SocialWebFoundation - what do people think?

I wouldn’t expect that. I’m not the only one that doesn’t like the name, not because it isn’t fair but because it’s not accurate. It’s obviously up to the SWF to consider that feedback or not.

I don’t know what “modulate your tone” means in this context. It’s fine to stop directly engaging on this topic here if you’d like. If giving you space means that I can’t express my opinions about SWF’s framing and objectives, in general, then I can’t make promises about that.

Believe it or not, I wish you the best with Foundation, even if I don’t agree with several aspects of it (I do support other aspects of it). You have the right to fight for your vision of the “Fediverse” and I have the same.

4 Likes

People can choose which protocols they want to use based on the people that are on those protocols, and what those protocols allow them to do. Back in the day when there were a ton of IM platforms, Pidgin allowed people to connect to multiple of them. This mostly works because those messaging platforms all share a common denominator of functionality, in 1:1 text chats. But there isn’t a common denominator of functionality in the fediverse except for POSTing a Follow to someone’s inbox. After that point, you have no idea what they’re going to send you other than “it’s an AS2 Activity”. You end up installing a dozen different apps, not to cover protocols, but to cover use cases and media types. I don’t think “multiple apps” is a good argument for or against anything. If someone built a super-app that did everything, then what?

Are you saying that if something uses UDP or QUIC, it’s not “internet”? I think that is not only untrue, but also incredibly myopic. “The Internet” refers to the overall network of computers, regardless of which protocol they use, as long as they can connect to each other. Generally, if you’re looking for a unifying protocol for “The Internet”, that wouldn’t be TCP, it would be the Internet Protocol (IP). Of course, other networks of interconnected devices could still exist, but for the most part there isn’t really a reason these days to use anything other than IP.

As for “web”, anything with links can be considered a “web”. Gemini is a web, for example. It’s just not part of “The (World Wide) Web”, which as you say uses HTTP(S) and DNS. If it uses HTML, I’d call it the “hypertext web”. If it uses something like RDF to encode semantic links between machine-readable data, then I’d call it the “semantic web” or “linked data web” . And so for a “social web”, I would define it as a web that lets us be social. For “The Social Web”, then this implies it is part of “The Web”. Which to me means publishing Web resources. Not so much building a new messaging protocol over HTTP (which is perhaps not well-suited for this role). I don’t think something is meaningfully “Web” if it doesn’t let you publish resources to the Web. The Web isn’t defined by the protocols we use, it’s defined by the ability to follow a link. For most people, this means things like Firefox or Chromium or Safari, which are “Web user-agents” mainly for the “hypertext web”. If you want to browse the “linked data web”, then look to something like https://browser.pub which lets you browse the JSON-LD directly and follow links inside the JSON-LD. But it doesn’t make sense to “browse the Social Web”, because the Social Web ought to be defined by the social links and mechanisms that we have available to us to interact with Web resources. Protocols like WebSub allow us to subscribe to updates about Web resources being published or updated, in much the same way that ActivityPub allows us to subscribe to activities, where those activities are often themselves Web resources on the linked data web.

So if we want to “interop over [a] unifying standard”, then it would probably be helpful to actually have a standard that is more than just “POST a Follow activity to someone’s inbox”. More than just “Likes should end up in the likes collection” or “Announces should end up in the shares collection”.

The “problem” with the SWF is that it skips several steps ahead and equates or equivocates “Social Web” and “fediverse” and “ActivityPub” as synonyms of each other, instead of as three different words that mean three different things. This is made more concerning by the list of partners being limited to a very particular interpretation of the fediverse. The center of gravity in the SWF’s interpretation of the fediverse (or “Social Web” as they call it) is located squarely within the vicinity of Mastodon, Pixelfed, Meta, Automattic, and other parties who seem to want to build platforms. I don’t think this should be the primary approach. Yes, it allows for more interconnection between those platforms… but then you ossify whatever protocol is shared between them, at a time when the protocol needs a lot more formal definition and further development and experimentation. Whether you like it or not, realize it or not, you are prioritizing the needs of your constituents above other needs. And those needs are focused on building platforms. Not on escaping the platform model altogether.

