I really like the idea! I'm not sure if I'd like to try it, though. I would fear that this might deteriorate into finger pointing / defensive behavior. Especially if there are lots of Commiserations about a single person / small group. What can I do to keep the overall atmosphere as positive as possible?

From your description, it sounds like you are doing very little estimation, and for the right reason. I don't know your exact circumstances, but I think we are in a similar situation. Except that we maybe produce the estimates not only for the right reasons. I also see some value in the "estimation process" my team uses (1.5 hours every two weeks, planning poker) - especially the discussions that happen there. And there's maybe even some value in the estimates we produce. But: I think we could get away with less. (Unfortunately, I am the only one who wants to try it). Our average user story is 2.something story points, and we usually accept stories <= 8 story points for development. So I think we would not lose much precision when we just estimated stories as "small enough for development" / "too big". Then we could move the (valuable) discussions closer to when the work happens - like, right after we start working on it. So, the real #NoEstimates question for me is: "Could we get away with a lot less than we do now". And the answer to that is probably often "yes". But maybe I didn't understand #NoEstimates correctly.