There have been quite a few statements by various scientists on Twitter who, in pointing to some preprint paper’s untenable claims, point to the manuscript’s identity as a preprint paper as well. This is not fair, as I’ve argued many times before. A big part of the problem here is bad journalism. Bad preprint papers are a problem not because their substance is bad but because people who are not qualified to understand why it is bad read it and internalise its conclusions at face value.
There are dozens of new preprint papers uploaded onto arXiv, medRxiv and bioRxiv every week making controversial arguments and/or arriving at far-fetched conclusions, often patronising to the efforts of the subject’s better exponents. Most of them (at least according to what I know of preprints on arXiv) are debated and laid to rest by scientists familiar with the topics at hand. No non-expert is hitting up arXiv or bioRxiv every morning looking for preprints to go crazy on. The ones that become controversial enough to catch the attention of non-experts have, nine times out of then, been amplified to that effect by a journalist who didn’t suitably qualify the preprint’s claims and simply published it. Suddenly, scores (or more) of non-experts have acquired what they think is refined knowledge, and public opinion thereafter goes against the scientific grain.
Acknowledging that this collection of events is a problem on many levels, which particular event would you say is the deeper one?
Some say it’s the preprint mode of publishing, and when asked for an alternative, demand that the use of preprint servers be discouraged. But this wouldn’t solve the problem. Preprint papers are a relatively new development while ‘bad science’ has been published for a long time. More importantly, preprint papers improve public access to science, and preprints that contain good science do this even better.
To making sweeping statements against the preprint publishing enterprise because some preprints are bad is not fair, especially to non-expert enthusiasts (like journalists, bloggers, students) in developing countries, who typically can’t afford the subscription fees to access paywalled, peer-reviewed papers. (Open-access publishing is a solution too but it doesn’t seem to feature in the present pseudo-debate nor does it address important issues that beset itself as well as paywalled papers.)
Even more, if we admitted that bad journalism is the problem, as it really is, we achieve two things: prevent ‘bad science’ from reaching the larger population and retain access to ‘good science’.
Now, to the finer issue of health- and medicine-related preprints: Yes, acting based on the conclusions of a preprint paper – such as ingesting an untested drug or paying too much attention to an irrelevant symptom – during a health crisis in a country with insufficient hospitals and doctors can prove deadlier than usual. But how on Earth could a person have found that preprint paper, read it well enough to understand what it was saying, and act on its conclusions? (Put this way, a bad journalist could be even more to blame for enabling access to a bad study by translating its claims to simpler language.)
Next, a study published in The Lancet claimed – and thus allowed others to claim by reference – that most conversations about the novel coronavirus have been driven by preprint papers. (An article in Ars Technica on May 6 carried this provocative headline, for example: ‘Unvetted science is fuelling COVID-19 misinformation’.) However, the study was based on only 11 papers. In addition, those who invoke this study in support of arguments directed against preprints often fail to mention the following paragraph, drawn from the same paper:
… despite the advantages of speedy information delivery, the lack of peer review can also translate into issues of credibility and misinformation, both intentional and unintentional. This particular drawback has been highlighted during the ongoing outbreak, especially after the high-profile withdrawal of a virology study from the preprint server bioRxiv, which erroneously claimed that COVID-19 contained HIV “insertions”. The very fact that this study was withdrawn showcases the power of open peer-review during emergencies; the withdrawal itself appears to have been prompted by outcry from dozens of scientists from around the globe who had access to the study because it was placed on a public server. Much of this outcry was documented on Twitter and on longer-form popular science blogs, signalling that such fora would serve as rich additional data sources for future work on the impact of preprints on public discourse. However, instances such as this one described showcase the need for caution when acting upon the science put forth by any one preprint.”
The authors, Maimuna Majumder and Kenneth Mandl, have captured the real problem. Lots of preprints are being uploaded every week and quite a few are rotten. Irrespective of how many do or don’t drive public conversations (especially on the social media), it’s disingenuous to assume this risk by itself suffices to cut access.
Instead, as the scientists write, exercise caution. Instead of spoiling a good thing, figure out a way to improve the reporting habits of errant journalists. Otherwise, remember that nothing stops an irresponsible journalist from sensationalising the level-headed conclusions of a peer-reviewed paper either. All it takes is to quote from a grossly exaggerated university press-release and to not consult with an independent expert. Even opposing preprints with peer-reviewed papers only advances a false balance, comparing preprints’ access advantage to peer-review’s gatekeeping advantage (and even that is on shaky ground).