Tags: * shorten review period for resubmissions * release reviews earlier * for discussion, author-reviewer * for author response * improve matching * enforce reviewer diversity * improve interface for opt-in / opt-out
"First, for me the ""resubmissions"" should have shorter reviewing period if the same reviewers are
assigned one hand (depending on previous scores, e.g., minor revision vs major revision need). On the
other hand, it would be nice if the reviews are released as soon as the reviewing period ends without
waiting for a meta-review (similar to the process before ARR), followed by a short ""response period"",
and then meta-reviews. I also think that introducing again an author response would help. This can reduce
the number of resumbissions, as now a missed details either by the reviewers, or by the authors in the
paper itself may cause lower scores, hence rejection. My previous experience with ""author responses""
was very positive, on one hand it helped me to understand better the work I'm reviewing and clear or
confirm important considerations that I had. On the other hand, as a author I felt I have an option
to convince the reviewers in the worth of my work. I also feel that the way currently reviewers are
assigned is not working great: (1) I think the automatic assigning of the papers is not great for researcher
working on multiple topics. My observations is that I'm getting papers to review, not based on my latest
work or interests but more on a intersection of other narrow topics I am (was) working on. I know I
can control this with excluding papers, however you have to tune this over multiple cyclesSEMICOLON
(2) I had cycles where I and co-authors of mine got assigned to review the same paper (probably because
of (1)). This, I believe, is not helping making the review process more fair. Moreover, reviews don't
know about this before the reviews are in, and they cannot take any action and ask for re-assignment
in time. (3) the opt-in, opt-out process should be improved, I am sure it's a limitation in OpenReview
at the moment, however for me it's not easy to follow how many papers can I expect in the next cycle,
i.e., as this is based on a response in a online form -- if I'm reviewing for this month or not, have
I filled in the form properly, should I fill it again this, etc."