What is hard to understand about the Feed Our Feeds BS is what Cory Doctorow @pluralistic, who normally is quite outspoken about these tech bro efforts to coopt legitimate social technology, is doing supporting it.

Mastodon Migration

Hmmm... sorry, but not understanding you. Unless the in house developers can prove multiple relays work in a loaded production environment with their technology, it is very hard to imagine an outside organization doing it.

Solarbird :flag_cascadia:

@mastodonmigration @joeinwynnewood @pluralistic The idea that their Relay isn't internally a fleet of coordinated - even mirrored maybe - Relay instances is... untenable. That's just not how server load balancing works. Literally no one would do that at the scale at which they're operating, and more importantly, want to operate, and I'm not even sure anyone even could.

Therefore, we already know it can work _within_ a single stack.

The next step is to build another, separate stack. And if you want that to be independent - _actually_ independent - you want that done by a separate organisation.

Just as was done for every other meaningful internet protocol.

Mastodon Migration

@moira @joeinwynnewood @pluralistic

Don't think that is how it works or what is meant by another relay, but admittedly not an expert on this stuff. Perhaps check out Chritine's paper on the AT Protocol.

Solarbird :flag_cascadia:

@mastodonmigration @joeinwynnewood @pluralistic I'm not saying it's the same thing. But it is the step you take _before_ taking the next step.

There is not one company in the world that is going to go build an entire new separate stack of its own protocol handling backbone solely for compatibility demo. How do I know? _Because I've worked this job_. You go buy _other people's versions_ and test against _them_, because that's the only thing that doesn't completely waste your time!

And right now there _isn't_ another stack, so people are trying to raise money to build one.

(1/2)

Solarbird :flag_cascadia:

@mastodonmigration @joeinwynnewood @pluralistic

People _have_ put up mini-second-Relays using the exact BlueSky code base, for experimentation purposes. At a tiny fraction of total traffic, it worked. The only real way to find out if it's going to work at scale, though, is to try it. And it probably won't at first, and then you _fix things_ until it _does_, or you figure out it's not gonna.

Expecting BSky to build a new second Relay stack to "demonstrate" anything before anyone else tries that is just nonsense. They have better things to do.

(2/end)

The Nexus of Privacy

Yeah. People are putting up larger-scale relays as well -- @edavis.dev recently mentioned experimenting with a non-archival relay for < $30/month (not bad at all!). Rudy Fraser has also posted about looking at doing a non-archival Relay for Blacksky. So I agree with your view -- there are llkely to be a few bumps in the road but there don't appear any barriers to solving the technical problem of running an independent whole-network Relay.

What's still an open question is the economic motivations for anybody other than Bluesky to running a whole-network Relay. The Free our Feeds approach of trying to get the community to pay for it as a public good is interesting. From Bluesky's perspective, it's great to have them be an independent organization -- and if the reports they're on track to closing another round of funding are accurate, it's very timely.

Of course Free our Feeds approach of having most of the "custodians" be AI-focused doesn't encourage confidence. Neither does the heavy involvement of Mozilla. So we'll see how it works out in practice.

@moira @mastodonmigration @joeinwynnewood @pluralistic

Joborg

@joeinwynnewood @jubei @cellularmoose @mastodonmigration @cwebber
The expensive part is "storing every post ever made on the network and its cryptographic signatures". A selective relay (only collecting posts its users want, as in Mastodon) would cut this by a lot, or it could just keep the last week or so of relayed posts.
But for that to happen someone has to build it.

Grassroots Joe

@joborg @jubei @cellularmoose @mastodonmigration @cwebber

Re: as in Mastodon

So IOW recreate ActivityPub (AP) in the middle of the rest of the Bluesky stack which seems a bit silly when AP already exists and is quite successfully deployed (as Christine notes someplace in her posts).

