s19e08: It’s Not 2006 Anymore; The Useful, Vernacular Institution; What Is Free Our Feeds, Anyway?
0.0 Context Setting
Monday, 24 March 2025 in Portland, Oregon where it is not raining.
0.1 Events: Hallway Track
No Hallway Tracks on the slate at the moment.
1.0 Some Things That Caught My Attention
1.1 It’s Not 2006 Anymore
I was at ATMosphere Conference 2025 in Seattle over the weekend. I’ll do some quick scene-setting shorthand so that some people will understand why I’m so excited coming out of it:
It felt like being at an O’Reilly ETech conference in the early 2000s. These were the conferences where web2.0 was invented, solidifed, and named. They were incredibly optimistic times and clearly with the benefit of hindsight, we were stupendously naive and privileged. You just have to look back over the last 18-odd or so years to see how what we thought would happen by default (“more people connected means good things will happen!) totally did not happen and instead ended up reinforcing and making existing power structures worse. Sorry.
But the energy from those conferences was there over the weekend. It felt like the beginning of something that could make a difference.
What matters is how different ATMosphere 2025 was. For one, the people who were there in the early 2000s were there to say amongst other things: these are the mistakes we made. These are the assumptions we made. Please make new mistakes. There’s no need to go over what we did. Here’s what we learned. Here were our blindspots. Here’s who wasn’t in the room. So it was different -- and better -- that Rudy Fraser from Blacksky was there.
Now is the time that we need to be making new mistakes.
So what got me so excited?
1.2 The Useful, Vernacular Institution
First was how Erin Kissane made clear the importance of what she introduced as useful vernacular institutions, and the concepts of legibility vs illegibility. These are the institutions that:
- are emergent
- are highly local
- prioritize needs on the ground over state or government bureaucratic needs
- are more useful than they are legible
Right now, as the administrative state is being dismantled in the United States, we’re going to need these illegible, vernacular institutions more than before. For whatever criticism you might have of the efficacy -- not the efficiency -- of the administrative state, it is supposed to keep people safe (yes there is a big caveat there). Or at least that’s the ideal that it strives for in our societal contract.
What are these illegible institutions like? In the network world, they’re institutions like IFTAS, which is (was?) an organization dedicated to independent federated trust and safety on the ActivityPub/Mastodon network. That’s one of the options we have for a decentralized social network, one that is more immune to takeover from a psychotic billionaire.
IFTAS emerged from the administrators of Mastodon servers needing to work together to deal with issues like CSAM - child sexual abuse material. You have to deal with those things now, not in the “wish you didn’t have to” or “maybe it will go away”, but in the “the real world is inseparable and legislation and consequences exist” sense. IFTAS was funded for a bit, but because it was illegible -- it was behind the scenes, and not something like a Formal Proper Institution, it was hard to understand by the kind of people who would give it money so that it could be sustained.
But one of the things about illegible institutions is that they survive and work precisely because they’re illegible. The work they do is so important to such a vulnerable population that it pretty much has to fly under the radar because it is against the wants and needs of the state.
“We” need to understand these institutions better and find ways to support them. And you can drive a truck through what “we” is because it actually needs to be defined and is very different depending on who the particular we is at the time.
I’m not entirely sure if this is clear -- Erin’s put up the reference page for her talk, so you should go look at that.
1.3 What Is Free Our Feeds, Anyway?
The second thing that got me excited -- or at the very least activated was learning more about Free our Feeds from Marc Faddoul’s talk and breakout session.
Here’s what I understand:
“We” (people who used to be on Twitter?) don’t want social media under the control of a small group of billionaires and we should do something about it. A whole bunch of people have signed up to an open letter supporting this position, ranging from Jimmy Wales to Roger “Oops I accidentally the torment nexus” McNamee, Brian Eno, The Internet’s Cory Doctorow, and Mark Ruffalo.
Marc was relaying the intention for Free our Feeds to raise, say, $30 million to... do something? To make sure that social media could remain independent?
What would that mean? What are the opportunities? It’s such an amorphous goal which would dictate very different approaches depending on what you’re trying to achieve or prevent.
For example: does “free our social media from billionaire control” mean something like “Europe wants data sovereignty along the lines that its regulators have started, belatedly, with the Digital Markets Act, that’s trying to reign in Meta, X, and Apple and so on?”
What exactly does data sovereignty mean? If this is predicated on ATProto, what part of the architecture is that? Is it at the PDS (personal data server) layer, which means that you could choose to have your data stored under European jurisdiction? Because you can just legislate for that. You don’t need $30 million to do that. You can do a China and say “hey, if you want to run iCloud here, then you need to run it in China so the government can do whatever it wants with it.” You can do it without legislation, I think, by the U.K. government saying “hey, if you want us to use Amazon Web Services to power government services, then we want those physical datacenters to not be in us-east-1 and actually in the United Kingdom”.
So you could “just do that”. Does it mean running a relay -- and here you need to really understand the architecture and what a relay is doing.
Do you think you’re in the sort of position that led to the creation of the Galileo constellation, Europe’s equivalent to the GPS satellite network, as a backstop in case the U.S. ever decided to turn off access? The first Galileo test satellite was launched in December 2005 in an environment I’m sure that included people (probably including me!) saying “it’s really expensive to create a global positioning satellite system, why not just use the one that’s already there?” and now we’re in 2025 where the U.S. is pretty much saying “we are going to annex Canada” and nation states are saying “we regret buying the F-35 because you will just turn it off.” We didn’t think those would happen!
But Galileo solves a specific problem which is “stuff can’t know where it is anymore, so then it is broken”. It is easy (I think?) to point to the infrastructure and architecture that needs to be:
- designed
- built
- launched
- maintained
- upgraded
Galileo is also... a single thing you can point at?
