“The dangers of open-access science in a pandemic.” Preprint servers make it easy for scientists to share academic research papers before they are peer-reviewed or published, and COVID-19 is leading to a flood of research being uploaded. That can be a good thing, getting new and cutting-edge research into decision-makers’ hands quickly, writes Gautama Mehta in Coda Story. But it can also spread misinformation.

As of Thursday evening, medRxiv (pronounced “med archive”), a preprint server run in partnership by BMJ, Yale University, and Cold Spring Harbor Laboratory, had 2,740 COVID-19–related papers.

It’s not as if just anyone can upload anything to a preprint server. There are screening processes in place, Diana Kwon reported in Nature last week, and those have been enhanced in light of COVID-19:

BioRxiv and medRxiv have a two-tiered vetting process. In the first stage, papers are examined by in-house staff who check for issues such as plagiarism and incompleteness. Then manuscripts are examined by volunteer academics or subject specialists who scan for non-scientific content and health or biosecurity risks. BioRxiv mainly uses principal investigators; medRxiv uses health professionals. Occasionally, screeners flag papers for further examination by Sever and other members of the leadership team. On bioRxiv, this is usually completed within 48 hours. On medRxiv, papers are scrutinized more closely because they may be more directly relevant to human health, so the turnaround time is typically four to five days.

Sever emphasizes that the vetting process is mainly used to identify articles that might cause harm — for example, those claiming that vaccines cause autism or that smoking does not cause cancer — rather than to evaluate quality. For medical research, this also includes flagging papers that might contradict widely accepted public-health advice or inappropriately use causal language in reporting on a medical treatment.

But during the pandemic, screeners are watching for other types of content that need extra scrutiny — including papers that might fuel conspiracy theories. This additional screening was put in place at bioRxiv and medRxiv after a backlash against a now-withdrawn bioRxiv preprint that reported similarities between HIV and the new coronavirus, which scientists immediately criticized as poorly conducted science that would prop up a false narrative about the origin of SARS-CoV-2. “Normally, you don’t think of conspiracy theories as something that you should worry about,” [medRxiv and bioRxiv cofounder Richard Sever] says.

These heightened checks and the sheer volume of submissions has meant that the servers have had to draft in more people. But even with the extra help, most bioRxiv and medRxiv staff have been working seven-day weeks, according to Sever. “The reality is that everybody’s working all the time.”

Confess I didn't fully appreciate level of work folks at preprint servers do, balancing the imperatives of sharing info & preventing harm ahead of peer review in this crisis.

As some1 who spends 65% of her days reading #COVID19 lit, TY to all involved🙏🏻https://t.co/Gvz61H8VrD

— Vanessa Sung | 宋宜臻, PhD 👩🏻‍🔬👩🏻‍💻🇨🇦🇹🇼 (@sung_vanessa) May 13, 2020

MedRxiv has a disclaimer at the top of the search page: “medRxiv is receiving many new papers on coronavirus SARS-CoV-2. A reminder: these are preliminary reports that have not been peer-reviewed. They should not be regarded as conclusive, guide clinical practice/health-related behavior, or be reported in news media as established information.” But publications don’t always heed that guidance — as in the case of a much-tweeted (and then much-criticized) LA Times article claiming there was a new mutant strain of the virus, for instance.

One of the issues that factors into media coverage of preprints is that the journalists covering the coronavirus are not always science reporters. [Fiona Fox, head of the UK’s Science Media Center] told me that many of the people now reporting about preprint studies have been taken off their usual beats and “have no idea what peer review is and have no idea what a preprint is, and are having to cover this because there’s no other story in town.”

This plays into another problem posed by preprint servers: they are essentially dumps of information which require scientific expertise to adjudicate or contextualize. “Everything comes out as it’s received,” [Derek Lowe, who covers the pharmaceutical industry], told me. “There is no way to know what might be more interesting or important, and no way to find it other than by using keyword searches. It really puts people back on using their own judgment on everything at all times, and while that should always be a part of reading the literature, not everyone is able to do it well.”