Joborg

@joeinwynnewood @jubei @cellularmoose @mastodonmigration @cwebber
If you will, though from a technical perspective ATproto is a better protocol, and more successfully deployed (in terms of user count and resource usage). ATproto already has distributed outboxes (and labellers, blocklists et c. missing here).

Joborg

@joeinwynnewood @jubei @cellularmoose @mastodonmigration @cwebber What is needed for a turnkey small server is an inbox implementation (AppView in AT lingo) with a cache.
Implementing this would enforce protocol compliance/interoperability on Bsky and thus restrict their ability to harm their users, as well as enabling independent alternative providers (including "server migration").

Joborg

@joeinwynnewood @jubei @cellularmoose @mastodonmigration @cwebber I'm hopeful for this to happen, since the current Bsky team seems open to this kind of work. Of course, the question is whether that lasts long enough to get that fire escape in place...

Mastodon Migration

@joborg @joeinwynnewood @jubei @cellularmoose @cwebber

Don't understand all the technical terms you are spooling out. Experience shows that when someone is saying something is simple to do, but it hasn't been demonstrated yet, then it is good to be wary.

If Bsky really wanted this to happen, they would at least have a multiple relays operational and a convincing story about making the technology public and, lockboxing it. Short of that it is just a 'trust us' play.

more...

#FreeOurFeeds

Mastodon Migration

@joborg @joeinwynnewood @jubei @cellularmoose @cwebber

Which is kind of crazy because the entire rationale of the foundation is that you can't trust Bsky.

Edit: Let's say that again. The entire basis of Free Our Feeds is that the public needs to invest huge sums of money in a technology because you can't trust the developer of that technology. ‍️

#FreeOurFeeds

Joborg

@mastodonmigration @joeinwynnewood @jubei @cellularmoose @cwebber
Not having dug into the details I cannot confidently say that it's easy, but at least AT was planned for multiple implementations, service providers and applications (cf pixelfed et c.).

As I understand it Bsky are actively trialling different relays. Their core technology is public and open source (on github) but not adapted to and packaged for small servers yet.

Mastodon Migration

@joborg @joeinwynnewood @jubei @cellularmoose @cwebber

Understand this is their pitch, but until there are multiple independent full scale independent nodes, it is still vaporware.

marcusb

@mastodonmigration @joborg @joeinwynnewood @jubei @cellularmoose @cwebber I think its worse than that. The rationale of FOF is to invest a bunch of money in a technology because you can’t trust its creator to be benevolent. But, the thing you’re building will only work if you can trust the company (that you don’t trust) to not hobble what you’ve built once you need it to work.

Joborg

@mastodonmigration @joeinwynnewood @jubei @cellularmoose @cwebber Not quite vapour: the protocol does the things that it claims and the core Bsky implementation is there as a reference and proof of concept.

I don't think we'd want more big Bsky nodes, apart from a second big relay, but rather large numbers of smallish ones.

Then, getting Bsky to support migration is touchy and would require there to already exist well-working "servers" to migrate to and where the new migration protocol works.

Grassroots Joe

@joborg @jubei @cellularmoose @mastodonmigration @cwebber

Re: better protocol

I don't know that this is true. There's nothing Bluesky has done that a few million dollars dropped in Eugene Rochko's lap couldn't get done for Mastodon.

The primary difference is Eugene won't give up some measure of control to VCs in return for their money.

The money being raised by this effort very arguably would be much better spent on enhancing the Fediverse.

Joborg

@marcusb @mastodonmigration @joeinwynnewood @jubei @cellularmoose @cwebber Yep, a surprise closing of any emergency exit that is built would be really bad. But the same goes for mastodon.social or other big fedi servers.

Joborg

@joeinwynnewood @jubei @cellularmoose @mastodonmigration @cwebber Scaling by 1M users per day in November was pretty convincing to me, and it's not at all obvious that AP/Mastodon has that potential without an incompatible do-over.

I agree that there is an argument for funding Fedi improvements (which I do), but improving future conditions for 25M Bsky users (and other AT applications to come) is not negligible.