Thank you, @annargrs for your work on this! I fully agree that anonymous review is important and worth preserving. (For #COLING2018, @LeonDerczynski and I preserved it all the way through, including for best paper awards!) https://twitter.com/annargrs/status/1283758686515732480

As we work in this constraint resolution space, though, I want to point out that one tactic available is to decide *not* to try to meet certain constraints. And I have a clear preference for which one to vote off the island:

You write "But NLP is moving at breakneck speed, and the authors would shoot themselves in the foot to not “publish” their work as soon as it was reviewed."

My take: while we're reengineering our field-level reviewing practices, let's also use them to promote slow science. >>

More bluntly: I see no value in supporting the ability of researchers to get things out quickly, and certainly don't want to bend over backwards for that. If it means that a large group of ML researchers abandon ACL, tant pis.