Science is self-correcting, but we don't act like it
Mistakes can never be completely eliminated
The Wall Street Journal is out with a big story today on the sleuths that are calling attention to what we call in the business “problem papers.” I spent a lot of time talking to the great reporter (Nidhi Subbaraman) who did a ton of work on this. I’m the only one from a journal or institution in the story, and I wish that wasn’t true. We need to engage rather than running in the other direction.
The story pays tribute to a number of sleuths, including the folks at Data Colada, who have brought attention to many problem papers, including most prominently those about dishonesty involving a prominent Harvard research who is now suing. Adam Marcus and Ivan Oransky - who have brought us Retraction Watch - have recently written about how this is not unusual. Subbaraman also highlights Elisabeth Bik who is well known for finding problems with images in papers.
As I said in the story, no editor loves it when we hear from these sleuths. It’s a ton of work to dig through the problems in these papers and negotiate with the various parties, which include the journal, the authors, the accusers, and the institutions (sometimes more than one). I’ve pointed out recently that the journals have simpler goals than the institutions: we just want to correct the record and are not interested in the personnel matters that the institution might have to deal with if there is deliberate misconduct. Unfortunately, the fear (and frequent reality) of litigation causes the institutions to clam up and that drags the whole thing out.
What’s needed is a generative system where instead of litigating and finger-pointing, everyone works together to correct the record, as I wrote about here. When we posted retractions in the Marc Tessier-Lavigne matter, I said in the LA Times that this was how science should work and not something we should be embarrassed about. We’re a long way from achieving that.
It’s impossible to imagine a world where scientists never make mistakes, both inadvertent and deliberate. It’s impossible to imagine a world where journals and reviewers catch every problem. Going back and correcting and retracting papers will always be required.
If we can get to a world where the shame and blame are replaced by collaboration and compassion, we would have a system that works.
Dear Professor Thorp,
If I may, would you consider:
Science is self-correcting, but scientists aren't
as a title for your post (cf. Marc, Poo et al. haven't confessed, but blamed the times that changed ;)
Admittedly, the proposed title needs to be qualified with respect to both science and scientists.
There are many scientists who voluntarily admit their mistakes and correct them. Professor F. William Lawvere (https://www.acsu.buffalo.edu/~wlawvere/) wrote to me many times correcting a mistake he made in an earlier email; he exemplified 'self-correcting' not only in private communications, but also in public posts, publications, and lectures (but for my pathological anxiety about preparing for my upcoming lecture on Education, I'd have listed all those, which I'll do after my lecture). For now, in this context, I'd like to note that Grothendieck (https://agrothendieck.github.io/), who insisted on publishing all the mistakes he made in the course of writing Pursuing Stacks (one of his many magna opera; https://www.acsu.buffalo.edu/~wlawvere/FarewellAurelio.htm), is a model scientist in many respects. In closing, just to hint at his brilliance, Grothendienck (re)defined a part of a whole as both itself and its relationship to the whole (e.g., a subset of a set is not a set, but a one-to-one function), which is readily understood by laypeople, although mathematicians in need of updating their understanding of 'subset/subobject/part' seem to find it purty perplexing to the point of acting unawares ;)
Thanking you,
Yours truly,
posina