ICML this year had no Borderline rating, I take it you prefer this setup as an AC? What if two reviewers put WA and two put WR and none of the reviewers particularly care?
IMHO the most important:
(1) Give actionable feedback to the authors (what needs improving, what parts are unclear, what results/experiments are lacking).
(2) Be clear whether this can be fixed in the rebuttal/final version. If not, be clear with a "reject" so they don't waste time on the rebuttal.
Good advice! My biggest problem is with papers that are technically pretty okay, but not so interesting that I would want to discuss them at a poster at the conference. I.e. the cockroach referred to by Freeman below. https://visionbook.mit.edu/how_to_write_papers.html
Comments
The rating can stay borderline, but a motivated forced decision greatly helps me understand and consider the reasoning, and lead the discussion.
(1) Give actionable feedback to the authors (what needs improving, what parts are unclear, what results/experiments are lacking).
(2) Be clear whether this can be fixed in the rebuttal/final version. If not, be clear with a "reject" so they don't waste time on the rebuttal.