That struck me too. The measures of usability were rather mechanically applied. If a site could check off 32 points on a usability checklist, it was considered usable, regardless of the mishmash of fonts and colors. So the example was pretty extreme and, despite my rather dismal design skills, I don't think I could ever turn out an abomination like the test case "ugly" site.
No doubt about that. I expected some of them because I knew there were problems there, but wanted to watch somone try to accomplish a couple of tasks for a reality check to make sure I was solving the right problems. The ones I saw as problematic were generally a problem (though not always).
The surprise, though, was some of the features that I thought were pretty straightforward. I was NOT surprised by the fact that the tester failed in places I didn't expect. I was surprised by the fact that the second she slowed down, I knew exactly what the problem was. It was so obvious where her gaze and cursor were going and they weren't where I expected.
So of course, an utterly unusable site is a failure, regardless of beauty. But the study did make me wonder about paying more attention to aesthetics. I'm such a text-oriented person by nature, that I tend to notice typography, but other aspects of aesthetics much less so when I visit a site (I'm here on WebmasterWorld aren't I?).
It doesn't surprise me that users find a beautiful site more authoritative. But it's interesting that they rate it more usuable, and I think Tedster's take on it (they think it must be their fault, not the site's fault) hits the nail on the head.