I would like to give an update on "federation" on Bluesky.
-
@bhasic @benroyce @swetland @mcc @aeva Alt text -
"One of the saddest lessons of history is this: If we've been bamboozled long enough, we tend to reject any evidence of the bamboozle. We're no longer interested in finding out the truth. The bamboozle has captured us. It's simply too painful to acknowledge, even to ourselves, that we've been taken. Once you give a charlatan power over you, you almost never get it back."
- Carl Sagan, The Demon-Haunted World -
@mcc @cthos @aeva I feel like fedi's cost/scaling problem mostly centers around issues with the implementation and deployment of Mastodon (the most popular server), with some features of the protocol being not entirely optimal, compared to the fundamental design of atp being hostile to lightweight independent instances.
Somebody could build a "better Mastodon" and instantly give people a lower resource / less complicated option for small or self-hosted yet fully interoperable fedi servers.
@swetland
That already exists in the form of microblog.pub, gotosocial, snac2 at the very least, and yes they all take very few resources compared to BigM. Even pleroma/akkoma is better on this front.
@mcc @cthos @aeva -
@mcc @cthos @aeva I feel like fedi's cost/scaling problem mostly centers around issues with the implementation and deployment of Mastodon (the most popular server), with some features of the protocol being not entirely optimal, compared to the fundamental design of atp being hostile to lightweight independent instances.
Somebody could build a "better Mastodon" and instantly give people a lower resource / less complicated option for small or self-hosted yet fully interoperable fedi servers.
@swetland @mcc @cthos @aeva Mostly true, but Pleroma and Misskey have most of the same features, and are simpler to install, yet haven't supplanted Mastodon.
I think the true killer feature of a "better Mastodon" would have to be in-place migration: allow upgrading an existing Mastodon server by reading its entire database, and provide a UI similar enough to Mastodon that it's a seamless transition for users.
-
@mat@friendica.exon.name I am not opposed to a category discussing alternatives protocols.
-
@mcc @lrhodes @esoteric_programmer @alter_kaker
Feel like the best 'simple' explanation for AT Protocol that makes some sense of the different components is this document from Kuba @mackuba
Introduction to AT Protocol
Walkthrough of the various parts and concepts in Bluesky's AT Protocol (ATProto), the types of servers involved and how it all fits together
(mackuba.eu)
Kuba is also a great resource here on the Fedi for all things AT Protocol. He may be able to clear up any misunderstandings.
@mastodonmigration @mcc @lrhodes @esoteric_programmer @alter_kaker BGS, Big Graph Server, that was the old name, got renamed to relay at some point.
And last November they made a switch to "non archival relays" which now only pass through, but don't store complete repos for everyone.
-
@mat @mcc @alter_kaker I can recommend joining ATProto Touchers discord (was renamed a few months ago)
-
@mat @mcc @alter_kaker I also use discord for Mastodon and Fedify development, but don't feel obliged to join, I'm just saying it's a good server
I currently work across both protocols for various things, learning how they do things that might be relevant to us
-
@lrhodes @mackuba @mastodonmigration @mcc @esoteric_programmer @alter_kaker
yes and no:
the switch was total in the sense that you can run the full network without any archiving function. zeppelin is a demonstration of that, that you can run a full network without any archiving functionality of the relay
in practice the relay that bluesky operates is still on the old version, so their relay still has archiving functionality present. dont think its actually used for anything tho
-
@lrhodes @mackuba @mastodonmigration @mcc @esoteric_programmer @alter_kaker
yes and no:
the switch was total in the sense that you can run the full network without any archiving function. zeppelin is a demonstration of that, that you can run a full network without any archiving functionality of the relay
in practice the relay that bluesky operates is still on the old version, so their relay still has archiving functionality present. dont think its actually used for anything tho
@laurenshof @lrhodes @mastodonmigration @mcc @esoteric_programmer @alter_kaker AFAIK they only run a non-archival relay rn, I don't think anyone runs an archival one at the moment
-
@squinky @mcc so much this
i will never forget how back in 2023 when i was just getting into more leftist thinking and i went to ask why capitalism is bad on fedi since i saw a lot of people shitting on it
and to say i had stones thrown at me would be an understatement
barely anyone was answering my question and the rest resorted to ad hominem attacks. ironically all of this had the opposite of the intended effect and only pushed me away from the ideology they were crucifying me for not blindly adopting
its such a shame that 2 years later this problem hasnt gone away -
@mcc I’m glad to see more discussion around this or any discussion at all because previously when black Fedi users have been outspoken about harassment on this platform and the fact that white queer folks are involved in it (because they are over represented here), they have faced denial, accusations, etc.
