Repost of something I put on Twitter, in response to a post about "Shuttering Google Reader Killed Blogging"

a repeat mistake: being excited about a decentralized system's success because a big player moves in, becoming reliant on it, not providing a better alternative, big player leaves, decentralized system dies from shock

see also xmpp and gtalk

hoping to do better...


"Providing a better alternative" also means more than just "make the UI nice and easy to use", though that's a critical step. What is required to become a participating node in the system? How hard is it to keep that software running?

And yes, before you point it out, I don't think ActivityPub is in a good place here. But I think we can do better (and if you look, you can tell how that's impacting my current research & development)

· · Web · 3 · 2 · 8

Yeah. More than once I've wondered what we'd do if Google did to activitypub what they did to email/xmpp, ie launch a service that uses activitypub, but eventually extends it and starts closing out the decentralized players more and more. It's embrace, extend, extinguish all over again.

Getting and using an activitypub account has to be as easy as possible for average people, without relying on surveillance industry players if we don't want to go the same way eventually.

The timing would be just right with Google+ closed down. They've already made competitive products to Apple (Android, Chromebook etc). They've shown a desire to create their own Facebook, trying more than once already. Combine that with the rise of activitypub and the pattern just talked about and it wouldn't surprise me if they were basically waiting to see if it can grow large and stable enough to experiment with turning activitypub into Gpub, or whatever they decide to call it.

@stardot @Blort pre-emptively blocking the big players isn't what's going to save the fediverse from their wandering in and then out, IMO. If all your friends and family show up on the fediverse finally, but they're using the Big G, I think it'll be really tough for network effect reasons, to not connect. Most of the network ultimately will, especially because whole new parts of the network will spring up that do.

So what's the alternative?

@stardot @Blort The alternative is to build our technology so that "where your content is hosted" matters less. The instance shouldn't be where the community is, or too critical to your character's identity.

Whether its Google or, if either leaves, the content you care about should be able to survive, and the user accounts too. You can see a bit of this here: says some of the rest (there's more, but hard to fit here)

@cwebber @stardot @Blort technically this is awesome, but socially I think the fediverse derives its character (and some of its benefits) from the sense of community your home provides. I know that it's mostly UX but still, the fact that your admin runs the service creates the feeling of being in someone's home, you have to behave as they expect you to. Facebook groups will never be the same.

@qwazix @stardot @Blort I agree that there's some degree of that, and some degree of that can remain! Let's reframe it in a different setting.

Imagine if you had something like a distributed virtual world. Your main character can travel anywhere they've been given access to... but somewhere is their "home", and that "home" is on some server run by other people... but that's also physically manifest. People show up to your home neighborhood, and there are friends nearby. (cotd ...)

@qwazix @stardot @Blort Even still, having a home somewhere shouldn't mean you can't travel anywhere. And how much should your home define you? I've moved around a lot, and my home in each place has had a meaningful impact on my life. But I think it would be also a problem if people immediately defined me by my place of residence. And it would be worse if I couldn't choose to move and still be the same person.

But that doesn't mean I don't have a home with a community.

@FiXato Zot has lead the charge on nomadic identity stuff, including that term. Though I don't think it does it in the way I think is really best. The details of that aren't super important for this post though.

But as for "the ideal", within the next six months I hope to release some demos that better explain what that is. I could ramble on here but I've come to realize that when I just talk it sounds like a soup of "ocaps sealers rights amplification blah blah"

demos show it better

@stardot @Blort Who says Google isn't already surveiling public posts on the fediverse?

I'm more interested in making our systems less surveilable where possible.

@stardot @Blort Sure, I absolutely agree that's a concern. EEE. Happens all the time and it sucks.

I don't know of good solutions completely, other than making it so that it's hard for them to lock in their users. Making the system resistant to lock-in can help... allowing for easy immigration between instances is more likely to help than putting up borders.

@stardot @Blort If you read my top post on this thread, you'll see there's already a large amount of agreement there.

But I don't think that the solution becomes walling them off. Not because we really want them here, but because I think it won't actually protect us in any way.

@stardot @cwebber @Blort we also need laws that forces online identity hosts to provide users a structured archive of their data that is easy to access and migrate at any time, with a generous buffer period before closing the service. In my opinion.

@cwebber The Google crawler (and all the other "search engines") have been indexing the web timelines all the time, by design. Content is being fetched regardless of robots.txt and headers. All it takes to turn the result of that into "surveillance" is some targeted metadata parsing. They don't need to participate in the Fediverse to that end.

On the other hand, public web content is currently the only workaround for incomplete federation, so it's hard to get along without.

@stardot @Blort

@stardot There is precisely ZERO assurance that there aren't multiple parties archiving Fediverse activity. Proving there are none is all but inpossible.

Yes, there's been loud outcry in a few cases intent has been explicitly voicd. But anyone who thinks that this is a surveillance-free zone ... has a lot of growing up to do.

Public is public, and non-public is likely little better.

@cwebber @Blort

@stardot Re; privacy, glad we agree, many people unfortunately don't get, or don't want to get, this.

On EEE my fears parallel @cwebber's I'm not as saanguine as them about the solution, though. Msss-relational systems such as communications platforms and networks tend strongly to centralise.

