Whenever something like this comes up, I think that a closely-related topic is replication of scientific studies. One of the best articles I've read on this was

Whenever something like this comes up, I think that a closely-related topic is replication of scientific studies. One of the best articles I've read on this was by Stuart Buck here [1], which I think offers a great starting point. I also think it's really relevant to this particular discussion; the added emphasis on attempted replication probably wouldn't prevent paper mills from working period, but they could drastically increase the cost of good forgeries and create an environment of accountability that could potentially make their use by researchers much more risky, which might drop their use even further. Of course, publishers could help as well; maybe only the most innovative papers that have been replicated by unaffiliated third parties that have been provided with the pre-print or low-impact initial publication get re-published in the higher-impact journals, and those that can't get watermarked with with "unable to be reproduced x of y times." [1] https://worksinprogress.co/issue/escaping-sciences-paradox/ https://worksinprogress.co/issue/escaping-sciences-paradox/