Sometimes, I find myself feeling burnt out about the #Fediverse.
-
@deadsuperhero I'm looking for answers why a new protocol was created when we already had the Activity Pub protocol.
I got some interesting answers from the creator of this protocol, I just shared some of them earlier today.
If you have some thoughts on technical problems of the Activity Pub, it would be nice to read them. Why a band-aid? May you elaborate about it?
Here are some links shared with me
http://evanp.me/2023/10/06/activitypu
Evan Prodromou (@[email protected])
'Prodromou, however, has strong words for any organization looking to enter social media with a new decentralized social media protocol. “I’m not interested in any protocol besides ActivityPub,” he says. “Anyone working on brand new protocols in 2023 should stop immediately. They are going to do more harm than good."' This may be surprising for people who have known me for a long time. I've generally been supportive of trying new protocols and tools.
Prodromou.pub (prodromou.pub)
-
Sean Tilleyreplied to Charles U. Farley last edited by
@freakazoid I’m sorry, was this supposed to mean something? I’ve been in this space for 15 fucking years, and have been documenting its growth and evolution for a really long time.
I’m sorry that I’m sometimes critical and express concerns about the future of this space, and don’t feel like being a constant cheerleader that only promotes what you want to hear about, when it agrees with your worldview?
To me, this network has always been about people, along with how the technology could enable people. I believe it has the potential to be the future of the Web. There are a lot of pain points and headaches though, from funding to labor to burnout to toxicity. In some ways, other protocols are doing a better job than AP. I’ve always maintained that we should try to adopt the best ideas for our own needs.
That whole “you just don’t know what the Fediverse is about” is a lazy insult from someone that I cannot recall doing a fucking thing for the network, aside from armchair posturing and shit-talk.
-
Sean Tilleyreplied to Charles U. Farley last edited by
@freakazoid This also speaks to how little you’ve actually done in terms of research. ATProto is not Jack Dorsey. Nostr is not Jack Dorsey.
Aside from helping Bluesky in its foundation and securing funding, dude has been practically absent, to the point that he terminated his presence on Bluesky, eventually left the board over fundamental disagreements about the vision of the project, and went over to Nostr.
Yeah, he’s building his new company on Nostr, and funding a bunch of efforts through a foundation. Yeah, it could have some effect on how that ecosystem takes shape. But he didn’t build the protocol.
And yeah, there is a certain degree of team loyalty between all three groups, where each one feverishly believes their solution is the only one, everyone else be damned. But, they’re all attacking problem areas in domains they believe to be important, and these explorations are both novel and fruitful.
-
@everton137 @deadsuperhero why is that even a question? ActivityPub isn’t the end all be all. Do you know how many protocols we have literally due to ideologies or something as trivial as missing a feature or two? What about coding languages that don’t deviate much from one another? It’s weird you and others want a lack of innovation and creativity. Screw having options
-
@freakazoid I’m honestly embarrassed there’s people like you on the fedi and I’m glad there’s spaces where people that don’t think like you are far as hell away from you and anyone like you
-
@everton137 I’ll try to address your points one at a time.
A good starting point for reading about the pain of implementing ActivityPub are these threads by @hrefna, who has been doing deep explanations of why it’s so painful:
Hrefna (DHC) (@[email protected])
I'd like everyone who is working with AP or who has worked with AP to go look at this resource: https://atproto.com/guides/applications Read it, beginning to end if you can. Then come back and look at the following: https://socialhub.activitypub.rocks/pub/guide-for-new-activitypub-implementers and https://jenniferplusplus.com/no-universal-translators/ and https://web.archive.org/web/20240224195830/https://gopiandcode.uk/logs/log-writing-activitypub.html
Hachyderm.io (hachyderm.io)
Hrefna (DHC) (@[email protected])
Attached: 1 image The average #ActivityPub interaction. You know mansplainer, you weren't talking to me, but that's exactly what I've considered doing entirely, and lean more toward that direction every step of the way. Also I have written one, or rather several, but thanks for playing along.
