Conversation

Notices

1. Embed this notice
   Sergio (galactus@col.social)'s status on Monday, 11-Nov-2024 04:36:31 JST Sergio

   • Evan Prodromou

   I just finished a first read of @evan ‘s ActivityPub book from O’Reilly. The definitive guide for anyone who wants to write applications for the fediverse, imho. The detail level is spot on.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Monday, 11-Nov-2024 04:36:53 JST Evan Prodromou

     in reply to

     @galactus Thank you! I’m glad you enjoyed it.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Monday, 11-Nov-2024 05:21:55 JST Evan Prodromou

     in reply to

     • Mushi

     @Mushi @galactus

     For me, the fact that AT is a proprietary protocol controlled by a single VC-funded startup, while ActivityPub was created and is maintained as an open standard with patent protection by a recognized standards organization, is what’s important.

   • Embed this notice
     Mushi (mushi@col.social)'s status on Monday, 11-Nov-2024 05:21:56 JST Mushi

     in reply to

     • Evan Prodromou

     @galactus @evan question: yesterday I was listening to the better offline interview with Mike Masnick where they touch on the AT protocolo I’m sure the comparative has been done before, but apart from the private vs open source nature of things, what are the main differences between the two

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Monday, 11-Nov-2024 05:22:42 JST Evan Prodromou

     in reply to

     • Mushi

     @Mushi @galactus

     I’d add that there are over 100 implementations of ActivityPub, dozens of clients, and tens of thousands of servers. The number of different organizations involved is surprising.

     Protocols are a social phenomenon, not a technical one. And on the social front, ActivityPub is a clear choice.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Monday, 11-Nov-2024 07:23:36 JST Evan Prodromou

     in reply to

     • Mushi

     @Mushi @galactus technically I don't know a lot about AT. I try not to use it or to read its docs since there's no patent protection.

     That said, I'm pretty sure it uses DIDs for identifiers, and that all content is public.

     Jan Wildeboer 😷:krulorange:, Tim Chambers and Michał "rysiek" Woźniak · 🇺🇦 repeated this.
   • Embed this notice
     Mushi (mushi@col.social)'s status on Monday, 11-Nov-2024 07:23:37 JST Mushi

     in reply to

     • Evan Prodromou

     @galactus @evan I completely agree, for what it’s worth.

   • Embed this notice
     Sergio (galactus@col.social)'s status on Monday, 11-Nov-2024 08:16:53 JST Sergio

     in reply to

     • Evan Prodromou
     • Mushi

     @evan @Mushi it seems to me ATProto’s descentralization is mostly theoretical (it is not realistic for a person or a small organization to run the main aggregator service and indeed no one but bluesky does) whereas for ActivityPub is not only possible but a reality.

     If Bluesky were to dissapear today, all the self-hosted PDS instances would be rendered useless immediately. If Threads or the biggest mastodon instances closed today, the fediverse would still be alive and kicking.

   • Embed this notice
     ch0ccyra1n :she_her::neocat_floof_cute: (ch0ccyra1n@emeraldsocial.org)'s status on Monday, 11-Nov-2024 08:24:15 JST ch0ccyra1n :she_her::neocat_floof_cute:

     in reply to

     • Evan Prodromou
     • Mushi

     @evan @Mushi @galactus I do think that it's worth incorporating DIDs into ActivityPub implementations (as I understand it, the current webfinger-based usernames fall outside the spec, please correct me if I'm wrong). Otherwise I couldn't care less about AT Protocol. I just want to be able to talk to my friends regardless of where they are.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Monday, 11-Nov-2024 08:24:15 JST Evan Prodromou

     in reply to

     • ch0ccyra1n :she_her::neocat_floof_cute:
     • Mushi

     @ch0ccyra1n @Mushi @galactus I'd definitely recommend using the Bridgy Fed bridge. It's really smooth.

     https://bsky.brid.gy/

   • Embed this notice
     mcc (mcc@mastodon.social)'s status on Monday, 11-Nov-2024 10:07:09 JST mcc

     in reply to

     • Michał "rysiek" Woźniak · 🇺🇦
     • viq
     • Evan Prodromou
     • Mushi

     @viq @galactus @rysiek @evan @Mushi (Important nuance, possibly not super relevant to the conversation preceding: you can always *add* stuff. The version mismatch rules viq mentions only refer to *changing* stuff. So you could layer a second "application" with its own orthogonal data fields on top of AT, it's just if your implementation of the bluesky distributed application differs even a little from the Bluesky Social mothership you can get unpersoned. I am not impressed.)

   • Embed this notice
     Michał "rysiek" Woźniak · 🇺🇦 (rysiek@mstdn.social)'s status on Monday, 11-Nov-2024 10:07:09 JST Michał "rysiek" Woźniak · 🇺🇦

     in reply to

     • mcc
     • viq
     • Evan Prodromou
     • Mushi

     @mcc so, BlueSky is *still* cosplaying decentralization.

     @viq @galactus @evan @Mushi

   • Embed this notice
     viq (viq@social.hackerspace.pl)'s status on Monday, 11-Nov-2024 10:07:11 JST viq

     in reply to

     • mcc
     • Michał "rysiek" Woźniak · 🇺🇦
     • Evan Prodromou
     • Mushi

     @galactus @rysiek @evan @Mushi
     IIRC @mcc tried to run PDS, and you have to have *exactly* same version as main bluesky. You try to add your own modifications, it won't work. You fail to update at the same time they do, it won't work.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 06:16:59 JST Evan Prodromou

     in reply to

     • Pierre
     • Mushi

     @okpierre @Mushi @galactus did you find the patent license that lets other codebases use the techniques patent-free? Not just their own Apache-licensed code?

