role in science communication. For now, Haustein says, “We are really not at the point where we want to get, which is, ideally, seeing the impact of research on the greater good of society.”

quotation mark

You may be calling into question the results of people who make hiring and firing decisions in your field.

  • Gideon Meyerowitz-Katz
  • University of Wollongong

Still, studies of Twitter use have revealed some intriguing trends. In a study published in Scientometrics in 2020, for example, Leiden University data scientist Rodrigo Costas Comesana and colleagues found that some Twitter accounts that spread antivaccine messages also tweet about the same, legitimate scientific papers as pro-vaccine accounts. Better understanding the behavior of these “bridging” tweeters could inform scientists’ strategies to engage with them and address their misrepresentation of facts, and help understand why their posts draw attention, the authors concluded.

As the pandemic wanes, scientists on Twitter may wonder whether the platform can also advance their own research or careers. The evidence is mixed. Before and during the pandemic, studies by Costas Comesana and others found little correlation between the attention a publication gets on social media and its number of citations—the usual measure of scholarly impact.

Many universities today consider scientists’ public service activities in professional evaluations as well, but there is no consensus about the role metrics involving social media should play in this. The Altmetric Attention Score debuted in 2012 as a way to summarize the attention drawn by a research paper across a variety of communication venues; in a single, weighted number, it amalgamates counts of Twitter retweets and likes and mentions in news stories and policy documents, for example. But university tenure and promotion committees haven’t considered the scores—nor should they, say scholars of social media metrics. The Altmetric score covers venues too disparate to be meaningful for tenure decisions, Haustein says. And attention does not always translate into social impact: A scientist with lots of Twitter followers may be ignored by policymakers.

Haustein suggests, though, that scientists consider strategically presenting metrics from Twitter or other social media to highlight papers that mattered to audiences outside of academe. And Dean notes that Twitter can offer researchers a way to build their professional reputations “by letting people know who you are and what you think.”

Even if social media metrics cannot replace citations and peer review to measure scholarly quality, Costas Comesana says, the new measures can serve as a complement and help researchers better understand how the public is perceiving, ignoring, or misunderstanding their work. “And that,” he says, “is where the thing gets more interesting.”