5 Likes

Friendica can be used as a frontend for an existing Bluesky account. This account can either be hosted on a PDS (Personal Data Server) that is provided by Bluesky or a PDS that you run on your own. Also you can use your Friendica handle as a custom hostname for Bluesky. So for example my Bluesky name is heluecht.pirati.ca (this is derived from my Friendica handle heluecht@pirati.ca). Friendica supports that you can use the handle user.domain.tld when your Friendica handle is user@domain.tld. (Some server setup is needed, but this is documented).

Except from direct messages, most other parts of the protocol are supported.

But: you currently need the “develop” branch of Friendica (or you have to apply a small patch to the core), since we hadn’t supported HTTP compression via gzip before.

2 Likes

The question is, whether you define “Fediverse” as “ActivityPub only” or as a network of distributed servers that communicate via a documented protocol. With Bluesky you can setup your own PDS (Personal Data Server) and you can use your own hostname as a handle to participate there. This - as far as I see it - is an indicator for a network of distributed servers.

3 Likes

I’d say that AT’s clearly a decentralized protocol, but not not in the instance-oriented way that current AP implementations are. Opinions differ on whether or not it should be considered a Fediverse protocol. But I know what Evan thinks on that front so that’s not what I was asking about.

And yeah, different people have different definitions of “Fediverse”, and its meaning changes over time; Definitions of “the Fediverse” has a bunch of examples.

Today, some (including Evan and SWF) define the Fediverse in terms of ActivityPub; others see it as multi-protocol, or prefer less protocol-centric definitions. Where it gets even trickier is that for people who define it in terms of ActivityPub, there’s a difference of opinion about whether networks that connect via bridges or multi-protocol Fediverse platforms like are included, and if so whether it matters whether the bridge is written by the same organization that implements the non-AP network.

So that’s what I’m trying to understand with my question to @eprodrom. To me, Bluesky and Threads are fairly analogous – they have their own tech stack internally and connect to AP via bridges. Does that mean they’re Fediverse instances? I personally would say yes, but others disagree for a variety of reasons. In his previous reply to me Evan said he sees Bluesky accounts as being on the Fediverse, so I think he sees this way I do, but am not completely sure so I wanted to double-check. It’s not a matter of “right” or “wrong”, there are a lot of different intellectually-consistent definitions, I just want to make sure I understand how he’s seeing it.

This is almost verbatim what I would say, yes.

I hope at some point Blue Sky makes the bridge official and gives Ryan Barrett the appreciation he deserves.

2 Likes

… until that one bridging server goes down, and then they can’t. And only after a fiddly permission dance, thanks to the Gated Community brigade.

The fediverse.party team talked this over recently, and agreed that we still don’t include bridged accounts in our definition of “the fediverse”. For the same reason nobody ever thought of Titter accounts as part of the fediverse. Despite the the fact there were once far more bridges to that (crossposting servers etc) than there are to BS.

You just presented the strongest argument I’ve seen for including BS accounts (do you include Nostr acounts for the same reason?). I’ll reopen our issue discussion and link to your comments, to make sure we consider it. But I’m still not convinced.

To be clear, I very much want BS and Nostr accounts to be part of the fediverse, whether that’s using AP or a successor. Just as I very much wanted the OStatus fediverse, the Diaspora Federation and the Zot-iverse to be unified, which AP has mostly achieved.

This was due in no small part to your efforts, both technical and organisational. Which is why I said in my unromantic history of AP post …

Also why my opinion of the SWF went from sceptical to optimistic as soon as I saw you were involved.

No analogy bears close examination, but see above. The issue is simply this; does it federate directly with (at least some fediverse servers), or does it depend on indirect bridging that might not always be there?

I’m itching to question some of your factual claims about internet protocols. Because AFAIK everything you mention sits on top of an IP connection, which is what allows it to be part of the rest of the internet. So the equivalent in the fediverse is subnetworks using different clusters of FEPs, on top of a minimal AP connection that puts them all in the same network.

1 Like

I’m not sure if it does or not (links please?), But to me, “social web” is obviously a broader descriptor, that includes any social software built on top of the open web. So it includes the fediverse, the Matrix and the ATmosphere, but not stuff like IRC, XMPP, FreeNet, Scuttlebutt, or Web3 stuff using blockchains and native apps. Not sure about Nostr yet, but I suspect it fits with the latter. In what sense is it part of the web?

