I certainly am not saying that server operators should feel obliged to content they do not like, especially if they believe it is illegal or immoral. After all, a huge draw of the Fediverse is the fact that you get to choose, right? Sure, personally I think all obscenity law is weapons-grade bullshit regardless of how despicable the subject matter may be, but also, server operators shouldn't feel pressure to compromise their ideals, attract a crowd of people they simply don't like, or (of course) run the risk of breaking the law in their jurisdiction, so what happens on the Fediverse seems like it is the right way for things to go, even if it is harmful to the federation in the short term.

But that's kind of the double-edged sword. You either have centralization where someone decrees "the ultimate line" or you don't. With Bluesky, there's a possibility that it will wind up being decentralized properly, but it could wind up being defacto centralized even if they uphold their promises and I think that strongly devalues the benefits of decentralization where they count most. Today, there is in fact one company that holds the line, and it's unclear if that's going to meaningfully change.

There are some aspects of AT proto and Bluesky that I think are extremely cool: an example is identity, identity in AT proto is MUCH better than it is in ActivityPub right now. However I'm not surprised they are not going to acknowledge this problem. Just know that they are 100% aware of it, and my opinion is that they really do want to find an answer that won't piss everyone off, but they also probably want to avoid the perception that Bluesky is a haven for degenerates, especially early on when there are less network effects and they desperately need to appear "better" than what is already out there. Unfortunately, I can only conclude that their strategy is most likely the best one for the future of their network, but it still rubs me the wrong way.

IMO, the point is, all of this architecture and the headache of decentralizing it is a waste of time if the most critical part for actually ensuring user sovereignty on a network winds up being centralized anyways. And based on the replies so far, what I'm getting is that nobody at Bluesky has a clear idea why anyone else would ever run a relay other than for the good of society. That's nice, but of the people running relays, none of the other entities that might run relays have the same level of stakes in it as they do. Bluesky is a company that presumably needs to eventually make money to stay afloat, but they basically built out the entire protocol, will run the app, and all of that jazz. They are the face of this. The users that are on Bluesky right now are by and large going through their infrastructure; presumably millions of users. Why in the world would someone else run a relay in this condition? You'd essentially be doing it to mostly support a network that is vastly controlled by one single entity, and there is no obvious reason why it will stop being vastly controlled by one single entity, too. It is a problem that the vast majority of users are on a small number of Mastodon servers, but with Bluesky there's little reason to believe right now that a vast majority of users ever won't be using Bluesky's infrastructure. If we're going to compare it to e-mail, then this is like if the only e-mail provider you could use for years was just GMail, and then slowly other providers were allowed to get whitelisted in. It is nice that the design of the network is so amenable to the potential of having it not be controlled by one entity, but there's the rub, I don't think they actually have any incentive to actually make serious efforts to fixing that problem. I don't know if they would even with substantial pressure from users to do so. Right now their messaging is basically "There might be a few entities that would run relays" suggesting that it is basically out of reach for people not at the scale of Google.

One could easily misread this thread to be about fringe content or more charitably, censorship, but I think the most accurate way to sum up my concerns is this: I don't think Bluesky's own incentives, nor the incentives that they are laying out with this design, are particularly fantastic for the health of the network they are building. It's not impossible that things could wind up working out anyways, but I worry that this is not actually going to be a sustainable system and that if there isn't substantially more thought put into this entire operation, it may basically wind up being just a very elaborate and overly complicated way to re-implement the exact same problems that centralization has.

That's not to say I believe they are working in bad faith or that they don't care or anything like that. Like I said, I assume they are actively trying to figure out how to make everything work without pissing everybody off. And I'd love to be highly optimistic and say that it will all work out, but I have my doubts. The silence is deafening.

I fully agree with this but I'm giving Bluesky a lot of benefit of the doubt because I think Mastodon has even worse incentives and I can't come up with anything better. (I can come up with different tradeoffs but they always leave somebody pissed off.)

Getting back to the Japan problem, I think "content we care about is blocked on the main relay" should be enough incentive to run an alternate relay and maybe a patched client. Obviously there's a bootstrapping issue there but you could say the same about Bluesky vs. Mastodon and yet here we are millions of users later.