Hacker News Reader

Comment by Ericson2314 | Hacker News Reader

I'm always contrasting this with Matrix in my head. E.g reading https://matrix.org/blog/2023/12/25/the-matrix-holiday-update...

A lot of sort of economics/approach stuff contrasts to mull over in the background.

View on HN

I'd love to know what your line of thought is on this (from my pov as the author of that matrix blogpost).

For instance, we tried to propose Matrix as the basis of bluesky (hence https://matrix.org/blog/2020/12/18/introducing-cerulean/) - but obviously they went and built atproto instead. Meanwhile, both projects have effectively taken VC funding (Element, founded by the core Matrix team has raised from VCs - although Matrix itself is governed by the independent non-profit Matrix.org Foundation). I'm honestly entirely sure how they contrast together :D

TBH I'd be 100% more interested in Bluesky had they built it on Matrix since Matrix's decentralization makes a lot more sense to me and has more history behind it and battle-testing in real-world use cases like German government and military than AT.

Eh - they solve different problems. Bluesky is optimised to be able to aggregate likes across billions of tweets happening in a as-federated-as-possible way. That’s a hard technical problem. I don’t believe any existing federated solution can be “dropped in” to solve it. It’s tricky even in an entirely centralised system.

> Bluesky is optimised to be able to aggregate likes across billions of tweets

That might be part of the problem. It's self-evident why you need to do that if you're building Facebook or Twitter (a single platform where all the communication happens across a single graph), but that seems counter to how federation works (where everyone's social graph is different).

A few things:

- I think the revenue streams you all have from a Federated EU make an actually federated Matrix ecosystem much more likely :).

  - It's a pity that EU economic policy is so terrible, and constitutionally so. I bet Matrix's finances would be a in a much better position if European countries could run healthy deficits like the US does.

- AFIAK their excuse was that Matrix didn't have account migration. I wonder if Matrix having had that would have made them use it, or they were looking for an excuse for NiH anyways.

   - Conversely, now that they *do* have it, it does make Matrix at some point getting it a bit more important.

> their excuse was that Matrix didn't have account migration

Neither does BlueSky… or if they do “have” it, it’s never been used because there is still only one server.

You're not wrong but you're also not fully right: there has been a second instance that the devs have been using to test out the federation code. It's not yet open to the public because they are still working on the implementation. At the protocol level, it is very much there, but you are right to point out that the advantages haven't been fully realized just yet.

That makes sense. I guess I would just add that the users don’t know if it’s a good experience yet.

Maybe it sounds good on paper but in practice it amounts to a nothingburger and is essentially the same as manually moving instances in the fediverse.

It is very different than manually moving instances in the Fediverse, in my understanding. For one, it's totally transparent to the people that are following. All of your data comes with you. You don't lose your posts, your followers don't lose following you. I know Mastodon has at least recently gained account portability to some degree, but https://docs.joinmastodon.org/user/moving/ has some serious drawbacks that aren't present in the AT model.

In general, it seems like ActivityPub is terrible, and all the mastodon people are deluded or worse about its technical merits.

Activitypub is the federation equivalent of an at-home 3d printer. It’s easy enough to understand how it works and do stuff with it. And very hackable. Bluesky is trying to be more like Firefox: it’s much more complex and built by experts to be well optimised for the problem it’s solving. But it’s much harder for lay people to understand how it works internally.

I feel like that is extremely uncharitable but you are entitled to your opinion.

Well, you are more public persona than I, so you have to be more diplomatic :)

Transparent account migration is one of the most interesting parts of AT to me, as a user, so I for one am glad they went their own way.

That said I don't really know Matrix at the protocol level, so I can't really speak to it, but just based off of what you said.

So Matrix also has account portability (almost) - https://github.com/matrix-org/matrix-spec-proposals/blob/keg... and https://github.com/devonh/matrix-spec-proposals/blob/cryptoI..., implemented in Dendrite. Unfortunately dev is paused on it currently thanks to lack of $ though.

The AP approach (prioritising portable identities over portable account data) is cute though, and perhaps we should have prioritised that as an alternative to fullblown cryptographic IDs & account portability.

Thanks for the context! I use Matrix daily, so I am interested in these details, I just haven't found the time to dig into things yet.

The ideal situation is that BlueSky and Matrix (/Element, let's be real) are giving hoards of money and are dancing around each other in a daring game of adversarial interop.

I want a Cold War-scale federation competition, dammit! :)

I would love both BlueSky & Matrix (and/or Element) to be given hoards of money and for us to race each other to the best federation imaginable (and then bridge it all together and live happily ever after)

To be clear, strong account portability predates Bluesky by ~5 years. Peergos[0], as reviewed by Jay before Bluesky Inc was created, has had this for years:

https://book.peergos.org/features/migration.html

So they didn't actually need to go their own way.

[0] https://github.com/peergos/peergos

I've tried to love Matrix multiple times in the past 8 years or so. Sadly my current feeling is that the spec is simply too complicated to build reliable and performant systems on. Hopefully that feeling changes at some point. I really want it to be good.

i’d have been in danger of agreeing a year ago, but thankfully we proved otherwise with Element X: https://element.io/labs/element-x. Bit embarassing that we didn’t get there sooner, but human fallibility and all that.

Reply on news.ycombinator.com