Whereas “the fediverse” is a specific network with a history. Because of that history, it is functionally synonymous with ActivityPub for now. But if the participants in that network mostly moved to another protocol, they wouldn’t be synonymous in any way.

The people behind Mastodon and PixelFed have been part of the fediverse longer than AP has existed. Automattic is a newcomer, but they’re a bastion of the open web, who have entered by hiring someone already working on fediverse software, again since before AP existed. Meta is … well … Meta. In no universe would I put those 4 entities in the same list, and I would respectfully suggest that anyone who is trying to equate them has an axe to grind.

2 Likes

Sorry to labour the point but …

That’s not the question. The question is whether you define it as;

a) network of distributed servers that communicate via a documented protocol with the pre-existing network that coined the name “the fediverse” and has an unbroken history of being known by that name (see the Wikipedia article on "fediverse).

b) any network of distributed servers that communicate via a documented protocol.

It seems pretty self-evident to me (and the rest of the fediverse.party team) that the answer is the first one. I have yet to see anyone explain why it would help anyone to expand the definition of “the fediverse” to the second one, when the more generic term “social web” is right there for that purpose. Although as you’ve stated it, b) is broader even than “social web”, because it would include networks that use servers but not web protocols.

It has changed as apps have started interoperating with the network known as “the fediverse” and as that network migrated from OStatus to AP. That doesn’t mean anyone’s definition of the term is as valid as anyone else’s. As you well know, because I’ve explained this to you on the fediverse, and here, multiple times. So I’m curious to know why you are so determined to dilute the meaning of the term, considering that I’ve only noticed you in discussions about fediverse development since Eternal November, and some of us have been working on this for more than a decade (and pioneers like Evan much longer).

2 Likes

Please correct me if I’m wrong, but Z-Wave and LoRaWAN don’t use IP, for example. Power consumption is a consideration in both cases, but Z-Wave is a mesh network that’s not a good fit for IP’s routing design. You can add a Z-Wave or LoRaWAN gateway/bridge to interoperate with IP.

I don’t know what “minimal AP connection” means, but I think it’s going to a challenge to describe an analogy with IP that withstands even superficial scrutiny.

Just like with the internets, different Fedi protocols are used to address different requirements and use cases. They can be and typically are bridged to allow inter-protocol communication. I don’t view the bridges from an ActivityPub-centric bias, but more peer-to-peer. For example, I wouldn’t claim that indieWeb is the true Social Web and ActivityPub is a secondary protocol that just bridges into it using bridgy (which was initially created to support indieWeb, IIRC).

2 Likes

Well, they are in the same list – as partners of the SWF. I’m not equating them or even providing a value judgement here; what I’m pointing out is specifically that partners of the SWF are largely taking a “platform” approach to the fediverse.

3 Likes

Thankfully, the Wikipedia article doesn’t say anything (AFAICT) about the “Fediverse” being a static historical concept that cannot be extended with new protocols in the future.

(This isn’t saying I agree with your other definition either… there are other possibilities than those two.)

4 Likes

Regarding an activitypub.rocks developer portal I answered here.

Other than that I feel - and I think you agree - that what makes the AS/AP Fediverse so unique is that it was entirely spun up by the Commons, including the W3C open standards that are its technological foundation.

I spent significant time and effort telling folks that - with increasing uptake and adoption - the commons stands to lose that unique position. That unless we manage to uphold and guarantee a bottom-up open and decentralized ecosystem, a corporate takeover is imminent. The trend for that is very clear.

As a firm representative of the grassroots movement / the commons, and active years-long community facilitator for SocialHub/FEP, here is where I prefer to spend time (and deliberately not in SocialCG), as a custodian of the 3-stage bottom-up standardization process.

To me a taskforce suddenly popping up was highly surprising, and a form of top-down decision-making: “Let’s take that away from SocialHub” without informing/consulting upfront.

I really, really hope not to wake up one day and see a msg of “The FEP is now at SocialCG and on Github, managed by its own taskforce” without prior notice and discussion with the community.

5 Likes

It is a bit late of me mentioning this now, given this thread is ultra-long and veiring off-topic at many points. Let’s try to keep things a bit focused on SWF / SocialCG / SocialHub, and have discussions about other protocol and types thereof in a different thread.

