The last SOSP of the second millenium--true no matter how one counts--seemed to be a great success in all respects. Attendance was excellent: 389 total, of which 175 were students, of which 60 had scholarships--out of 90 who applied. Moral to students and their advisors: apply!
The core of the conference remained the strong refereed paper presentations and the rich hallway interaction with a "who's who" in operating systems research. However, three unusual special events enhanced SOSP this year, two of which are summarized in the following pages. Mark Weiser, who recently suffered an untimely death, was honored in a moving special session. His friends and colleagues shared a wide range of reminiscences of this visionary researcher and very special person. In another session, Jerry Saltzer and Butler Lampson contributed two insightful, humorous, and reflective invited talks. In a late-breaking talk that is not further summarized herein, Ed Felten gave an entertaining after-dinner talk on the Microsoft-Dept. of Justice lawsuit. He focused on his role as an expert witness--what the experience was like--and included an "Antitrust 001" tutorial. He used an analogy accessible even to CS researchers: a company with a monopoly on flour is legally constrained in its actions in the sugar market.
This particular SOSP seemed to offer especially good schmoozing, at least to me, and I have heard from a number of people that they found it an especially good conference. In a novel perk, students who served as volunteers were offered "lunch with the big-shot of their choice." The receptions, supported by generous industrial sponsors, were frequent, convenient, and liquid. They were held right amongst the posters during two nights, so for once the 17 posters (summarized below) apparently received their due audience.
SOSP has always aimed for seclusion, to encourage interaction among attendees. The carefully-groomed Kiawah Island venue certainly was secluded--the guards and checkpoints on the road reminded me of my youth in Haiti. But the beach was close, wide, and breezy, occasionally serving as a place to clear the head, play Ultimate, or watch the meteor showers at 1:00 a.m.
The bulk of this report, however, focuses on the papers, the most important part of the conference. John Wilkes, program chair, deserves a huge amount of credit for organizing a successful reviewing process, doing more than his share of reviewing 60 or so papers, and helping to line up the special events on the program. 89 papers were submitted of which we provisionally accepted 19, all of which improved through the now-routine editorial review process called "shepherding." The OS field appears to be unique in the quantity and quality of the reviews generated for its top conferences, in this case SOSP and OSDI. In this SOSP, second-round papers typically had 14 reviews, often lengthy. The program committee and external referees generated over 800 reviews, totalling over 2MB of text and 280,000 words. Based on this mass of data, the program committee selected four papers to forward to TOCS, recommending them for expedited editorial review. Their brief titles are "Porcupine," "Cellular Disco," "Click," and "Soft Timers." My congratulations to their authors.
We should all be grateful, as I am, to the student scribes who provided the following session summaries. The students were a pleasure to work with and very capable. They provided text that, typically, we edited only very lightly--as can be noted from the non-uniform format and level of detail, for which I bear responsibility. The scribes also gathered each speaker's slides, for regular papers, work-in-progress, and invited talks. The slides should be linked off the SOSP web page by the time you read this in OSR. Other material may also be there, such as pointers to the relevant home page or software. Gretta Bartels' eloquent summaries of the invited talks may have longer versions created; if so they will also be linked to the SOSP page: http://www.diku.dk/sosp99/ and http://www.cs.dartmouth.edu/SOSP99/.
I thank Darrel Anderson of Duke, whose name does not appear on a summary but helped by taking detailed notes that served us in other ways. I am very grateful to Eric Eide of Utah for his great assistance editing and formatting this report. We thank the scholarship and general SOSP sponsors: SIGOPS, Compaq, HP Labs, IBM, Lucent Bell Labs, Mercury Computer Systems, Microsoft Research, and Xerox PARC. Finally, we should all thank David Kotz for his great job as general chair, and all the other organizers who mostly worked behind the scenes. The next SOSP will likely be at Lake Louise in Banff or Sun River Resort in Oregon. See you there!
[next]