When a server A federates with server B, it allows itself to scrape information off it and synchronise with it so that its users may comment, receive updates etc. Because these are individually run, and decentralised, each server will need to comply with its own local laws.
If server B decides to host a community with kiddie porn, Server A now will synchronise and share that information with its users, storing elements of the information on it’s own servers. This would put the owner of Server A at massive risk and would be a responsibility of the admin to keep themself protected and in line with the Law. Since you can’t control anything that happens on Server B.
If server B gets taken over by Russian spam bots, and they are able to comment wherever they wish, you wouldn’t have any control at moderating those accounts as they are not on your server. It’s easier to just block that server entirely.
Defederating but allowing commenting means that you aren’t actually disconnecting, and keeping a tunnel open and swapping data and updating fresh comments.
It’s a gray area as the line can become blurred with things like The Donald, which might be largely unpopular and racist but it doesn’t break any laws. so deferedating willy nilly can easily become a tightrope walk between protecting users from bad actors and quelling free speech. That’s why we can block those communities ourselves.
True that it’s not the perfect system and as servers connect and disconnect from one another, communities may fracture and users may need to create new accounts or have multiple which can be inconvenient, but it’s the most logical system for now.
There may come a time where a middle step is needed, once instances become so large that defederating can be a massive issue, it may be nice for an instance to block off communities from other instances while not actually fully defederating.
I too have been sent here. It’s at the top of hot right now.
Seems to be a Lemmy zombie post glitch that needs to be addressed.