An interesting thread to discuss might be Poll: Should SocialHub be scoped to AS/AP or Social Web?

From the SWF Launch Announcement (fediverse:creator metadata identifies Evan as the author):

The “social web”, also called the “Fediverse”, is a network of independent social platforms connected with the open standard protocol ActivityPub.
Launch of Social Web Foundation | Social Web Foundation

Compare this to Coates’ description (which I believe is much more reasonable):

The main Fediverse approach is through projects like Mastodon – which are effectively small, local social networks that can be hosted by an individual or company, but whose users can still communicate with — and reference posts and people — using other similar networks. Most of these products are built with at least some reference to the ActivityPub protocol co-written by Evan Prodromou. (emphasis mine)
On the Social Web Foundation… – plasticbag.org

2 Likes

Then this list is not a great example of that. Because the first 2 in your list are fully Free Code software that anyone can run, and hundreds of people do. The second one is a company that maintains a number of fully Free Code packages (including WP) that anyone can run, and thousands of people do. Yes, all the entities behind all 3 run one or more flagship instances, but as I said, they’re also longtime supporters of the fediverse in good standing (1).

In neither of these cases do I see a “platform”, in the obviously negative sense of the word that you’re using (by analogy with what Meta do). Can you explain why you do, and more importantly, what you think it means to be …

Also, I’m still waiting for a link to demonstrate that SWF use “social web” and “fediverse” interchangeably.

EDIT: This was provided by @stevebate . Thanks for this. But that doesn’t really clash with what I say below, especially in the context of the the blog piece you linked by Tom Coates.

My interpretation is that Evan’s definition of “social web” very much includes networks that he doesn’t yet consider part of the fediverse (even if he does include accounts on the bridged services within them). With the aspiration that the SWF can help the various social web networks to unify. In a similar way to how the SWWG helped the OStatus, Diaspora, and Zot networks to unify to form the fediverse that existed when Eternal November began.

If I’m right that this is Evan’s understanding and aspiration, I thoroughly endorse it.

(1) Automattic by extension, as the new employer of Matthias Pfefferle who has been developing fediverse extensions for WP since it was still using OStatus

You may well be right, but the point is that tries to claim that the internet isn’t the network of networks unified by IP networking, and nobody says “internets” (except you, today, but have you ever written that before?).

Exactly what it says on a the tin. Implementing enough of the core AP spec to have some level of interop with other software that does the same. As I’ve said elsewhere, my ideal AP would be as minimal as possible, like IP and XMPP-core, with everything else standardised in FEPs. Including meta-FEPs defining AP+FEP bundles for use cases like the threadiverse, or video federation, or events federation (Mobilizon, Gancio etc)

@trwnh seems to think we should try to fold all of this into AP itself, and I have no idea why they think this is a good idea.

I have no idea how to interpret this because I don’t know what you mean by “Fedi”. If you mean “the fediverse”, no, that’s never been the case. If you mean “federation protocols in general”, then yes, of course. People use email for some things, XMPP or Matrix for others, and AP or AT Proto or Nostr for others.

I’ve talked to the developer of Prosody (also founder of ModernXMPP and Snikket) quite a bit about the pros and cons of bridging. He says it’s a brittle and resource intensive way of doing things, that’s only useful in adversarial interop (eg Mastodon<>Titter crossposters), or as a stopgap while trying to converge on a common protocol. This matches my observations over about 2 decades of using and researching decentralised network software.

Would you claim the opposite, that the IndieWeb is a secondary protocol that just bridges into it using bridgy? The IndieWeb and the fediverse are 2 separate federated social web networks, bridged together. The fact of that bridging doesn’t automagically fold one into the other, in either direction. Adding BlueSky and Nostr doesn’t automagically make them part of the the fediverse, any more than it makes the fediverse part of BlueSky or Nostr.

I’d ask you the same question as Jon. Why do you seem so determined to dilute the well-established meaning of the term “the fediverse”? I don’t remember seeing you in discussions about fediverse development before Eternal November either.

