In terms of whether it’s poor/reliable design, I think the failure mode and dependency chain is much more important than capabilities. The glovebox example has two key features your windows example does not:
1. If the glovebox fails for any reason, there’s no workaround and it goes to closed. (I don’t know whether it’s true there’s no workaround, but that’s what we’ve got suggested here). This would be akin to the window failing, and reverting to full-open position with no way to physically pull it, or convince it to stay in current position.
2. If the infotainment system fails, the glovebox is taken out with it. Because of 1, the glovebox is taken out 100% — not just the convenient infotainment style of usage. This would be akin to, well, the same. The death of the infotainment system is the death of your windows.
3. There could also be the reverse dependency, not in play here (afaik) — the death of the glovebox mechanic may equate to the death of your infotainment system.
That is, the question isn’t whether the individual component is reliable in and of itself (which is less a statement about Tesla and more of their suppliers), but whether they’ve tied things together such that one failure creates a cascading effect — and done so unnecessarily. And in such scenarios, whether backup strategies exist.. (your tire shouldn’t blow, but if it does… there’s an extra tire in your trunk and shitty-but-enough tools to get by). And as I understand it, Tesla has produced a masterpiece of unnecessarily coupled components, particularly around the UX.
So it’s not surprise to me that they see higher UX failure rates — they seem to design the car intentionally to have no degraded experience allowed; either everything works, or nothing does