This fucking reeks of Pressure to Publish in academia.

Except that publishing stupid papers does not harm millions of users, but yes.

It harms thousands of researchers who read them and base their further research upon.

The average paper only gets one citation.

If the average type of paper deviates over time, it doesn't matter whether the average number of citations is low - the type (direction) of research that gets cited shifts as well, and ultimately the direction of all research done shifts.

I don't understand how this scheme (in companies or academia) manages to emerge and sustain everywhere.. surely there's an answer but I don't know it.

I think the long and short of it is that systematically incentivizing anything is hard.

When organizations are small, you have the resources to evaluate performance on a case by case basis because everyone works with everyone else to some degree. When organizations grow and you have layers of bureaucracy, inevitably some asshole middle manager gives someone a bad review for a bullshit reason (jealousy, racism, dislike, etc).

That's when HR swoops in and starts making processes and standards for everything to shield the company from lawsuits. Crucially, these processes and standards don't necessarily prevent the problem, but just let the company's lawyers argue that they tried to prevent the problem.

Now that you have processes and standards for everything, employees start gaming the system. The goal is no longer "do the thing that makes the company succeed". The company is so big that relatively few individuals have the power to swing the company's fortunes one way or the other. Instead, the goal is now "maximize my career growth".

Then you end up in the nonsense FAANGM situation where individual little teams are putting out crap and making decisions that make the company look dumb. Things that would be easy to ignore if a significant fraction of teams weren't doing it. All because they're doing things that maximize their "impact".

This is hard to correct because you don't want to be too harsh with your teams and hurt morale, and it's not like it's every team. So you try to be flexible, be tolerant of missteps as long as people show improvement, because if you just fired everybody who made boneheaded decisions in the name of "impact", the hit to moral would cause terrible attrition problems. And worse, some of them would sue the company, which was the problem you were trying to avoid in the first place.

s/company/university/g as needed.

It's one of those classic situations where every individual step in the process does at least kind of make sense, but the end result is bonkers.

That's the problem with "impact", they're not asking for "positive impact", and they're not getting it.