not that there aren't other problems here, but mastodon is one of the world's largest open source socio-technical systems and we've got people banging it with wrenches to try and make people work correctly.
a mastodon server is like a pet wolf: a bad idea for all involved unless you really know what you're going into and also have maybe taken more than a few zookeeping seminars. Because it takes a team. at best.
Hachyderm grew to 40k people and now needs basically a volunteer SRE/Devops team -- the events have a higher time between them, but some things cause the server to absolutely be slammed -- someone shared a popular thing, or a popular post was shared to other servers.
the protocol is not particularly efficient, and wasn't intended for a popluation this big across the fediverse, in a lot of ways.
This is why most instances are closed to new members -- once you're medium and not small, there's a hard cap based on donation income (you need better infrastructure, and to pay for that bandwidth) or volunteer labor willing to give free labor or free hardware. Hachyderm has some legit corporate donations because some guy from digitalocean just happened to see Nova struggling and offered infra and my impression is Nova is still slowly burning herself out (Even with a team).
this isn't like self-hosting SyncThing or a polycule media server. I think a lot of people don't get specifically that -- but mastodon has big factors that make it much more complicated. you have several exponential factors: local members, servers that federate with you, high-following accounts on big servers, external things that posts were shared to, etc. But also? No one tells you things will suck when you get big, until they get big.
As I wrote here:
https://cohost.org/NireBryce/post/1010134-anyway-related-to-t
a big part of the masto problem is that even server admins were falsely advertised to, imo. They weren't told just how deep they were getting into things. And users don't realize the costs of those things.
It shouldn't require that much, if we're talking about what's best for the world. But best isn't what we've got, instead with the overarching Mastodon project, people have been left with a Rails monolith that has a (largely) uncooperative project lead who doesn't really understand delegation. (Not Nova, she's great so far, but Eugen.)
Which isn't to say the masto.lol or whatever admin's reasoning is good. But that, well, the discourse around it outlines a failure cascade, and that's worth learning from even if the moderation techniques aren't.
(A talk Nova did about the hachyderm oral history is at https://www.youtube.com/watch?v=5KU10K3EXK4)
