Yep, it's really hard to keep things blind with a reviewer who's determined to unmask the author! Even the title of a paper in a conference list, or on a list of "works in progress" can be turned up by google pretty quickly. That may be addressed by changing the title for a submitted version. But  if the paper is online, googling random sentences is indeed another way in which it can be unmasked. (Lest it seem that I spend my time unmasking authors whose papers I'm refereeing---this is the same process one goes through to find the source of unreferenced quotations, etc).

Of course, blind review is still in place if referees *don't* google search for titles or random references (or do searches in philpapers). It seems to me that there has to be a certain level of trust in the good faith of referees *not* to unmask authors (at least before the report is submitted---afterwards is a bit of a grey area).

Personally, I have been sent papers to referee where I'd already read in draft (and so knew the author) a few times. Each time, I've contacted the editor to let them know that this is the case, and usually they've asked me to go ahead with the review anyway. So that wasn't a fully blind process, even at journals that go through the right process. I don't really know what to think about that (I've treated it as the editor's responsibility to make such decisions).

I do know people who, in the light of this sort of thing, don't put drafts online. But that isn't cost-free: the whole point of the enterprise, ultimately, is about sharing research, and I find it incredibly valuable when people put stuff online. And for myself, I've got valuable feedback from online drafts in the past. So for me, the benefits of making drafts available outweigh the costs. But I'd be interested to know what others think.