Everyone Wants Servers And Nobody Wants Servers
5 points by hongminhee
5 points by hongminhee
(I think) no one runs HA ActivityPub servers mostly because they are not available, they don't care, or they are more expensive. But there's no reason you couldn't have them.
I think people frame ActivityPub in a slightly "incorrect" way. In my opinion, "organizations" can decide to run an ActivityPub service for the organization and their members. By doing this, the organization and their members "own" the communication they perform on the ActivityPub service.
ActivityPub allows interaction between accounts on different services, yes, and this is pretty convenient, but if you think about it, ActivityPub IMHO works better when people use it to interact within the same service- no different moderation policies, no defederation, etc.
There is something wrong (in my head) with large open communication channels. I really don't want to have "public" conversations where anyone can post whatever I want publicly that shows under my threads. This works well enough for me, but we've seen so many times that this is used for evil.
(So I think the traditional web, with RSS/Atom are better for "public broadcasting" than ActivityPub. Unfortunately RSS/Atom are "dead to the general public". I kinda think that one of the best things we can do in this area is to promote RSS/Atom; make good feed readers 0-cost to people, campaign for everyone providing RSS/Atom feeds.)
There is something wrong (in my head) with large open communication channels.
I think you're probably right about this.
The protocol that did not design for a resilient network, and did not even specify the entities that cause the network to be resilient, turn out to currently be fairly resilient to DDoS attacks, while the protocol that does explicitly design for resilience turns out not to be.
Blacksky was mostly available while Bluesky was down, other mastodon servers were available while mastodon.social was down. For the vast majority a large portion of users on both networks, it was a downtime, and that has everything to do with the actual topology of the network being deployed than the protocol. Am I missing something?
I don't think it's equivalent for the two networks. Bluesky is a much bigger piece of the ATmosphere than Mastodon.social is of the Fediverse. Mastodon.social is much too big, at like 30% of the microblogging Fediverse, but that's nothing like a "vast majority". 70% of the microblogging Fediverse only knew about the downtime because it was fertile ground for discourse. (And I didn't hear about it until just now, because my main daily Fediverse interactions are on the Lemmy/threadiverse side.)
Had to take a look at the numbers again, you're right even based on active user count mastodon.social is only 40% of the "mastoverse." I think my broader point still stands though, this doesn't have much to do with the protocol, or at least not evidently so. bsky's self-hosting story (pre-packaged software, operation manuals) is just much more immature than mastodon's.
Part of that is because it's far more expensive to be completely independent from the server infrastructure of Bluesky-the-company than it is to be independent from the server infrastructure of Mastodon-the-company, which AIUI is at least partly due to protocol decisions.