My theory is that it is irreducible cost functions (such that larger == more expensive) or mutually incompatible communities which are required to avoid exessively high levels of scaling and concentration, not mere (theoretical) mobility.


The alternative that will work, but is hard to do, is real-world organizations that support their own members needs and also support their own technologies and vice-versa.

They may mostly be local. Like, the participants will know each other in real life.

They will want what works for them, and will not be seduced by big players, because the big players will only work for them when they make informed choices to allow that for tactical purposes.

A bit more to come.

@stardot @Blort

@cwebber @stardot @Blort

I know two examples where this has happened (real-world orgs supporting their own tech and vice-versa):

@bhaugen @stardot @Blort I agree with choosing not to work with the big players and ignoring them in favor of local, cooperative interactions. I actually think that's a different thing than trying to put up border walls.

@cwebber @bhaugen @stardot @Blort maybe this is just a daft point from a non-technical user, but i have often wondered why fedi/activity pub goes the open/public route (more twitter-like) rather than the fb closed model. is it easier, for ferder-able, or just what some devs were into? as in, google have oft been hella pissed at fb coz its closed. wdn't it make sense for a decent/community network space to be walled-garden in a way? (i hate walled-gardens, but maybe a kinda anti-one?) this wd also seem to match the fact that fedi etc are much more abt communities than they are abt broadcasting (power). or do ppl rly use fedi things for communicating with non-users?

@mousebot @bhaugen @stardot @Blort the activitypub spec isn't public-oriented, it's built around direct delivery though it does support public, but the way it's rolled out has been primarily by twitter-style applications so that's colored public impressions of what the protocol is

(can't say more than that, completely unrelated to any of this shitty stuff is happening over here right now and I'm putting out fires)

@cwebber @mousebot @stardot @Blort

ActivityPub can totally be used for more private groups, like the Hometown fork.
And I and several other people use single-person pubs as well as participating in group pubs.

@AbbieNormal Datashards is one component, yes! And application towards the fediverse is a large motivator there.

The next big component is the CapTP work happening in Spritely Goblins.


Ludovic Cortes published a reference to a project made in Guile a few days ago

And that resembles some of what you are doing, it mentions "capabilites" explicitely

Do you know about that ?

@AbbieNormal Yes, I think you are referring to @pukkamustard's ERIS and openEngiadina systems

@pukkamustard has been in regular communication with the loose group of people defining Datashards at the moment (facilitated and organized by @emacsen) and we've definitely been in communication. We're hoping for convergence.

@AbbieNormal So yes, @pukkamustard's stuff thus is very similar to Datashards, might very well converge. What I don't think @pukkamustard is doing *yet* is the distributed *programming* ocaps stuff that I'm doing in Spritely Goblins. But I hope that can change; I loosely have on my agenda to try to evangelize that approach to them, with sufficient interest!

(I also plan to port Goblins back to Guile too, so that should make it easier.)

@cwebber @pukkamustard

You're right, I overlooked distributed programming

That's because I don't understand what it is

Instead, I pretty well understand the opportunities of distributing structured information (semantic balh blah) in a NOT client/server fashion

@AbbieNormal @pukkamustard You're not alone. It'll be easier to understand once I have more demos available :)

@cwebber @pukkamustard

I attempted to distribute a bunch of triples along the standards of the semantic web myself, some time ago

A server was lent to the project by a local unversity lab

But that didn't last, because managing servers is challenging

If I had the opportunity, I would have distributed the data myself !

@pukkamustard @cwebber

I know something about RDF and graphs !

And that's why I'm excited about your work !

Further, it's made in Guile !!!

@cwebber @pukkamustard @emacsen

yes ! That's exaclty what I was referrign to !

I was impressed by it !

And I'm glad you know about each other !

@cwebber i want us to do better than matrix too (a LOT better)

@hankg @cwebber dont want this to sound mean, but: a better protocol, specified and described better, allowing other implementations (which gets into cwebber's "freedom to move")

@hankg @cwebber @robey Ironically I've just started studying the Matrix protocol this week. There are a lot of things about it that I like over D* and AP federation. I thought they had a lot of white papers, code samples, reference implementations, etc. to go build something else too. I haven't done as much with it yet so I can't draw any final conclusions yet though.

@hankg @robey Matrix is doing a lot of cool stuff, and a bunch of it I hugely respect. More than anything else, I respect that they manage to push out software and get it in the hands of users. That's huge.

That said I think their architecture of a bunch of bespoke REST endpoints isn't a robust enough foundation to do the kind of things I want to do going forward. (Still some gems in there though.)

@hankg if you have any good links to share, i'd appreciate it! i looked at matrix last year during a holiday, and the spec didn't seem very ... finished. maybe i was looking at the wrong pages!

i was/am very interested in writing my own server implementation of a secure federated chat network :)

@hankg @robey The two documents I spent a bunch of time on were here:……

This site in general has lots of other good links too including an API endpoint "playground":

@hankg this looks interesting, thanks! i wish they would write a spec instead of a handful of blog posts, but i'll take what i can get... this is at least way more than i could find last year :)

i'm really curious why they invented a new conflict resolution instead of using an existing CRDT, hopefully there are some clues about their motivations hidden in here somewhere...

Sign in to participate in the conversation

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!