100 likes | 224 Views
Report on SIGMOD 2005. DeWitt reporting for Widom. My True Feelings About Panels. What does “ PANEL ” stand for? P ain in the A **, N othing E ver L earned. Report on SIGMOD 2005. Notable Changes from past including PC Groups Reviewing Load Mandatory PC Meeting Detailed Review Form
E N D
Report on SIGMOD 2005 DeWitt reporting for Widom
My True Feelings About Panels • What does “PANEL” stand for? • Pain in the A**, Nothing Ever Learned
Report on SIGMOD 2005 • Notable Changes from past including • PC Groups • Reviewing Load • Mandatory PC Meeting • Detailed Review Form • Review Monitoring by PC Chair • Author Feedback • No quota on Number of Papers Accepted
PC Groups • 57 PC members divided into 9 groups w. goals: • All papers in an area (e.g. XML) went to the same group • Papers balanced among groups • Higher than normal load since smaller than normal PC • Group leaders • Assigned papers to group members • About 10% of the paper assignments were to members in different groups • Monitored reviewing process, initiating and moderating discussions of controversial papers BEFORE the PC Meeting • Ran group meeting at the PC meeting
PC Meeting • Mandatory attendance!!!!! • Made clear when PC invitations were extended • All 57 PC members showed up • Meeting had 3 phases: • Separate group meetings to decide on papers to rate papers as accept, possible accept, or reject (4 hours) • Group leader meeting with Widom to decide on PC papers rated “possible accept” (1 hour) • All PC meeting to decide fate of non-PC papers rated as “possible accept” (2 very long hours) • Followed by a one day symposium for young faculty to pitch their stuff
Review Process • More detailed review form • Separate questions for technical correctness, depth, novelty, impact, … • 3 strong points and 3 weak points were required for each paper • Automated review monitoring by Widom for “harshly worded” reviews • Many false positives but her goal of “kinder, gentler” reviews was certainly achieved
Author Feedback • Text portion of reviews returned to authors one week before PC meeting • Goals: • Minimize technical errors by reviewers • Eliminate unprofessional reviews • Authors allowed to respond (up to 4000 characters) • Minor battles between authors and Widom over whether LF characters counted!!! • Authors felt obligated to respond (about 75%)
Feedback Redux • Need 2 weeks: 1 week for author to prepare feedback and 1 week for PC to digest feedback • 4000 characters too long • Authors probably not satisfied with effect of feedback on outcome of their papers • Almost certainly had a positive effect on tone of the reviews in the first place
Paper Quota • Like VLDB 2004, no preset limit on how many papers would be accepted • Like most recent conferences acceptance rate ended up being 15%
DeWitt’s thoughts • Best SIGMOD in years • There were no talks that I listened to that led me to think “they accepted that?” • Why? • Face-to-face PC meeting with 100% attendance • Group organization led to better decisions • Author feedback encouraged reviewers to write careful and thorough reviews • Widom is a tough taskmaster