1 Like

Yeah, it’s worth underscoring that the solution to the protocol wars was not “Okay TCP/IP wins the rest of you go home” it was to understand the problem differently (which was controversial in itself in some ways). TCP/IP can be viewed as the “winner”—not inaccurately—but what happened was significantly more complex, happened over the course of decades, and happened not just because it was popular but because there was an entire mental model shift being facilitated at the time.

5 Likes

I like this-- I would also mention that in 2017 the Social Web WG tried to formalize this a bit in this document which people often overlook when trying to keep straight what’s a platform and what’s a protocol and what’s a federation of platforms and/or citystates.

Hey, let’s keep it above the belt here! I didn’t create a socialhub profile or write to the CG list until 2023 even though I was at the ActivityPub conference in Prague in 2019.

I have personally been grappling a lot with different usages of the word “platform” lately, ranging from economics to regulation to software. Basically, any protocol/rule-driven and stable environment where pots find lids and sellers find buyers is a “platform”. It can mean very different things whether you’re talking about a:

1.) low-level technology (“the web platform” === everything built on top of the sum of browser requirements and APIs, i.e. the entire web and half the world economy),
2.) a developer platform (i.e. facebook connect, a platform matchmaking developers and users on commercially-controlled but stable APIs), and
3.) a user platform (commercial social media), where users interact with each other within some kind of perimeter.

ActivityPub is a “social web platform,” to some degree, and I suspect that people who say it is inadequately specified to serve that function are probably conflating level 1 and level 2; to them, I would recommend a slow read through the HTML specs (AP is a dream by comparison!). On Layer 2, the “Mastodon API” (loosely speaking, and including the Mastodon dialect of AP, etc) has supplanted C2S and made monolithic clientic-servers the “de facto standard”-- here is not the place to argue whether or not that’s optimal, but it has definitely happened already, and fuses these two layers into one “de facto platform”. Every server which has exposes a “local timeline” is, literally, also creating its own micro-platform on the third layer, and I don’t think that’s an anti-pattern at all (it’s actually a really neat aspect of Mastodon that I think explains much of its success!). I would even say that the coziest, “small-fedi”, “I won’t federate with servers over 100 users” crowd conceives of federating these micro-platforms without ever creating a governable platform at layers 1 and 2 as the optimal outcome. The useful comparison on Layer 3, however, is not the size of the local timelines, but the perimeter-- how differently each shows its content to unauthenticated viewers, whether cross-platform DMs are allowed, etc etc. Here Threads is operating not much differently than an ultra-cozy 50-user Mastodon with aggro mods (except the ads).

All of this is to say that whatever your vision of the fediverse, you are federating platforms on 1 or more of these 3 layers. The frustrating part is that we keep talking past each other or cudgeling each other to say one layer matters more than the other 2, when in reality all 3 are important and neglecting any of them hurts us all.

I would even argue that the three layers map pretty well to the 3 stages in Arnold’s process. I think normative work really only affects Layer 1, and the SWF seems to be targeting work at that layer (how the social web interacts with the actual web-web, both technologically and commercially). I would prefer Layer 2 diplomacy stay in the CG, where some degree of IP protection is still useful in the long-term but volunteers and amateurs need to be on equal footing as large corporations, and Layer 3 needs to be independent of Layer 2 to work well.

3 Likes

So, here, the clear winner in the Social Web Foundation picture, after two decades of fighting Gaggle and Fakebooz, Twatter and Yourude, is Mata, the continuation of Fakebooz. Nice prospect.

I’m not sure it’s that cut-and-dry just yet, and I’m pretty cynical.

By reference to my mammoth wall-of-text above, I would point out that currently, Threads is only interoperating deeply with two implementations (Mastodon and GoToSocial) of the Mastodon API, i.e. on Layer 2, with even more limited interop on Layer 1 (WordPress, Lemmy, Discourse, Fedify), and the project has so far evidenced no interest in interop or federation on Layer 3. And that would make sense, because the hypothetical “developer portal” that @trwnh mentioned would be something Meta has built many times before-- I already mentioned Facebook Connect and the pivot to platform-as-infrastructure, and I’ve previously linked @JasonJV 's excellent article on how that could play out across Layer 2.

Counterintuitively, Layer 2 might be the only horizon on which all hope is lost, and Layer 1 might actually be where the most effective interventions are possible.

2 Likes