Message from the Program Chairs

Once again, we coerced colleagues into serving on the program committee and were rewarded with an outstanding group -- 7 from industry and 6 from academia. They each read and reviewed an incredible number of papers and produced often extensive comments for the authors, whether or not the papers were accepted. We deeply appreciate all their efforts.

As with the second OSDI, we requested full papers instead of extended abstracts. These were reviewed in two rounds. In the first round, each paper was read by at least three members of the program committee, and often by external referees as well. At the end of the first round, about two-thirds of the papers were selected for review by at least another three program committee members.

Armed with reams of paper, the program committee met for a day and a half at Harvard University in Cambridge, Mass. By that time, papers that had high variance in the reviews had been read by even more committee members, and several papers that were heavily debated on the first day were read overnight by still more members. Many a committee member gave up valuable hours of sleep so that we'd have the very best data possible on all the submissions. By the end of the second day, we had selected 20 papers out of the 96 submitted.

Two of the accepted papers were co-authored by program committee members, out of six such submissions. As in the past, a significantly higher standard was applied to these submissions, and program committee authors did not know who their reviewers were, nor did they see their reviews or participate in discussions of their papers. In fact, they learned of the fate of their papers only a few hours before the rest of the authors did.

There was one paper selected for fast-track publication in the ACM Transactions on Computer Systems: "IO-Lite: A unified I/O buffering and caching system", by Vivek S. Pai, Peter Druschel, and Willy Zwaenepoel, of Rice University. We thank Tom Anderson for serving as our official interface between OSDI and TOCS.

Another tradition, that of "shepherding" the accepted papers, was continued this year -- each paper was further revised after acceptance, based on the feedback from the reviews and with the guidance of a program committee member.

In addition to the refereed track, we have augmented the program with a Works-in-Progress session, organized by John Hartment, and an invited keynote by Jim Gettys. The keynote, "The Blind Men and The Elephant", discusses the intersection of the World Wide Web not only with operating systems but with other computing fields as well. We hope that these sessions will stimulate interesting and heated discussions both during and after the conference.

As is always the case, there are a number of people behind the scenes without whom such events never take place. Liz Pennell of Harvard University did an extraordinary amount of work processing the submissions, arranging for the program committee meeting, collating papers and reviews, interacting with authors, and picking up all the pieces at the end of the day. David Sullivan was the program committee scribe, taking notes of the committee's discussion so that authors would be provided as much feedback as possible. And, of course, there is the amazing Usenix staff: Judy DesHarnais, Ellie Young, Jane-Ellen Long, and Toni Veglia -- they make putting on a conference seem almost easy.

Last but certainly not least, we would like to heartily thank all the authors who submitted their work, without whom the tradition would not continue.

Paul J. Leach
Margo Seltzer

January, 1999