An incident in the career of the late and sorely missed Angelika Amon illustrates an ugly truth about scientific publishing.

The scientific world is still mourning Professor Angelika Amon, who recently died from cancer, aged just 53. A native of Vienna, Amon did her PhD with Kim Nasmyth in the early days of the Institute of Molecular Pathology (IMP), before moving on to a glittering solo career in the USA. The number of tributes and obituaries published since her death is a testament to the extraordinary impact she had not just on research, but on the people around her and in the wider scientific community.

Of those eulogies, Kim Nasmyth’s is probably the most moving, and highlights one episode in her early career that will strike a chord with many young scientists.

During her PhD on cell cycle control, Amon showed that cyclin destruction was not the trigger for anaphase, contrary to the consensus view at the time. This finding proved unpopular, however, as Nasmyth writes:

“Angelika and [her teammate] Uttam were both dismayed that their paper was rejected by Cell and it appeared instead in the EMBO journal. I had a long correspondence about this with Marc Kirschner, who argued forcibly that showing others were wrong did not constitute an important discovery. Needless to say, we did not agree. As Francois Jacob so eloquently put it, the real driving force of science is the recognition that the world cannot be explained by current knowledge, which forces one to imagine new forces. By eliminating cyclin proteolysis as the anaphase trigger, Angelika and Uttam showed that there must be another force out there.”

This, sadly, is an all-too-familiar predicament. An attempt is made to build on previous findings, only to discover that things are not as clear-cut or even as compelling as first claimed. In worst case scenarios, the published data may not even be reproducible. A problem immediately arises when one has amassed a large volume of data showing that the published record is at best questionable, and at worst unreliable.

Everyone knows it’s hard to dislodge or even perturb conclusions once they have been published in a journal, doubly so if the journal in question is perceived as a high-quality one. Everyone can probably also point to various papers that have elicited surprise if not outright astonishment or disbelief upon publication.

When challenged about this, a standard defence on the part of the authors is that the story cleared peer review and satisfied the editor, so clearly it has merit. It is also true that just as it’s possible to come up with an explanation for almost any observation (witness the contortions people go into to protect their favoured hypotheses), it’s possible to come up with an answer to any reviewer criticism. No demonstration of antibody specificity? Well, we attempted this and it turned out to be more challenging than expected. Not able to do something that everyone else considers standard? Well, we did it a hundred times and in our hands it never worked. And so on. If authors are willing to double down on their interpretation and simply refuse to consider alternative interpretations – even if those alternatives require fewer explicative contortions – then they dodge a reckoning.

Editors can similarly play Pilate and choose to overlook a reviewer’s concerns and accept the author’s explanations at face value; if challenged as to how a paper cleared peer review, they can always shrug and say that the scientific record corrects itself in time. That’s small comfort to the people tasked with doing the corrective work, especially when the very existence of the unreproducible work creates a significant barrier to their own work’s acceptance.

It is all too often the case that where the high-profile paper turns out to be wrong, the evidence confirming this gets published in less highly-regarded journals. This has the knock-on effect of disincentivising young scientists from challenging published work, as they know that they are unlikely to publish the challenge at the same level, with inevitable consequences to their status and careers. Such behaviour, encouraged by the disproportionate respect shown to the original paper, runs totally counter to the spirit of scientific enquiry as encapsulated in the Jacob quotation above, and stifles scientific debate – or at least until the first paper’s authors have moved on.

In science, it is better to speak ill of the dead than of the living.

In this respect the operation of the scientific record resembles libel laws, and the contrast in the application of libel laws in the UK and USA is illuminating. People claiming to have been libelled are more likely to win in the UK than in America, because in America, the burden of proof in such cases is on the person who brings the case. In British courts it is the person who has been accused of libel (usually an author or journalist) who carries the burden of proof, and usually loses. In the UK, the rich, the powerful, the privileged, can silence voices who accuse them.

Science publishing is more like the UK libel system than the American one. Should new work challenge a published finding, the onus is put on the accuser to prove their case rather than letting their results stand and thereby inviting the authors of the earlier work to defend their record. A valid though debatable justification for the bias in UK libel law is that the rich & famous are usually more at risk of malicious, salacious and prurient claims and deserve more protection, but this protection is unnecessary and unhelpful in science, where the timing of a work’s release should have no bearing on its reception.

In fact, the new findings should not be treated like libel at all – there is nothing wrong with publishing work that conflicts with the record, so long as it is of high quality. Why should one paper passing through peer review make the subsequent passage of a dissenting paper more difficult?

In individual research fields however, things can be very tribal and personal feuds can supersede good scientific practise. Authors routinely block some experts from reviewing their work on the grounds of a conflict of interest, silencing some of the people best able to critique the work – but with the sometimes valid justification that those experts might not treat the work impartially. Similarly, submitting a paper to the “wrong” editor can waste an enormous amount of time.

Challenges to the consensus may not be discovery as we usually view it, but they will very likely lead to many new ones or at the very least stimulate a reconsideration of dogma. In today’s hypercompetitive research landscape, people have become jealously territorial of research tracts, and questioning past or current conclusions is treated as a hostile incursion, to be met with an aggressive show of force. This is unprofessional, acts to perpetuates consensus views that may be unreliable, and is also not scientific.

We mustn’t be possessive of our ideas and our research areas. Scientists should disagree. They should discuss. They should find what they can agree on, and then explore where they differ, either in their results or in their interpretation. This is the lifeblood of research and enquiry.

In film and literature, the most dramatic discoveries are usually not the finding of something wholly new but rather the realisation that everything you thought you knew was incorrect. As every bench scientist knows, there’s nothing wrong with making a mistake – it helps you understand the thing better, and is the best way to learn.
Acknowledgement:
This posting suggested by and co-authored with Graham Warren.