I actually think he's right in dismissing those studies as ludicrous. You can't honestly take one or two (that I can recall) "studies" with only a dozen participants and then brand your product as "usability tested." That's just paying lip service. When you're changing a complex, deeply-ingrained interface for hundreds of thousands of people, doing two studies on a dozen people--and AFTER the fact!--and then calling it a day does indeed seem ludicrous.

Though at least Canonical can say they've done SOMETHING, which (at least I think) is more than Gnome has done. Gnome Shell gives itself a handful of breathless superlatives but I don't think they've ever once actually done a study of how real people use software. (Correct me if I'm wrong.)

In the end, though, such studies are very expensive and time-consuming, so free projects would have a tough time pulling it off. Though that doesn't discount the silliness of Canonical's attempts.