Saturday, October 7, 2017

No, science is not fundamentally broken just because one person had problems with their supervisor and prefers preprint servers

While on holidays I learned about an interesting case in food science, where it is suspected (at least according to some statisticians who have dived into the literature) that a very prominent researcher may have P-hacked, self-plagiarised, potentially reused the same dataset for several publications while making it appear as if they represented independent studies, and used self-citations to support statements that they don't. I find stories like those at the same time interesting, inspiring and very, very frustrating.

Interesting because, well, human nature and all that, and I am continually puzzled whether the people in question really think it won't come out in the end. Inspiring because this is how science self-corrects; not necessarily by individual scientists changing their minds (although that would be the ideal), but by open debate between scientists and careful re-examination of data. And frustrating because of exaggeration, over-generalisation, and naiveté about the solutions or alternatives to the problems that are identified. This post will mostly be about over-generalisation as found in the writings of one Jordan Anaya on Medium.

He evaluates and criticises the relevant food scientist's working practices, as do many others. But in doing so, he writes the following:
But I am interested in how academia selects for bad science, is free from any outside regulations that might prevent a crisis like the housing bubble, and how its power structure allows senior members to behave like dictators.
...
Science in academia is not about performing science, it is about your brand.
...
We are in the midst of a reproduciblity crisis in science,
...
all the problems science is currently facing
...
As it stands now, the wrong papers get published, the wrong researchers get funded. There is no incentive to share data or perform careful science. The only thing that matters is your brand, and your ability to leverage that brand into publications and grants, which circle back to feed the brand. If that means performing sloppy research, exaggerating results, and then refusing to acknowledge any errors, so be it.
...
Most of the literature is wrong, this is just a reminder that we need to be vigilant. It is also your daily reminder that peer review is useless and everyone should instead be preprinting their work.
I find this nothing short of astounding. Here is a bioinformatician generalising from his own bad experience in what appears to be one research group and one case in food science across all of academia and across all of scientific research. He does not write, "food science seems to have a problem", no, it is all of science, based on n=2.

When he writes that academia selects for bad science, can he really say with confidence that that is the case for, say, Australian entomologists? How would he know?

When he writes that it has no outside regulations, does he really mean to claim that it is not ultimately voters who decide through elected governments what kind of research will get funded? Is there so much money in cancer research in spite of or because of the wishes of the public? And this is before mentioning industry collaborations and industry-funded grants. (I will grant him that science isn't a democracy. The question is whether it could work as one, but that is beyond the scope of this post.)

When he writes that science as currently practiced in universities is not about science but about branding, can he know that this is the case for inorganic chemistry in southern Germany?

Do astronomy and phylogenetics really face a reproducibility crisis? I at least am not aware of that, and indeed I would be confident that if somebody were to repeat pretty much any phylogenetic study published in a serious journal with different molecular markers they would be able to reproduce all major results.

Is surface physics really facing "all the problems" that food science does? Again, how could Anaya even pretend to know?

How does he know that the wrong papers get published and the wrong researchers get funded in, say, plant ecology or archaeology? He claims that there is no incentive to share data - has he never heard of GBIF, Genbank, TreeBase or Dryad, or of all the journals that will not even accept your paper if you haven't deposited your data in a publicly available repository?

"Most of the literature is wrong." Seriously? Meaning I could pick a random article from a well-respected taxonomic, systematics or evolutionary biology journal and there would be a more than 50% chance that it is "wrong"? (And what qualifies as "wrong"? Does it mean this group of plants dispersed to Australia 16 million years ago instead of the currently accepted 15, or does it mean evolution is a lie meant to destroy Christianity?) And the alternative is to do away with all quality control whatsoever? But I am getting ahead of myself, this post is meant to be about over-generalisation, not solutions.

There is no doubt that the way science is being practiced leaves room for improvement, in particular in the areas of research funding, incentive structures, publishing, and recruitment. Okay, the same is probably true for any collaborative enterprise that humans have ever undertaken or will ever undertake, but tu quoque arguments don't change the fact that a lot can be criticised. I would have quite a few ideas for improvement myself.

Still, I hope one thing is clear: the fact that there are problems and some people acting in bad faith does not mean that an entire enterprise is broken. When we find that there are incentives for teachers to inflate grades we do not conclude that all of education is broken; when we find a certain percentage of police officers are racists we do not conclude that all criminals should go unpunished. And you may have heard that when we pour out the bathwater we usually take care not to pour the baby out with it.

Of course it is a question to be debated whether the entire system is broken. It could be. But from what I can see in plant systematics, ecology and related fields, this is so much hyperbole. In fact I am struggling to find a description of the act of dismissing all of science and the careful work of thousands of researchers with a mere "most of the literature is wrong" that is reasonably polite and does not involve phrases like "breath-taking arrogance".

It is also, of course, ridiculously irresponsible. Think of anti-vaccinationists, creationists, climate change denialists, alternative history cranks, expanding earthers, and any other set of conspiracy theorists. They already claim that science is all broken and the literature can't be trusted. Now they can say that a scientist frustrated with the review process in his field confirms it. Believe what you want, because the scientists can't be trusted!

Ye gods. Did the plain truth - science publishing and funding have serious issues that we need to tackle, and there are a few frauds just like in every other profession, but all in all we can trust most of our colleagues to be well-meaning and most of the literature to be useful - not sound sensationalist enough?

No comments:

Post a Comment