Jonathan Gitlin wrote for Ars Technica earlier this month:

If a paper posted to arXiv regarding a particular flavor of subatomic particle turns out to be erroneous or flawed, no one’s going to die. But if a flawed research paper about a more contagious mutation of a virus in the middle of a global pandemic is reported on uncritically, then there really is the potential for harm.

Indeed, this is not an abstract fear. We are in the middle of a global pandemic, and a recent study in The Lancet found that much of the discussion (and even policymaking) about COVID-19’s transmissibility (also known as R0) during January 2020 was driven by preprints rather than peer-reviewed literature.

Nobody claims that the conventional peer-review process is perfect. And “a kind of de facto real-time peer review has emerged in the comment sections of preprint studies, as well as in discussions on Twitter,” Mehta notes. “These are precisely the places where large numbers of scientists gathered to discuss the flaws in the Indian study on similarities between the coronavirus and HIV before it was retracted.”

On the balance, preprints and early open sharing of results are still vastly preferable to sitting on data and results until a journal publishes it. We have seen that pre-publication review and debunking work.

But we need to maintain standards to improve signal-to-noise. 8/8

— Richard Neher (@richardneher) May 14, 2020

In the worst-case scenrios, though, scientific research may also be becoming a partisan weapon, Northeastern’s Aleszu Bajak and Jeff Howe wrote in The New York Times this week.

How coronavirus research is being weaponized: Focuses on how the right seized Stanford's Santa Clara study to support the protests against lockdowns and other social mitigation efforts. https://t.co/pmpd78sy5a

— Victoria Colliver (@vcolliver) May 14, 2020

The response we've gotten to our @nytopinion piece on weaponization of Covid-19 preprints has been incredible https://t.co/tUpmNwAY1Y

Thanks to @stuartathompson and @iarynam for edits and graphics. Before-and-afters below. And here's a Shiny w/ more data https://t.co/z4Dy174VKp pic.twitter.com/Tf6Vohuf3S

— Aleszu Bajak (@aleszubajak) May 14, 2020

Conspiracy theories and election disinformation on TikTok. Rolling Stone’s EJ Dickson found COVID-19–related conspiracy theories in abundance on TikTok:

Some of the most popular videos exist at the nexus of anti-vaccine and anti-government conspiracy theorist content, thanks in part to the heightened presence of Qanon accounts on the platform. One video with more than 457,000 views and 16,000 likes posits that Microsoft’s founding partnership in digital ID program ID2020 Alliance is targeted at the ultimate goal of “combining mandatory vaccines with implantable microchips,” with the hashtags #fvaccines and #billgates. Another popular conspiracy theory, among evangelicals in particular, involves the government attempting to place a chip inside unwitting subjects in the form of a vaccine. Some Christians view this as the “Mark of the Beast,” a reference to a passage in Revelations alluding to the mark of Satan. The #markofthebeast hashtag has more than 2.3 million combined views on TikTok, and some videos with the hashtag have likes in the tens of thousands.

On The Verge’s podcast this week, Alex Stamos, director of the Stanford Internet Observatory and former chief security officer for Facebook, talked about bad actors on TikTok (“If I was the Russians right now, I would put all of my money, all of my effort behind TikTok and Instagram”).

“Over one-quarter of the most viewed YouTube videos on COVID-19 contained misleading information.” Researchers in Ottawa screened the top COVID-19–related, English-language YouTube videos on March 21 and found that more than a quarter of them contained inaccurate information. The sample size here is small: The reachers started out with an original set of 150 videos, but after screening for duplicates, non-English language, a lack of audio, and so on, they had 69 videos to work with; those videos had been viewed over 62 million times. More than 25 percent of the videos contained “non-factual” information — including inaccurate statements (“A stronger strain of the virus is in Iran and Italy,” racism (“Chinese virus”), and conspiracy theories. “Government and professional videos” contained factual information but “only accounted for 11 percent of videos and 10 percent of views.”