170 likes | 195 Views
This paper addresses the flaws in the current paper selection process used by program committees, highlighting issues with conflating reviewer standards and making early acceptance decisions without sufficient information. The proposed alternative advocates for comparing papers based on relative quality and deferring decisions until all submissions are evaluated thoroughly.
E N D
what PCs do now Paper Ratingvs.Paper Ranking what I am advocating John R. Douceur
Step 2: PC Meeting 2.0 0.9 3.3 3.2 1.6 2.3 1.9 2.8 1.5 1.3 1.4 1.0 0.6 1.9 3.9 3.6 1.0 3.0 3.1 2.5 2.8 2.2
Submission Quality Varies per Year papers papers papers 2006 Quality Quality Quality good good good bad bad bad 2007 2008
Problem #1 with Ratings The reviewing process conflatesreviewers’ standards of stringencyand leniency with the reviewers’assessments of the merits andweaknesses of each paper.
Problem #2 with Ratings In the PC meeting, early acceptancedecisions are based upon insufficientinformation; these decisions becomeentrenched for psychological reasons.
What’s the Alternative? I’m glad you asked...
PC Meeting – Phase I I thought was this paper much weaker than This paper has that one. a great evaluation. How well evaluated are higher ones?
PC Meeting – Phase II min max
Summary Rating what PCs do now Conflates reviewers’ assessments with standards. Entrenches decisions based on insufficient data. Ranking what I am advocating Compares each paper to its competition. Defers decisions until relative quality is known.