Right, that’s why I said the key metric is retractions per 1000 papers, or any metric which measures the proportion of legit vs retracted papers.
Of course, there is the issue of whether a paper is truly contributory, or just ‘salami slice science’, where you examine some very narrow and trivial issue. The usual measure that attempts to quantify this, is “citation number”, journal impact where the paper appears and such comparatives. Unfortunately, as with all attempts at measurement, they can be gamed. Same as it ever was. The old publish or perish imperative is a perfect example of a legit measurement attempt of productivity, and the result of gaming that measure, which in turn resulted in salami slice science. Of course, in recent times, many have abandoned such “innocent” tricks and simply plowed into full on fraud.
Until there are real consequences to such fraud, it will continue to proliferate virtuallu unchecked. The number of offenders so outpaces the “checkers” and investigators, that it’s a losing battle.
The worst is that the institutions are frequently complicit in this fraud. They have a “star researcher” who attracts grant money, endowments, attracts students, raises prestige and so forth, well, if the star transpires to be a fraud, the institution administrators are not eager to discipline, fire, present consequences, instead, they cover up. Everyone is in on it, from the admins to the other authors and reasearchers at the lab all of whom ride along on the coattails. Reviewers scratch each other’s backs. The loop closes. The big loser is science, and trust in science, and downstream from that (in medicine) drugs that don’t work or injure patients. And sadly, again, nowhere is immune, including top institutions like Stanford, Harvard and so on. By the time you get to some obscure institution in China, India, Russia etc., you are really looking at lottery type odds - like in economics, bad money pushes out good money. When the fraud is up and down the hierarchy, any honest researcher is just pushed out.
It’s hard not to become cynical, but the answer is always the same: follow the science, replicate and verify. And good ones still exist (ITP!).
I think it’s inevitable that something has to be done. Proven deliberate fraud should result in a permanent ban for those responsible. Fraud, perhaps should even have legal consequences, same as if you lie on a loan application or receive money (research grant), but deliver fraud (don’t fulfill the terms) - pay restitution and go to prison. Negative, null results and studies that simply attempt to replicate results should all be elevated in publishing and valued on a CV.