EDIT: As noted in the linked post, the defederation was a mistake and has been reversed. Thank you to the lemmy.world admins for reviewing this quickly.
I'm not a lemmy.world user, so I'm not directly affected by this defederation. I am a fan of both anime and the fediverse in general though, so I'm concerned about an apparent crackdown on my hobby and by what seems to be an increasingly damaging flaw in the fediverse model.
A few days ago. lemmy.ml defederated from ani.social, a lemmy instance specialized in anime. The only explanation given by Dessalines (lead dev of lemmy, owner of lemmy.ml and lemmygrad.ml) was that it was "full of CSAM". As far as anyone who's commented can tell, there doesn't seem to be any evidence of this whatsoever (see post link for an overview of the discussions). The only CSAM here appears to be in Dessalines' head.
That defederation was annoying, but not all that surprising. Dessalines and his fellow lead dev Nutomic are tankies, and tankies often seem to have a weird hatred of anime fans. The two of them have a history of making self-marginalizing decisions, so the obvious course of action is to just point them out so that people gradually abandon lemmy.ml, and hopefully eventually fork the lemmy codebase.
However, today I found out that lemmy.world had also defederated ani.social, again with no evidence presented for the decision. It looks like the LW admins are just rubber-stamping the bad decision of the lemmy.ml admins, without bothering to investigate at all.
We really can't afford to have the most popular lemmy instance behaving like this. The LW admin team has generally shown more professionalism in that past, so what gives?
Hey folks,
So what happened? We manage our defederation process with fediseer. It's a system that allows us, instance admins, to follow the defederation happening on other selected instances. We generally follow the defederation of sh.itjust.works, lemmy.dbzer0.com, literature.cafe, lemmings.world and a few other instances that have proven to be trustworthy in the past. However, in most of the cases I will also review them manually instead of blindly following them. That's why I added a censure for ani.social with tag "Loli" as was being reported - contrary to the "CSAM" that was being reported by Lemmy.ml.
When I was informed that might have been a mistake I delved a bit deeper and saw there was indeed no proof of any CSAM or even LOLI being posted - and I deleted the censure and we federated again. After this I made a post on the ani.social instance about what happened.
We always put censures in Fediseer to be as transparant as possible but we usually don't announce blocks on CSAM, Loli or Spam instances. Why? Because for Loli and Csam we don't want to give those instances extra exposure by announcing them. And for Spam, there's just too many and it would be hard to keep up.
Again, yes we were quick on the buttons but the csam/loli blocking is treated in a "better safe than sorry" way. We do this to keep our users and team safe. And everything was corrected within an acceptable time imo.
Fair enough. Thank you for the transparency and for resolving it quickly.