-
So what I understand from this is that it would be relatively possible for BlackSky or NorthSky to add ActivityPub as a secondary protocol in ways that would likely be unfeasible to do the other way (ActivityPub instance using ATProto as a secondary protocol)
-
...this is just a surface level commentary on what happened, but if you were BIPOC or followed prominent BIPOC folk here you could eventually see a pattern in the interactions.
I know Bluesky is definitely not a panacea in this regard either (otherwise why the tremendous effort out into blacksky?), but they did address the wants and needs of nonwhite people better at least initially, at the app level.
Anyways what I am seeing is that both ecosystems here are making similar mistakes and facing similar challenges, and it is to do with some aspects of insufficient diversity both technically and socially.
@msh @ikuturso @mcc @swetland @gbargoud arguably bluesky could address those needs better because they maintain control over the full vertical and can apply advisory policies that don't have any real security backing. you can use a different appview or just browse the data directly and *not* apply blocks, gates, etc... this horrifies most people who learn about it and horrifies them further that there's nothing they can do about it. all the policies only apply if you're using the official apps.
-
@msh @ikuturso @mcc @swetland @gbargoud arguably bluesky could address those needs better because they maintain control over the full vertical and can apply advisory policies that don't have any real security backing. you can use a different appview or just browse the data directly and *not* apply blocks, gates, etc... this horrifies most people who learn about it and horrifies them further that there's nothing they can do about it. all the policies only apply if you're using the official apps.
@msh @ikuturso @mcc @swetland @gbargoud however, with that said, the vast majority of people are using the official apps. so the blocks and gates "work" in the sense that the vast majority of people are subject to them. but it remains trivially easy to just *not* respect those blocks and gates, because all the data is forever public
-
-
@mcc @lrhodes @esoteric_programmer @alter_kaker I think they just straight up renamed the "Big Graph Server" or BGS to Relay.
They announced it in Nov 2023 here: https://github.com/bluesky-social/atproto/discussions/1847
@ikuturso @lrhodes @esoteric_programmer @alter_kaker yeah, that thing
-
@esoteric_programmer @alter_kaker @mcc as I understand it the cost of a relay has gone down because you can run it without keeping full account of the message history for all time and instead restrict what you have to a time-window.
Raises the question of whether that's good enough if we want real alternatives to the official company infra though.
@ikuturso @esoteric_programmer @alter_kaker @mcc that $34/month is enough for 2 hours per the article
-
@mcc what exactly is the benefit of migrating to a non bluesky PDS? I understand being on an entirely different vertical stack like Blacksky or Northsky but what does being on a PDS give you? Aren't you still (almost) entirely at bluesky's mercy?
@eniko the advantage is the data is yours and you can recover it later if Bluesky goes berserk. This is a mild advantage but in keeping with the mild advantages operating a PDS is also a mild imposition. It's practically a git repo with a weird frontend. I'm not saying self host but if people are starting to host alternate bluesky bits, this is the most likely thing for someone to alternate host (because it is the least demanding thing) so it's the most likely option to appear.
-
@mat @lrhodes @mcc @alter_kaker @esoteric_programmer the pds/storage can change because the identity is a separate layer. the authority is a DID (most likely did:plc but did:web is also supported). the DID document points to your current PDS. this allows the pds to change as long as the did stays the same.
example: you are did:plc:whatever, and your pds is shiitake.example, but you migrate your data from shiitake to puffball. the did document updates the service pointer: https://web.plc.directory/did/did:plc:ewvi7nxzyoun6zhxrhs64oiz
-
@mat @lrhodes @mcc @alter_kaker @esoteric_programmer the pds/storage can change because the identity is a separate layer. the authority is a DID (most likely did:plc but did:web is also supported). the DID document points to your current PDS. this allows the pds to change as long as the did stays the same.
example: you are did:plc:whatever, and your pds is shiitake.example, but you migrate your data from shiitake to puffball. the did document updates the service pointer: https://web.plc.directory/did/did:plc:ewvi7nxzyoun6zhxrhs64oiz
@mat @lrhodes @mcc @alter_kaker @esoteric_programmer it's kind of like updating dns records. the did:plc stuff is fully in control of bluesky pbllc of course, so it's equivalent to everyone having an id of https :// plc.directory / whatever which is itself equivalent to serving http redirects.
