ethical-ai

In AI research conferences are main way to publish, if I understand you correctly. It is generally considered to be a bad practice to submit "such an early draft" for peer review. Moreover, if it was accepted, it would be "circulating online" in conference proceedings. Why is anyone surprised that the draft which is considered not good enough to share with public by authors themselves is banned from sharing (in form of conference publication) by internal review?

> Though Bender asked us not to publish the paper itself because the authors didn’t want such an early draft circulating online

Should not you avoid submitting early drafts which are not good enough to be published yet?

I'm not sure why everyone refer to the process as "retracting" rather than "withdrawing", given the paper was not accepted yet. It sounds for me pretty reasonable to withdraw the paper which has flaws to fix and resubmit it, rather than rush into fixes before conference deadline (given that those fixes should probably be reviewed too). I didn't see any claims that the requirement was to never publish this paper. It also seems to be reasonable that the papers published by company are internally reviewed, like the papers by lab members are reviewed by PI before submission.

I see different claims from Google researchers, but it seems that at least for some internal review is used to access quality of the work [1,2]. It might potentially harm the freedom of the researcher, but once again, unless I missed something, there is no evidence that Google asked her to withdraw the paper and never resubmit.

As I already wrote in different comment, I can not think of a single reason to ask for reviewers identities unless she wanted to confront them personally.

[1] https://twitter.com/julianibarz/status/1334623651719643141 [2] https://twitter.com/vivnat/status/1334710506045669376

What I really cannot understand is how someone who works on ethics can think that revealing names of anonymous reviewer is OK?

Why would they need those names anyway? I'm also unable to think of any reason they might need the identities of the reviewers.