Caveat Lector - Peter Kevan on scholarly publishing, artificial intelligence, and the slow erosion of trust in science

You've described a recent run of papers hitting a wall. What happened?

Unprofessionalism seems rampant in the scientific and scholarly publishing industry. That's what it comes down to. We submitted a paper and it turned out the web page was out of date: the journal was no longer handled by the particular publishing houses named, and nobody bothered to inform us nor offer redirection. No link, no notice. We only got anywhere by making enquiries with the journal's own society editorial office rather than the publishing house. And it happened twice on the one paper through two publishing houses. Months of hair-tearing to unravel the snarl.

Once that submission was in the right hands, the reviews went through fairly smoothly. Good reviews, handled to the editor's satisfaction. Then everything went wrong again inside the publishing house, with nobody to communicate with, until I started to make some noise and be a little bit unpleasant. None of it had anything to do with the reviews nor lack of diligence by the editor of the journal; it was all unprofessionalism within the publishing houses.

Is that getting worse, or have you simply had a bad run?

Yes, a bad run, but it's definitely getting worse in my view. A paper we submitted a year ago is still ensnared. We finally reached the editor-in-chief of the journal, who seems to have no authority over how the publishing house is treating him. The house had changed its electronic editorial management system and the article had just been mired in geekism at the publishing house, although it is now in revision following intervention by the editor-in-chief and ensuing rigorous and helpful peer-review.

What galls me is the anonymity. It may be some sort of policy that authors aren't allowed to know who the handling editor (if not the editor-in-chief) is. Prospective authors should know who that person is, so there's a contact, rather than all the automatically generated stuff from robo-geeks who have likely rarely published a scholarly paper in their lives and have no idea what transparent communication is about.

How did we get here? It wasn't always like this.

When a lot of journals were published by societies, they were a service to members and the public. They might have made money or lost money from time to time, but that wasn't the point. Such journals still exist, but so many have been taken over by the commercial publishing houses so that scholarly publishing has become very much a business rather than a service. The decisions (some seem to be ethically questionable) are made now for the profit motive, not for the good of the research.

You've floated a fairly radical remedy.

My feeling is the scholarly community might consider going on strike: not submit anything to any journal (except possibly to publication-free open access journals) for a whole year, until the journals get themselves together and reinstate some ethical principles for scholarly publication, with transparency, honesty and integrity.

We're running a piece by Wim Hordijk in this issue on the misuse of AI in academia. You use these tools yourself. What's your read?

Deep suspicion of anything that comes back to me from AI. Caveat lector: let the reader beware. I can only advise readers of AI to find the original source of the information, find out what was actually said, and look into that.

Hordijk's finding is sharper still. AI inventing citations to his own papers that don't exist.

That's the thing that ought to frighten people. References to work he never wrote: the spirit of what he'd said, attributed to him, appearing nowhere in print. And it isn't only that someone asked a machine and took the answer on trust, but whoever passed it through review didn't do their homework either.

So, what should be done about it?

Scholarly publication needs scrutiny: national, then international. The major societies ought to lead it: the Royal Society in London, the American Association for the Advancement of Science, the Royal Society of Canada, the Committee on Publication Ethics, etc. But I think, in terms of scholarly publication, commercial publishing houses are rather antiquated in what they concern themselves with, still worrying about whether somebody deserves to be a co-author, whether there's some hanky-panky going on behind the scenes. Meanwhile the whole business of AI, publication ethics and the commercialisation of scholarly publishing is an open book, really requiring rigorous and deep scrutiny.

The original ethical standards for scholarly publication were fairly rigorous. There were odd transgressions and fabrications, but by and large those were caught and exposed. It's different now when it's a machine generating information/disinformation/misinformation and people aren't even bothering to check. Something's wrong.

Let me turn you to your own field. What would you point a general reader to?

Pollination. It's a central ecosystem service. It drives the terrestrial ecosystems of the world, save for Antarctica. I first published on it in the early 1970s and became interested in the conservation issues implicit in pollination. That was my first venture into the field, and that paper, alongside one by Dan Janzen, set a great deal in motion: agriculture, policy, conservation, biodiversity, all of it flowed from there. The number of papers coming out on pollinators and conservation is still growing. It's a very active field, and it's had an enormous impact on policy.

And it leads into a story about how research does and doesn't get funded. For example, the problems of neonicotinoids and beekeeping in France.

An ugly one. The French beekeepers claimed the neonicotinoid pesticides were basically killing off their bees, and the agrichemical companies hotly denied it. There wasn't any scientific evidence, the companies said. Well, the reason there wasn't any evidence is that nobody had collected any.

That was the trap. Certainly in Canada, unless one had industry or private sector buy-in to a project, one couldn't get federal research grants. The companies under suspicion didn't want information that might become public, published off the back of a grant, so they were happy to not contribute and say there's no evidence to support research to investigate the idea of concern. Essentially. Nobody had funding to collect the evidence, bar a few not-for-profit and independent organisations.

So who's going to do the research if there's no money for it? There's no money because industry won't give any, they don't want the answers to the questions at all, and the government says it can't give any because there's no interest from industry. It was the proverbial “Catch-22”. It was pretty corrupt, when you get right down to it. "Unethical" would be the softer word.