Which is a bit silly, isn't it? Grad students are poor and overworked, it seems easy to find one to trick/bribe into signing off your code, if you wanted to do something malicious.
If this is foolproof, then no-one should be talking about the replication crisis.
People don't do bad things _expecting_ to be caught, if they haven't already convinced themselves they're not doing anything bad at all. And I suspect it's surprisingly easy to convince people that they won't get caught.
Replication is really a different problem. It's possible for you to do nothing wrong, run hundreds of trials, get a great result and publish it. But it was due to noise/error/unknown factors, and can't be replicated. The crisis is also that replication receives no academic recognition.
When people fabricate results they know it's an offence, the problem with these guys is they don't even acknowledge/understand the ethical rule they are breaking.
In late January I submitted a patch with no prior contributions, and it was pushed to drm-misc-next within an hour. It's now filtered it's way through drm-next and will likely land in 5.13.
https://github.com/torvalds/linux/blob/master/Documentation/...
Right? It's true that all systems can be gamed and you could no doubt fool the right maintainer to take a patch from a fraudulent source. But the point is that it's not as simple as this grad student just resubmitting work under a different name.
Maybe?
My point with the above comment was more to point out that there is no special '"presumptive good faith" pass' that comes along with a .edu e-mail address, not that it's possible to subvert the system (that's already well known).
Everyone, including some random dude with a Hackers (1995) reference for an e-mail address (myself) gets that "presumptive good faith" pass.
One of the reason this worked is likely that submissions from large US research universities get a "presumptive good faith" pass. A small company in the PRC, for an example, might see more intensive review. But given the history of open source, we trust graduate students maybe more than we should.
[1] Originally legal/copyright driven and not a security feature, though it has value in both domains.