Done all the visits! They were all great. For each school, I'd rather do a PhD there than not do a PhD there. And for each city, I'd rather live there than not. But I've got to decide...
I think it's UW or CMU. (Which is weird, right. Ask me a month or six months ago and I'd say I want to try something new! I want an adventure! ... I feel a bit like Harrison, but also feel pretty confident that CMU or UW would be the best choice.)
Why not Georgia Tech or Toronto? Ask me in an email, I guess; anything I say to the internet in general could be misinterpreted negatively. They're both awesome schools.
UW CSE is organizationally more traditional: it's a Department of Computer Science and Engineering, within which there are a bunch of great HCI people. But the HCI faculty are mostly a bit younger, and really inspiring. Microsoft Research is right next door, and interning is super common (there and elsewhere). Their graduates win all sorts of awards. Their recent grads (since, say, 2007) are about 55% industry, 23% faculty, 20% postdoc.
Surprising good thing: I had more than one really inspiring conversation about grad school and research while I was there, both with students and faculty. Very idealistic, very pure scientific motives. This is, I think, a good sign in a grad school.
CMU HCII is organizationally more innovative: it's the Charles Xavier academy for HCI researchers. There are about 20 faculty and 40 students, all working on HCI primarily. So instead of learning all about computer science theory and systems and stuff, you learn some CS, some design, some behavioral science, some general HCI research methods. Something like half their graduates have gone straight to faculty jobs (which is unheard-of; faculty jobs are really hard to get). They steamroll through conferences: I think they have the most papers at each of CHI, Ubicomp, and UIST. (though UW is not far behind.) Their grads are 20% industry, 44% faculty, 26% postdoc.
Surprising good thing: it's like a big family! Everyone knows, and supports, everyone (faculty and students). They go to conferences together, have parties together, even buy houses together. And (surprisingly for CMU) I found them all wonderfully charming, with only an endearing touch of CMU weirdness. (we arrive at a party and I'm greeted with "hi! talk to me." "what?" "tell me something!" "uhh... er, what do you want to know?" "you're terrible at this!" Oh, CMU!)
But more important than any of this, I suppose, are the particular faculty and students. I've talked a few times with most likely advisors at each school, and I'd love to work with them both.
To do still: make lists of other profs at each school and their specialties, make lists of students to make sure I remember the atmospheres pretty equally, send some emails, sit quietly and think a lot, and sit quietly without thinking a lot.