JunkThoughts

[JunkThoughts @thoughts_junk]

@nft_cryptogang @CarlHigbie stupid police

[JunkThoughts @thoughts_junk]

fuck the ICE agent who murdered an American citizen.

[JunkThoughts @thoughts_junk]

@iclr_conf support your decision! For fairness, we need to get rid of all reviewers' replies during the rebuttal period, because those are compromised too. keeping them is not fair for paper who have not received rebuttal feedbacks.

[JunkThoughts @thoughts_junk]

@DamienTeney @WenhuChen it is possibly a made up hoax. people are dramatic these days

[JunkThoughts @thoughts_junk]

@iclr_conf suggestion for
@iclr_conf: to avoid review issues going forward, ML conferences in general should eliminate the rebuttal mechanism, and let authors and reviewers pay more attention in their initial submissions and reviews. I think the rebuttal thing is useless anyway.

[JunkThoughts @thoughts_junk]

suggestion for @iclr_conf: to avoid review issues going forward, ML conferences in general should totally discard the rebuttal mechanism, and let authors and reviewers pay more attention in their initial submissions and reviews. I think the rebuttal thing is useless anyway.

[JunkThoughts @thoughts_junk]

@mahdi_tcs_ @iclr_conf We have acs to evaluate all the review comments and I do not think the end results/most reviews are very noisy. It is extreme to throw out the whole conference just because of noisy reviews.

[JunkThoughts @thoughts_junk]

@puneeshdeora @iclr_conf they should not have allowed ac to look at earlier reviewer-author discussion.

[JunkThoughts @thoughts_junk]

@iclr_conf I support ICLR's decision to revert scores where they were. This will get rid of many artificial high scores resulting from doxing reviewers, and is the only fair treatment for who has not received reviewer responses.

[JunkThoughts @thoughts_junk]

@mahdi_tcs_ @iclr_conf we need papers to make progress in research. I am sure good papers will have a higher probability of getting accepted. Initial reviews are not necessarily bad, many conferences only rely on initial reviews, no rebuttals.

[JunkThoughts @thoughts_junk]

fully support ICLR @iclr_conf decision to revert to original scores. The reason is obvious 1) reviewers had been forced to raise scores because of doxing, due to leak; 2) Other papers not having doxed reviewers cannot get score raised or reviewers have not responded yet

[JunkThoughts @thoughts_junk]

@shruuuum But the reviews are poisoned due to reviewer doxing. The only fair action is to go back to the initial score. Many authors have not even heard back from reviewers yet.

[JunkThoughts @thoughts_junk]

@mahdi_tcs_ @iclr_conf not a good idea, we need the papers.

[JunkThoughts @thoughts_junk]

@marwinsegler @chrsmrrs @iclr_conf but that is unfair for authors who have not got rebuttal feedback yet

[JunkThoughts @thoughts_junk]

@akshitwt it is good to reverse the score and get rid of the reviews, because many of them are poisoned or coerced.

[JunkThoughts @thoughts_junk]

@adilsoubki @paahrsa @iclr_conf Mass reviewers raised the scores under the pressure or threat of already being revealed online

[JunkThoughts @thoughts_junk]

what is ICLR's target acceptance rate?

[JunkThoughts @thoughts_junk]

@MAghajohari @iclr_conf no, a lot of them got affected. I have many reviewers who rushed to improve scores dramatically after the leaking under pressure of doxing

[JunkThoughts @thoughts_junk]

@sainingxie @BlackHC @iclr_conf ICLR should erase any discussions posted in the discussion period. It is unfair some papers have them and some do not. Some of those discussions are already compromised due to doxing.

[JunkThoughts @thoughts_junk]

@JunzheZhang12 @iclr_conf There are some corrupted reviewers doxed by others, who tend to give higher scores, and there are others who reviewers have not got around to respond, so the best is to revert back, maybe