If you're thinking about setting up your own single-user Mastodon instance, I urge you not to do it. The user experience is so broken that it's baffling that anyone would find it acceptable.
I agree with this take, and recently I actually read this article that criticizes how server centric fedi is as a whole. If it’s hard and expensive for a layperson to self host, but you need to have an account associated with a specific server, then you’re going to end up with a system where you’re under the whims of a instance owner still. Not to mention the whole pick a server step severely hurts our adoption rates.
I like the idea of having an account just being a public and private key pair. Theoretically you could make one client side, use it to sign your messages, and servers could verify the signature and distribute your post without needing to have an explicit account for you. You could send every message to a random instance and it’d still work. You wouldn’t have to worry about links to the “wrong instance” and you wouldn’t have to attach your identity to a instance that might shut down or be bought by a bad person. The server would be essentially irrelevant.
I align with that article 's conclusion; in fact such a “fediverse browser” is exactly what I think the fediverse needs to fully replace closed/proprietary/traditional social media.
However, some of their arguments seem off. For example, for the client to be able to choose/implement it’s own sorting algorithm, it seems to me that it would need to have access to all posts. At that point, your client is just another server, with all the problems that we’re originally trying to avoid.
I have the same problem with your proposal / nostr’s approach: you may obtain a portable identity but all the “content” tied to that identity still has to live somewhere - someone else’s server or your own.
Yeah, I disagree with that part as well. I think it’s fine for servers to store the content and provide endpoints for specific queries/sorts, and expecting the clients to have all the posts is a tad extreme.
In this case, yes the data needs to live somewhere, but that’s the nature of having data be retrievable.
Nostr does some interesting things! What I mentioned here is actually just the identity part of what I think could be a significantly improved version of the fediverse. I have ideas on how to support subreddit style communities and decentralized moderation and things like that that make the whole idea a bit different from nostr.
I suspect retrofitting a whole new identity system to Fediverse will never happen because server admins, or instance admins, will come up with all kinds of reasons why they don’t like the idea of not knowing who their users are. Some of them would probably allow it, but I bet a whole bunch of them wouldn’t, and we’d get into this fragmentation where some servers won’t allow posts from those types of identity, etc. It seems to me much easier to take Nostr and just give it the functionality you get inside the Fediverse.
I agree with this take, and recently I actually read this article that criticizes how server centric fedi is as a whole. If it’s hard and expensive for a layperson to self host, but you need to have an account associated with a specific server, then you’re going to end up with a system where you’re under the whims of a instance owner still. Not to mention the whole pick a server step severely hurts our adoption rates.
I like the idea of having an account just being a public and private key pair. Theoretically you could make one client side, use it to sign your messages, and servers could verify the signature and distribute your post without needing to have an explicit account for you. You could send every message to a random instance and it’d still work. You wouldn’t have to worry about links to the “wrong instance” and you wouldn’t have to attach your identity to a instance that might shut down or be bought by a bad person. The server would be essentially irrelevant.
I align with that article 's conclusion; in fact such a “fediverse browser” is exactly what I think the fediverse needs to fully replace closed/proprietary/traditional social media.
However, some of their arguments seem off. For example, for the client to be able to choose/implement it’s own sorting algorithm, it seems to me that it would need to have access to all posts. At that point, your client is just another server, with all the problems that we’re originally trying to avoid.
I have the same problem with your proposal / nostr’s approach: you may obtain a portable identity but all the “content” tied to that identity still has to live somewhere - someone else’s server or your own.
Yeah, I disagree with that part as well. I think it’s fine for servers to store the content and provide endpoints for specific queries/sorts, and expecting the clients to have all the posts is a tad extreme.
In this case, yes the data needs to live somewhere, but that’s the nature of having data be retrievable.
You realize you’re describing Nostr right?
Nostr does some interesting things! What I mentioned here is actually just the identity part of what I think could be a significantly improved version of the fediverse. I have ideas on how to support subreddit style communities and decentralized moderation and things like that that make the whole idea a bit different from nostr.
Agreed, but that is exactly how NOSTR identity works, and the reason I think NOSTR is such a great protocol.
I suspect retrofitting a whole new identity system to Fediverse will never happen because server admins, or instance admins, will come up with all kinds of reasons why they don’t like the idea of not knowing who their users are. Some of them would probably allow it, but I bet a whole bunch of them wouldn’t, and we’d get into this fragmentation where some servers won’t allow posts from those types of identity, etc. It seems to me much easier to take Nostr and just give it the functionality you get inside the Fediverse.