ISMIR 2008 seems extremely well organized. I've been only watching from the side line this year (for the first time since 2002 I haven't submitted a paper myself) but the bits and pieces I've seen seem great.
I like how reviews are double blind this year.
I like how authors get a chance to respond to reviews. And reviewers get a chance to reconsider their ratings after seeing the response of the authors, and what their fellow reviewers wrote.
However, and this strikes me as fascinating: I get to see the name of my fellow reviewers! (To be more accurate: I only get the see reviews and names of reviewers who reviewed the same content I did.) First I thought it was a bug in the system (I even felt the urge to instantly report it to the program chairs). But it seems like the whole system is designed around exposing the real names of the reviewers to the fellow reviewers: I was just sent an email with the email addresses of all my fellow reviewers.
I think I've never experiences such openness in any of the review processes I've been involved in. It's fascinating, but it makes me wonder if that might lead to reviewers being more reluctant to write critical remarks in the future? Especially in such a small community as the ISMIR community, one of the fellow reviewers might be a colleague of one of the authors etc. While I think it's a good idea to publish reviews and a response to those (I did so for one of my publications last year here), I don't think it's necessarily a good idea to expose the reviewers without their consent.