Much better to make reproducibility a valuable research contribution in and of its own and let "the market" take care of it. Create better incentives for all authors. For example, make it possible to submit the code for analysis alongside an article and highlight articles who do (similar to the open-access movement), create journals dedicated to reproducibility, measure successful reproduction attempts alongside citations of each article, etc.
"the market" is the same force that has led to the perverse incentives of constantly pushing new "discoveries" above validations. The small variations you have prescribed have been tried (to various degrees) and have not had an effect on aggregate behavior.
> Why make it a PhD requirement? You'd end up with half-assed attempts to reproduce easy results and little mafia rings of my-reproduction-for-your-reproduction.
The why is obvious. The purpose of a PHD is for the furthering of knowledge, so it's a perpetual source of motivation.
> You'd end up with half-assed attempts to reproduce easy results and little mafia rings of my-reproduction-for-your-reproduction.
That's how it might end up (largely based on the rigor of the PHD program), but it's better than what we have, which is no good incentive to reproduce results.
I can think of plenty of reasons why it won't happen (notably the downward pressure on the candidate pipeline, which for a variety of reasons isn't an outcome research institutions don't want), but I'm wondering if the best place to apply the pressure would be the accreditation institutions and the Department of Education.
So for each publication, you'd need to provide two serious attempts of reproducing other published results. You would make the reproductions yourself, but also anyone else could do them for you (here's where PhD students come in), but once used for a publication, they'd be "spent tokens". (No, I'm not going to even mention NFT. ..wait a minute, crap..)
From here it would follow that eventually all results would be reproduced (or failed to do so), until we eventually run out of results to reproduce and would need to shift the burden from two reproductions down to one—probably go from 2 to 1 in some stochastic manner. (Maybe also similar number of serious peer reviews of such reproductions would be required to keep the system running.)
Maybe we could make a new currency out of this. Actually, it sort of automatically does that by itself. I foresee no issues in this whatsover.
Speaking for my own PhD (now around 25 years ago): it's hard enough getting one's own damned work to reproduce, never mind the work of a complete stranger.
Reproduction in science is a though problem -- not only it requires specialist lab equipment, methods or procedures, but sometimes "a trick" is needed. Inconveniently, it may not be mentioned in the publication for whatever reason. Therefore, giving an unexperienced PhD a task of reproducing somebody's work may end up to be a futile effort. But the idea is neat in general, I think.
I would even go so far as to say that peer review is worthless, it may well be time to replace it with peer reproduction as a prerequisite for having an article published.
This also ensures the most famous results are the most replicated. Hugely valuable to society since the most famous results also drive the most thinking / decisions / policies.
The challenging researchers are simply implying that the original study did not account for the positive effects of red meat (vaguely 'vitamins and minerals).