@tbodt @j0ebaldw1n But he doesn't need market feedback, he's literally cloning Discord. It already owns the market.
Discussion
Loading...
Discussion
@j0ebaldw1n @lina my single concern is that "coder guy builds first product in solutide for 6 years without market feedback and immediately flops" is a trope
@tbodt @j0ebaldw1n But he doesn't need market feedback, he's literally cloning Discord. It already owns the market.
@lina @j0ebaldw1n you're probably right!
@j0ebaldw1n@mastodonapp.uk The issue with such a federation model imo is, that it doesn't scale to people running servers temporarily. I have a lot of friends, who temporarily hosted a Matrix server until they got tired of it and switched to a different server hosted by someone else. We still use mostly the same chat rooms as we did originally. With something like XMPP or the linked federation model, that would have required a migration every time. Now, that isn't as big of a concern. Switching rooms is okay and you have to do it eventually anyway. But it is just a lot cooler, if you don't have to do that and people can host their servers with unreliable internet at home and it still mostly works fine.
Obviously I am biased there as a rather technical user. For most normal users a federation model like the linked one is probably the better UX. But imo it does encourage more centralization, which I think is an issue in the very long-term. Imo there should be no difference from me hosting my own server to store my data or registering an account on an existing server.
@deepbluev7 @j0ebaldw1n I mean what Matrix tries to do is cool and all but the Fluxer dev is very right and pragmatic in that it doesn't work and would get in the way of building the product people want.
If it were easy to build true reliable distributed message federation, that would be one thing. But Matrix has been trying and failing to make it reliable for 11 years. So that's not going to work. Never mind the moderation problem...
What I want to see is just built in data migration. An instance to instance flow for one shot migration of guilds, home accounts, and DMs. It won't help if a server goes offline suddenly, but it covers the important issue of instances winding down. The good part is this can be built after the fact, it doesn't have to be a core design feature upfront so it's not blocking.
But yeah at the end of the day, it does require server operators to be responsible. FWIW I host a Synapse server that used to live on a Mac Mini at a friend's apartment. It has now moved through two production servers and is now on redundant infra. Keeping servers alive is not *that* hard if you care... and if you don't, well, you probably shouldn't be hosting stuff for people if they don't know the risks!
@lina@vt.social @j0ebaldw1n@mastodonapp.uk I don't think the core issue with Matrix is the federation. It certainly does add complexity, but I do think it is something that could be done well and Matrix suffers a lot from being somewhat the first system trying to do this kind of federation.
I think the major issue is, that Matrix never focused on making that work well and instead got distracted by trying to add other features (E2EE, video calls, threads, widgets, etc). That's something I think the Fluxer dev gets right. You need to focus on which features you want to polish and what you can actually achieve with the resources you have. Matrix lacks that focus and as such a lot of the Matrix features are half baked. So it is probably the right call to skip on federation, however I still think a distributed federation can add a level of reliability, that we simply lost from services nowadays.
I have been dependent on services hosted by others way too often and it sucks when those become less reliable over time or even got shut down for various reasons. I think most of us have been there, where we set up a service for friends or family, but then because of life you may need to deprioritise it 5 years later.
But I will wait and see, what comes out of Fluxer. My primary concern is that federated usage will feel like a second class and have quite a few usability downsides. I think only Matrix and Email have come close to federation feeling seamless (if you don't need to operate the servers). However, the Fluxer dev seems to have the right ideas in a lot of ways and I do very much suffer from being too developer focused in how I approach technology and if the end result turns out great, I will change my mind on this. But otoh we have a lot of open-source chat platforms and even without federation, it is not an easy task.
> I think most of us have been there, where we set up a service for friends or family, but then because of life you may need to deprioritise it 5 years later.
This is true, though the only time it really happened for me was email (I got tired of self hosting that after years of it). This has the same issue as Fluxer though, and that's why I'd like a migration path, just as with email you can change the MX and imapsync the mailboxes over. I've never just downed something without giving advance notice. In fact I still have selfhosted email on one host and I'm currently tracking down the few accounts relying on it before I shut it down (it's mostly on me, it's not that the people aren't responsive).