> My experience is that managers don't like it when people rock the boat
That's what's missing from Graydon's analysis: risk-aversion is also a strong incentive in many corporations. I would argue it's the rule for middle managers, with growth being the exception.
Also missing is telemetry and coordination, where companies use FOSS to find out what other companies are doing, or to coordinate policy, esp. when they fund a leading contributor from whom other companies need buy-in.
Put another way, a FOSS contributor is not an individual, they are a company representative, and their opinion has the weight proportional to the companies' influence.
The contributor's influence also depends on the composition of the other contributors. Alternate influences become impossible when a company dominates the contributors; hence e.g., the CNCF tracks metrics for ensuring that it takes a plurality to dominate.
But really this posting isn't about FOSS contribution at all; it's about the under-valuation of avoided future costs. But that's a much harder problem, because you get all sorts of illusory accounting when people project potential costs they're avoiding.
E.g., I um heard that at (big firm with ~1000 developers), the QA team was successful arguing for additional funding because they were finding more bugs. So the kernel team started tracking edits as fixing potential bugs, to restore the balance of funding. They hated the game, but had to play it.