many hours over a few days going back through Twitter with various search terms, back to include tweets during the months when pre-releases were being tested. I now know that Twitter offers advanced search syntax that would have made the task less arduous, but I don't regret the work because ploughing through so much, in a concentrated way, left me with a strong impression that public comments on the looks of the pre-release were mostly favourable (point 10 above). I can't treat it as qualitative analysis, but I did get that impression.
That exercise, and others, with Twitter also gave me the impression that the majority of post-release tweets about the looks of Yosemite some of which are posted by professionals are not favourable.
A fraction of what I bookmarked
http://tinyurl.com/1010uglystick might have been posted to Twitter by bots of some sort but I'm not beating myself up about that. There's plenty to be ridiculed/disliked/hated and if a bot can perform the ridicule, then I sort of admire whoever wrote the bot. (Note that I was not careless. Some of the Twitter accounts did appear odd and if I suspected that a bot was at work, I refrained from bookmarking. I blocked and reported a few accounts as a result of the exercises.)