> A single frame that's 5x too long or 4 frames that are 2x too long have the same impact on the average delta of frame staleness on the total sum.

It's not clear to me: are you saying those two scenarios should be quantified as equally bad? Because it seems pretty obvious to me that the 5x outlier is qualitatively much worse.

The fact that the 5x outlier may be qualitatively worse is something that you pin down using percentiles, which is what everyone does, either 1% or 0.1% low. The argument about time-weighted frame staleness is not mathematically valid.

There’s nothing inherently mathematically special about 1% or 0.1%, you can also use 0.0042% or 0.00069% which highlights this kind of extreme outliers more.

Benchmarking sites use 1% or 0.1% because they better highlight the commodity PC hardware rather than the games bugs and architecture.

That's not true. You can't use 0.0042% because then you don't have enough frames to be statistically significant without running for a very long time. It's also true that a frame pacing issue that is so rare that it isn't even impacting the 0.1% frame time isn't going to be noticeable, really.

These benchmarks are controlled for game bugs and architecture by simply averaging over ~40-50 games, on repeatable loops.