Creating an SOSP program consists of at least the following steps: pick a program chair, pick a balanced committee of outstanding scientists, refine the call-for-papers, read the papers, write comments about the papers, review the comments, rank the papers, meet to refine the reviews and rankings, choose the papers, choose TOCS recommendations, choose posters, conduct shepherding, receive final papers and organize proceedings. Each committee member read all submitted papers, and the committee together wrote more than 250 pages of 9pt single-space 2-column comments about the 84 submitted papers. These comments were our guide to discussion and selection, and were also sent back to the authors of each paper so that all could learn. (Several authors thanked the committee for the unusually detailed feedback.) During the committee meeting itself we had these comments in front of us, and we also used a wall-sized poster of information about the papers to help reach consensus.
The most incredible part of reading these 84 SOSP papers is how interesting they all were! We had almost no poor papers, and reading all 84 is a lot of fun and leads to a feeling of, for a few moments, really being on top of the O.S. field.
These papers were so excellent that we recommended four papers for publication in the ACM Transactions on Computing Systems, and we extended our program of 22 papers with 8 additional posters to be presented only to attendees.
I hate to always do the same old thing, and yet SOSP is not broken. I made only a few small changes:
I continued the fifteen year tradition of asking all committee members to read all papers. We again have a Work-In-Progress session, organized by Mike Jones, and a debate, organized by Mary Baker. And of course the conference tradition of animated conversation, sometimes long into the night, will continue.
It is my fond wish that all future SOSP's are online, publish a CDROM, and include large amounts of source code that enable independent researchers to verify and surpass SOSP results.
This may be the last SOSP to use so much paper. Although I had a few requests from authors who wanted to mail me the Postscript of their papers, my experience is that Postscript does not work for universal printing exchange without the occasional intervention of a human. I don't know how many times colleagues have mailed me Postscript, only to have it break on a local printer because their word processing software generates Postscript with imbedded 68000 binary instructions, assumes strange fonts, or otherwise is broken. I did not want to be responsible for the hand fixing and verification of 84 papers.
This year I put the final proceedings together in the traditional way: starting with the final hardcopy from authors, then adding the front matter (e.g. this "foreword"), writing page numbers in pencil on the back of each sheet, and forwarding the resulting stack of papers to ACM for copyright notices and printing. (Something new: all authors got to review the draft front-matter on the web and make corrections before I sent it to ACM). I would have been willing to work with 22 papers of Postscript and handle the whole thing electronically, but it would certainly have been more work than the paper method. The benefit would have been more uniform, and in many cases higher quality, output.
Perhaps by SOSP '97 these problems will be solved, or a more energetic program chair will have the energy to go the last 10%.
It has been a pleasure working with the general chair, John Bennett, who helped set the pace and whose arrangements I look forward to enjoying. Many people emailed me with helpful SOSP suggestions; past SIGOPS Chair Hank Levy, and past SOSP chairs Barbara Liskov and John Ousterhout, provided especially significant advice (not all of which I followed). The program committee was a joy to work with, giving weeks of time and decades of wisdom with humor and energy. My deepest thanks go the wonderful Operating Systems community without whose papers and willingness to volunteer there would be no conference at all. I joined the O.S. community only ten years ago, relatively late in my career. In the vitality, warmth, and unmatched striving for excellence of its people I continue to find a home.