   • Embed this notice
     Pierre (okpierre@mastodon.social)'s status on Tuesday, 12-Nov-2024 06:17:00 JST Pierre

     in reply to

     • Evan Prodromou
     • Mushi

     @evan @Mushi @galactus

     Please share how atproto is a proprietary protocol

     According to the license file atproto is Dual MIT/Apache-2.0 License which you can view here: https://github.com/bluesky-social/atproto/blob/main/LICENSE.txt

     Also included a screenshot

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 06:17:42 JST Evan Prodromou

     in reply to

     • Pierre
     • Mushi

     @galactus @okpierre @Mushi thanks. This is correct.

   • Embed this notice
     Sergio (galactus@col.social)'s status on Tuesday, 12-Nov-2024 06:17:43 JST Sergio

     in reply to

     • Pierre
     • Evan Prodromou
     • Mushi

     @okpierre @evan @Mushi the license you pointed at concerns the distribution if the at protocol reference implementation. The propietary character of the protocol itself is a different matter.

   • Embed this notice
     Sergio (galactus@col.social)'s status on Tuesday, 12-Nov-2024 06:17:44 JST Sergio

     in reply to

     • Pierre
     • Evan Prodromou
     • Mushi

     @okpierre @evan @Mushi but that's not how openness works for network protocols. Governance is critically important.

   • Embed this notice
     Pierre (okpierre@mastodon.social)'s status on Tuesday, 12-Nov-2024 06:17:44 JST Pierre

     in reply to

     • Evan Prodromou
     • Mushi

     @galactus @evan @Mushi

     I'm trying to understand how atproto is proprietary. Are you saying the license is meaningless?

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 06:25:58 JST Evan Prodromou

     in reply to

     • Pierre
     • Mushi

     @okpierre @Mushi @galactus I've asked @bnewbold.net and @mmasnick.bsky.social about making a patent license for ATProto.

     One way they could do it is to publish the protocol as a Report of a W3C community group, like the SocialCG or even a new community group, which would give it the patent licenses that open standards require.

     https://www.w3.org/community/about/process/final/

     This would make protocol experimentation much more reasonable.

     Only the Bluesky team can do this.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 06:27:06 JST Evan Prodromou

     in reply to

     • Pierre
     • Mushi

     @okpierre @Mushi @galactus @bnewbold.net @mmasnick.bsky.social I know that "open source" feels like it should be enough to be an "open standard", but unfortunately not. One of the hard things the distributed social network community have learned over the years is that the standards process matters.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 09:01:08 JST Evan Prodromou

     in reply to

     • Pierre
     • Doctor Popular
     • Mushi

     @okpierre @docpop @Mushi @galactus @bnewbold.net @mmasnick.bsky.social

     https://en.wiktionary.org/wiki/proprietary

     "4. Nonstandard and controlled by one particular organization."

   • Embed this notice
     Doctor Popular (docpop@mastodon.social)'s status on Tuesday, 12-Nov-2024 09:01:09 JST Doctor Popular

     in reply to

     • Pierre
     • Evan Prodromou
     • Mushi

     @evan @okpierre @Mushi @galactus @bnewbold.net @mmasnick.bsky.social fascinating convo. It's interesting to say "open source" is not the same as "open standard" and that both are needed for true decentralization. That makes sense, but I'm not understanding what BSKY needs to change in order to be open standard.

     Are you saying the code is open source, but still requires a proprietary piece (like a company-controlled server)?

   • Embed this notice
     Pierre (okpierre@mastodon.social)'s status on Tuesday, 12-Nov-2024 09:01:09 JST Pierre

     in reply to

     • Doctor Popular
     • Evan Prodromou
     • Mushi

     @docpop @evan @Mushi @galactus @bnewbold.net @mmasnick.bsky.social

     I'm glad I asked. This is more about the additional patent license that allows others to use the techniques patent free

     It's misleading to say atproto is proprietary. I do wonder if Nostr can also be mislabeled the same way

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Tuesday, 12-Nov-2024 10:16:13 JST Evan Prodromou

     in reply to

     • Pierre
     • Doctor Popular
     • Mushi

     @docpop @okpierre @Mushi @galactus @bnewbold.net @mmasnick.bsky.social there are two main things.

     First is governance. Turning over the protocol to a multi-stakeholder standards organization like the IETF or W3C. As long as one company controls it, it's not open.

     Second, like I said, is eliminating the other IP issues. Patents, trademarks, copyright. Those are usually handled by the IP agreements of standards organizations, or they also can be done unilaterally.

   • Embed this notice
     Evan Prodromou (evan@cosocial.ca)'s status on Wednesday, 13-Nov-2024 02:49:26 JST Evan Prodromou

     in reply to

     • Pierre
     • Doctor Popular
     • Mushi

     @okpierre @docpop @Mushi @galactus @bnewbold.net @mmasnick.bsky.social I agree that if your only understanding of the word "proprietary" is "not available under an OSI-approved license", it can be hard to comprehend the phrase "proprietary protocol" or "proprietary extension". The term is commonly used in standards circles, though.