Richard Smith: Journals, fraud, science, and misaligned incentives

richard_smith_2014Journals, like the mass media, have a major part to play in exposing scientific fraud and other kinds of misconduct. In contrast, as I’ve argued many times, there are better ways now to disseminate science. Yet sadly and ironically, exposing fraud is risky and expensive, whereas publishing science is often highly profitable. The incentives are all wrong. Can they be changed?

I’ve been prompted to write this blog after answering an email from a researcher who has spent many months trying to get an article published that exposes serious fraud. And I’ve recently been involved with another case where months have been spent trying to achieve publication. In my 25 years in medical publishing, I experienced many similar cases.

Imagine the editor reading a paper that purports to expose major scientific fraud. Her/his first thought is: “Can this possibly be true?” This is especially the case if the person or organisation accused of misconduct is important, prestigious, and rich. I’ve argued before that just as celebrities got away with sexual abuse of children for decades, so famous researchers can get away with fraud for a long time—indeed, perhaps forever. The case of R K Chandra, “the father of nutritional immunology,” provides an example.

Many editors, particularly those from smaller journals, will simply decide that they do not have the resources to manage through to publication such a paper.

The editor on the larger journal will try to peer review the paper, but will be anxious about sending the paper to reviewers in case that might count as “publication,” potentially triggering a libel action. If she/he consults the journal’s lawyers or publishers they are likely to advise that “the risk of an expensive libel action doesn’t justify the potential return.”

There will be questions on whether the journal has an insurance policy to cover libel actions. If it doesn’t (and many, perhaps most, don’t) then the risk becomes huge. If it does, the insurers may need to be informed, again prompting the advice that “the risk of an expensive libel action doesn’t justify the potential return.”

But let us suppose our bold editor persists. Finding neutral reviewers may be difficult, and the truly neutral may not be useful as they may not be able to confirm or deny what is alleged. Other reviewers may be supporters or enemies of the accused, so helpful reviews may be hard to find.

If, eventually, the journal reaches the stage that it wants to publish, the editor will send the article to the libel lawyer. She/he will want evidence that what is alleged could be proved in a court of law and, unfortunately, there is a big gap between something being true and it being possible to prove it’s true. So the paper might be ditched at this point. Or much time and expense might be needed to try and produce enough evidence that the lawyer will think there a good chance that the allegations could be proved.

The lawyer is also likely to want changes in wording to reduce the chances of a successful libel action. Actions may sometimes be decided on just a few words. The hapless and long suffering author may resent the changes proposed, and the editor has to arbitrate.

The next difficulty might be a discussion around “trial by media.” If serious accusations are made against a person or an organisation, then ideally there should be a legally legitimate inquiry with due process with both accusers and defenders having a say. Journals can’t provide legitimacy and due process, leading to intense editorial debate about whether it is right to publish or whether the accusations should be referred to a legitimate authority. The editors might decide not to do so—because the legitimate authorities have had their chance and failed, nobody has any confidence that the authorities will act, or because it’s a “great story.”

Editors can, of course, allow the accused to respond, but this is a risky process—because the accused might apply for an injunction, blocking publication.

If and when the paper is finally published all sorts of consequences may follow: libel actions, complaints to owners or media regulators, and abuse from supporters of the accused (if the accused is a charismatic fraud he may have many ardent supporters). Libel actions can run for years, consume all the time of editors, and cost a fortune, possibly bankrupting the journal. Actions can also be lost even if everything alleged is true, or you finally win and can never recover your costs because the accused flees or is bankrupt.

So you can see why editors and journals are reluctant to publish accusations of fraud and misconduct—even though that may be the most important role of the media.

In contrast, publishing science is quick, easy, and lucrative. The value is in the research, which is sent free to the journal. After peer review of dubious value and light copy editing the science can be published. Subscriptions or author charges bring in money, and if the research happens to be a trial of a new drug or device then hundreds of thousands (even millions) might be made in sales of reprints. It doesn’t much matter if the research is wrong or misleading because that’s true of much of what’s published.

Could the incentives ever be changed? It’s hard to see how. I foresee a future in which journals are no longer the route for publishing science. At that stage most of them will disappear, but a few should remain to digest the science that is posted elsewhere, promote debate of important issues, and expose fraud and misconduct. If there are only a few then perhaps people will pay, but it seems likely that they will need support from research foundations, potentially setting up uncomfortable conflicts of interest.

What is clear is that it will continue to be easy to publish weak science, but hard to publish well argued and supported accusations of fraud and misconduct.

Richard Smith was the editor of The BMJ until 2004. 

Competing interests: None declared.