Hachyderm.io (hachyderm.io)
Hrefna (DHC) (@[email protected])
This second point here is actually key. When you find something easy that someone else finds hard, your instinct needs to be not "what are they doing wrong" nor "what am I doing right?" but rather "What are the ways that the problems we are solving are different." Sometimes it is because one of us is solving the wrong problem which easier or harder, but often it is because our perceptions of the problem are _different_ and those differences are _telling_. https://hachyderm.io/@hrefna/113150086741090819
Hachyderm.io (hachyderm.io)
Hrefna (DHC) (@[email protected])
This has really proven to be one of my biggest frustrations with ActivityPub. Nothing you do with it works without breaching multiple abstraction layers. https://hachyderm.io/@jenniferplusplus/113026563544821873
Hachyderm.io (hachyderm.io)
Hrefna (DHC) (@[email protected])
This is one of my other recurring frustrations every time I dip back into ActivityPub, tying into the lack of well defined semantics. People will readily say "oh you can just do this!
" That's a research protocol. It doesn't give me guidance on building a usable system. There's no built-in mechanism for feature discovery or protocol handshakes, so there's no way for a server to know how to interact with another system here. https://hachyderm.io/@hrefna/113051946822578560 Hachyderm.io (hachyderm.io)
Hrefna (DHC) (@[email protected])
Something I want to note here is that AP could work with separating identity from hosting, but the major providers who use AP* do not take advantage of the features for that, and doing so would go against how the guidance docs for security are written. The basal assumptions made by mastodon et al are largely incompatible with the more straightforward approaches here, but there's nothing that requires it. (*I'm talking mastodon-derived protocols, not hubzilla etc). https://martianbase.net/@mackuba/113120949169540812
Hachyderm.io (hachyderm.io)
Regarding why Bluesky made their own protocol, rather than building on AP, it boils down to wanting to implement their own approaches from first principles. ATProto ties heavily back to the identity layer, and you could make the argument that its design is heavily influenced by Secure Scuttlebutt, Solid, and Nostr.
There is nothing wrong with wanting to roll your own thing. Paul Frazee, the protocol dev, was heavily involved with Secure Scuttlebutt and Beaker Browser, so of course his understanding of distributed systems informed some of the decisions he’s made with ATProto today.
-
@everton137 To broadly summarize some of the pain points, a lot of it comes down to how the standard is put together, as well as how it is used.
Most ActivityPub implementations initially looked to Mastodon for examples on how to implement the protocol themselves. However, Mastodon took a number of liberties as First Implementor to declare how things should work, such as: using Webfinger for pulling in remote content, using a field for content warnings that other systems were using for summaries, and not implementing the Client-to-Server portion of the spec.
Part of the problem is that some parts of the AP spec are under-defined, to the point that it’s not totally clear how things are supposed to work by default, as presented in the spec. Most ActivityPub implementations are a revision of what Mastodon first did, and most of these implementations were built against Mastodon for compatibility, rather than a standards-compliant target where everybody could be compatible with everybody else.
This has actually caused a lot of long-term growing pains, as developers have to play whack-a-mole in trying to be compatible with one another. I’ve heard that some aspects of the protocol cannot be easily abstracted into a library in a matter that’s both comprehensive and standards-comformant, and there are a lot of open questions on what a “proper” standard implementation should be for any given ActivityStreams Verb. Most of that stuff is loosely defined, at best.
-
@damon @deadsuperhero I don't want a lack of innovation. One of the main co-creators of Activity Pub just said that the existence of AT Protocol pushed Activity Pub to improve some aspects (second screenshot). That's positive, in my opinion.
There's a technical aspect, I still have to study it. but we cannot forget several cases where the for profit goal smashed or swallowed good initiatives with less resources.
I still have to do my homework and read several references I saved because of my question on the reason behind the creation of the AT Protocol.
