This is the irony of good QA.

1) Company starts to notice a lot of quality problems.

2) Company institutes a good QA program.

3) Company notices less quality problems.

4) Company cuts QA program because there seems to be less need for it.

Wash. Rinse. Repeat.

(The same can be said for safety when dealing with low probability events).

Yes.

It is raining.

I use an Umbrella.

Still raining, but I am no longer getting wet.

I must not need the umbrella.

Close the Umbrella

Get wet again.

In that case, the causality is easy to determine. But that’s not the case with more complex systems, so it’s easier to rationalize a bias (I don’t need this expensive QA)

This doesn't explain Microsoft though, as that's probably a special case. They simply eliminated the Windows QA department, and instead let users do the QA. Why not? It's not like their users are going to switch OSes.

I don’t know about that specifically, but I’ve seen that as a common occurrence as a cost cutting measure in low risk situations. In that case, the risk is low because 1) it’s not safety critical and 2) the switching moat is large enough to ensure they don’t lose customers