The easy sciences stay scientific much longer than the difficult sciences. E.g. puzzlingly, IQ research continues.
Hypothesis: the easy sciences filter out the frauds. When a termite tries to burrow into a grant-writing or journal-editing position, they get thrown out for obvious incompetence. Even if the State appoints an apparatchik, they have to withdraw because the contrast is too embarrassing. In the difficult sciences, a termite can trivially Sokal Hoax their way into whatever position they desire, and thereby shut out anyone attempting to honestly approach the subject.
The easy sciences only get politicized either by State force majeure or when they luck into a genius termite who can actually do the work yet is still more interested in pushing the religion.
Conclusion: in most cases, gatekeepers are not seized by force majeure. It occurs due to a failure of internal policing. Too much mercy, generally by a very wide margin.
As always, the answer is Exit. Defend the gatekeepers if you want, but sooner or later the gates are going to be seized by hostile parasites. The question pertains to what to do once this inevitably occurs. The answer: leave the gates. Let them have it. Every other solution fails both theoretically and empirically on quis custodiet grounds.
As always, the more Exit you have, the less you need. If folk can leave your gated community whenever they feel the gatekeepers are hostile, then you don't end up with a community riddled with rotten pockets of hostility. The hostile wannabe-parasites tend to be somewhat lame and crippled, so the easier it is for them to Exit, the less you have to eject/execute them.
Or: there is no alternative to the NSF, which is how you know the NSF is corrupt. The termites know that if the virtuous could leave, they would.