-
Sam Sethi :pc2red: ⁂replied to Sean Tilley last edited by
1/3 @deadsuperhero As someone new to developing ActivityPub apps the poor documentation was a big turn off. @fedify have done a great job in making it easier to build new apps and their documentation is much better. The new book by @evan is also a great read for new AP developers.
However given that the Activity Vocab was defined way back in 2017 by the W3C and Mastodon the biggest AP client still doesn't support all the verbs is a disappointment. e.g summary, listen etc.
-
Sam Sethi :pc2red: ⁂replied to Sam Sethi :pc2red: ⁂ last edited by
2/3 - @deadsuperhero @fedify @evan I still think ActivityPub has a strong future. Sometimes it takes a decade before a technology finds its mojo. In recent months I have observed Flipboard, Threads, numerous new AP clients and my own app TrueFans all adopt ActivityPub. @fedidb has shown a significant growth on AP user numbers and @fediforum was great.
Also Apple, Linkedin, YouTube, Twitter and others are scraping our data to train their AI's. The Fediverse may be the only safe haven.
-
Sam Sethi :pc2red: ⁂replied to Sam Sethi :pc2red: ⁂ last edited by
3/3 - @deadsuperhero @fedify @evan @fedidb @fediforum Sean I wouldn't get down. In the Podcasting 2.0 community we have similar challenges and problems. The big players like Apple and Spotify are not supporting the new RSS tags/standards. The collective open nature of developing new RSS tags means it moves slower than proprietary platforms like Spotify or YouTube.
Sean I don't have an answer to your questions but I do see momentum in both ActivityPub and Podcasting 2.0
-
Sean Tilleyreplied to Sam Sethi :pc2red: ⁂ last edited by
@samsethi Honestly, I feel the same way most of the time. There’s a lot of exciting stuff going on, and I do still believe ActivityPub has time to evolve and mature. It’s just that, having watched the space for so long, some days feel like everything slows down and speeds up at random.
I think we still have a bright future ahead of us. The difficult part is figuring out the growing pains.
-
@deadsuperhero In spite of how many years it has been, we are all still in the very early phases ("tech enthusiast") of the tech adoption cycle. What we see in the #Fediverse right now is straight out of the textbook: lots of tinkering, bandaids, very geek heavy, no money.
But we are about to move out of this phase into the "visionary adopters" of which we now have a few. That will change the ecosystem quite a bit, but towards sustainability and higher quality. -
I suppose I should have known what a "visionary adopter" is but nothing came to mind.
I'll have to look it up.
-
I'm referring to the version of tech adoption cycles commonly associated with McKenna and Geoffrey Moore's books Crossing the Chasm etc.
Also, this is a multi-sided "market", so the model applies differently on the user side and the developer/vendor side. It's possible we are further along on the user side than the developer/vendor side.
-
Oh come on. @[email protected] has been part of the fediverse for years and has as deep an understanding as anybody I talk to. OK, you disagree with him on this -- but he's far from the only long-term fediverse person who's highlighted the dismissiveness (and often ignorance) many people here have when they talk about Bluesky.
@[email protected] @[email protected] -
smallcircles (Humanity Now 🕊)replied to Sean Tilley last edited by
That's a great summary and glad you kept a list of @hrefna posts, as I didn't have that ready.
Before Bluesky really started they created a document comparing all existing decentralized protocols having any amount of uptake. That doc lived at Gitlab, but looks to be gone + not archived either.
I did find a snippet by @cwebber - co-author of ActivityPub spec - on AP with OCaps:
Christine is leading great innovation at https://spritely.institute
-
smallcircles (Humanity Now 🕊)replied to smallcircles (Humanity Now 🕊) last edited by
Oh, btw, @hrefna has brought us many more insightful discussion threads than those referenced in that list.
Unfortunately, unless one is a heck of a dilligent note taker - a fedi scribe - everything that's discussed in fedi timeline history is bound to be lost forever.
Btw, there's a current topic at SocialHub collecting a wishlist for future version of AP protocol: https://socialhub.activitypub.rocks/t/desired-changes-for-a-future-revision-of-activitypub-and-activitystreams/4534