Meta optimizing employees have issues hiding at smaller companies where one person has a vision of what needs to happen and quickly identify someone not pulling in one direction.
Once you get big enough, upper management has tough time figure out who is meta optimizing over work-optimizing. Not to mention there might be multiple meta optimizing employees.
I've seen high performance organizations at 600-800, thanks to execs that spent quite a bit of time talking across levels: When at that size, some ICs get CEO 1:1s, you have some chances of quality control. After all some execs had been coders. The problem is that none of them had ever been middle managers, which meant they had no idea of how to tell a good one from a bad one.
TAs the company kept growing, and hired middle managers from bigger tech, they that Jira was the way to go, as it allowed for nice reports aggregating "insights"across the organization. In under a year, point-centered management arrived, and with it an exodus of top talent, all of which had massive amounts of equity anyway. Execs then wondered what happened, and why ability to ship features kept declining. I think they still don't know.
I'd also be curious as to how it relates to span of control (~4-15 direct reports) and therefore levels of management for a given org size, as information hiding about actual work performed seems tied to managerial masking.
Once you get big enough, upper management has tough time figure out who is meta optimizing over work-optimizing. Not to mention there might be multiple meta optimizing employees.
IME, it's around 100-150 employees which lines up with Dunbar's Number. https://en.wikipedia.org/wiki/Dunbar%27s_number