ISCA and Importance of Conferences

I'm leaving momentarily for the International Symposium on Computer Architecture, in Boston this year. Although Mark Oskin holds the honor of authoring the first ISCA quantum computing paper, I believe, in 2003, this will be the first time there is a full session on the topic. Three papers, one from Berkeley, one from a mixed group including Davis, Santa Barbara, and MIT, and ours (Keio, HP Labs, and NII), will be presented on the last day of the conference (Wednesday). A good performance by the presenters will go a long way to convincing the architecture community that there is important and interesting work to be done, and that direct involvement of architecture folks will accelerate the arrival of viable quantum computers.

For you physicists who are still learning about the CS conference circuit, the top CS conferences review and publish full papers only, and are very competitive (I think ISCA's acceptance rate this year was 14%). Journals are important, too, but in many ways less so. Of the top ten most-cited venues in 2003, eight are conferences and two are journals, according to CiteSeer. I think CiteSeer's publication delay list is very suspect, but it gives you an idea -- conferences are much shorter. Many ACM and IEEE journals and transactions average more than a year to return of first reviews, let alone final publication. Recognizing that this is a problem, many of the newer journals, such as JETC, JILP (well, at seven years, JILP might not count as "new"), and TACO are working hard to keep turnaround time for reviews down. But the dialog on open access is, I think, further advanced in physics than in CS, which is not what should have happened -- IMHO, CS should have been the leader on this topic.

Anyway, I'll try to blog from ISCA. The program this year looks exciting, though interestingly, there are no storage papers this year. Perhaps FAST and the new Transactions on Storage are getting the best storage papers these days?