Seems like they were just citing anecdata anyway, so calling out a statistical error doesn't seem worthwhile.

I believe their main point is that humans are much more tolerant of latencies than is commonly assumed. I remember reading about Carmack's early work with VR and sensor delays and screen refresh delays on the order of 100ms and as high as 400ms and it still being enjoyable[1]. At 60fps you've got about 16ms per frame to work with, and I'm sure the human mind doesn't need input -> change in image within a single frame (though I'm sure pro gamers would notice).

I even see people using and enjoying shadowpc + vr headsets, where delay between head turn and change in image can be nauseating.

[1] https://www.pcgamesn.com/virtual-reality-john-carmacks-battl...

Have you used it recently? People are definitely blowing the whole input lag thing out of proportion. Sure it might not hold up if you're a competitive gamer, but for the leisure gamer it's definitely not noticeable assuming you have a decent internet connection.

I have a gigabit connection with ~10ms latency and Stadia feels awful even for single-player games.

Yeah but when did you try it? They've made tons of improvements since it was first launched.

Of course it is to a degree (I don't remember anyone I personally know complaining about the latency, but there is very likely a self-selection that those that would didn't try or didn't talk about it). But I'm not sure why that matters when addressing the claim that the product is a priori useless?