Some of his results are biased because he pretends that 30-35 fps at 2560x1600 4AA is playable in a first person shooter or a racing game.
Minimums near 30-35 FPS are acceptable for most. I think it depends on the reason and the frequency. The averages are generally higher, sometimes much higher.
There is always commentary following the graphs that detail what is happening at low spikes which
allow readers to make their own conclusions for their own intended usage
If, for example, a down spike is due to heavy action and someone expects to be in heavy action a lot, then they can figure on lowering video settings to be acceptable. On the other hand, if a down spike is due to some quirkyness following a level load, or something like that, then they note that and you can expect it to be the same at any setting.
They're about the only place I've seen that shows instantaneous FPS graphs so you can draw your own analysis. In their Skyrim charts, for example, you can look at GTX580 SLI scaling and see it's poor, but you also can note it's higher FPS than anything else in the article... then you can see areas in the instantaneous FPS charts where scaling is very near zero in some areas and +80% in others and conclude that they're definitely bumping against CPU limitations in some areas. So, if you spend the time, you can actually SEE the CPU limitations, which makes sense as the TES series have always been fairly CPU intensive with such large worlds.
Then you can look at these areas and focus in on the same area on OTHER charts to give you even more useful information, or read the comments on other cards to find out what parts of the games those CPU limitations are seen.
No other site gives the reader that kind of power. Min / Max / Average charts are near useless unless the minimums are put into context like [H] does.
You don't like his criteria for playable... that's fine, he gives you enough information that you can then take that information and shape it to your specifications for what playable is in a way that no other review site does. I find it difficult to believe that people would outright dislike the amount of information [H] gives to readers.
Have you ever FRAPSed your own play and noted areas where you
notice performance degradation to learn how to define what playable means to you? When I did it, (years ago when I was playing shooter games more) I was surprised that there were some times when FPS dipped to 30-35 FPS I didn't notice, and other times when it was quite bothersome.
Most of the time I didn't even notice. It was a surprise to me, because before I tested myself, I never would have thought a dip to 32 FPS would be tolerable based on the media and forum information I had been fed.
It was time consuming, but I learned a lot about my own perception and hardware requirements. This was way before [H] started doing things like they do now, and I really appreciated it when they started doing full instantaneous FPS graphs like I had been looking at during my own play at various video settings. It really gives a lot of info at a quick glimpse, and the comments following usually give further insight.