Again, the question is: what is freeing our feeds?
Is it: Europe is jealous that it doesn’t have a big successful social network with tens of millions of users? Because you can’t “just make” one of those, they come about because people choose to use them, not through mandate.
Is it because one failure mode is:
- I’m a President
- I use “Twitter” (or Bluesky!) to communicate with my nation state’s population (and more)
and then:
- Someone bad is in charge of “Twitter” now
and then:
- I am being censored
- My account has been compromised
- My data has been compromised
Are you hoping some sort of interoperability helps with this? Then you need to really think through how that interoperability would work in practice.
ATProto, which Bluesky the Thing is built on, is an open protocol. Bluesky itself is built, more or less, as a Single Thing On ATProto. While you could have your PDS somewhere else, most people have their PDS hosted by Bluesky The Company. What is Bluesky? Which parts of the below are more important than the others, or are they all equally important? Which parts can be separated out, and how might that work in practice?
- Is it your graph of connections?
- Is it the posts you’ve made, the interactions, and the direct messages?
- Is it the moderation that’s built-in? What moderation? Accounts? Posts? Spam? Dealign with CSAM?
- Is it the ability while using Bluesky The Thing to also add other third party moderation services, the whole composable moderation thing?
- Is it Bluesky The App You Use On Your Phone That Accesses Bluesky the Thing (Network Edition?)
- Is it the ability aggregate all the posts in your graph and view them in a browser or on your phone?
And what can you practically protect against? There’s “is bought by an evil billionaire” and there’s also “happens to be physically located in a nation state that has no problems whatsoever and no resistance against jack-booted masked agents of the government going to grab servers at the behest of someone who literally is fine with just pulling random cables out to see what happens”. What are you looking to mitigate here at, say, the nation-state-with-physical-access level, and is that practical or reasonable on any sort of risk/benefit analysis? Is it even possible?!
I’m clearly thinking out loud. I think this stuff is important! But it needs detail, which is one of the things that I felt was sorely lacking in terms of preparation from the representative of Free our Feeds, which is especially distressing in terms of the stated request for funding to “do this”. One of the requests was even -- if I’m hearing it right -- the chance to have a do-over so that Europe would have more than one online success other than Spotify.
This... is not a way to do that? There are many reasons why Europe has only produced something like Spotify once! They are so structural in nature, so systemic, and so “well, you’re in a post-/late-capitalism environment what are you gonna do?” and that was before this shitty new world order we’re in now that we’re post Pax Americana, whatever that was.
You want to run “another relay”, where relays are the things that aggregate data from individual PDSes -- so aggregate all the posts that people are posting, for example, and then emits a stream/firehouse to be consumed for someone to make an “application”. Bluesky at the moment a) hosts PDSes for people, b) has its own relay to produce a firehose, and c) runs its own application that consumes that firehose to end up with a social network that people use.
I think -- but can’t remember -- that you can tell Bluesky The App On Your Phone to use a different PDS. I think you can have your own PDS and “be on Bluesky” which means “people who are Bluesky users can interact with you”. I think you can... tell Bluesky Your App On Your Phone to use a different relay... for what?
And then if you want to “run a European relay” honestly one of my first reactions is “you know, relays are great for surveillance”, but then another one is “okay specifically what do you mean by “run a European relay”?
Do you mean that somewhere physically located in Europe, there are some servers that are crawling PDSes everywhere in the world and producing a firehose to be consumed by... some other things (you can make up your own rules as to what the relay will listen and talk to! You control the relay!). Which nation state is this in? Who is paying for it? How are you doing ops?
We have models for this that are, you know, nation state infrastructure. They are things like utility networks. They generally don’t start off privatized or funded by philanthropic institutions. Normally this is because states see them as in their interest. In the UK we used to have this with the General Post Office which for a while ran telephony as part of its remit to make sure that people in the UK could talk to each other. Then telephony was spun out into British Telecom, and then British Telecom was privatized. Other utilities in the UK bounce between state-owned and privatized depending on who manages to be in power at the time.
So, okay. Are we looking for a situation in which Mark Ruffalo cannot be silenced? I have no problem with this at the moment, there has been no milkshake ducking and he was fantastic in Poor Things. So No Caesars Jay goes evil and bans Mark Ruffalo. Marc then... migrates his PDS to another service, and is able to do so without losing any data and preserving his identity. Can you make Bluesky The Service use relays that crawl Mark Ruffalo’s now not-on-Bluesky PDS? Probably not! Are you able to tell the client you’re using to access Bluesky The Service to also use a relay that is crawling Mark Ruffalo’s PDS? Does that even work?
Here’s a made-up user experience in my head. I use Bluesky. I follow Mark Ruffalo. One day, Mark Ruffalo is persona non grata. The way I understand it is that Mark Ruffalo would then not show up in my Bluesky-The-App-And-Service feed. Someone else could make a feed available that is the All The Hollywood People Bluesky Hates, and I could subscribe to that feed, which does aggregate from Mark Ruffalo’s now not-on-Bluesky PDS. Got that?
But the actual experience is that while I can still access Mark Ruffalo in that feed, if I’m using Bluesky The App On My Phone Which Is Synonymous With Bluesky The Social Network and Mark Ruffalo is not in the default feed then... Mark Ruffalo is not there... because I would have to swipe to a different feed.
Now -- still thinking out loud -- if you wanted to have an experience in which Mark Ruffalo never disappears from your Default Feed, you could get around this by saying that you have a Bluesky-Lexicon-Speaking End User Application that retrieves feeds from whatever aggregator you want, and retrieves a feed from Bluesky The Social Network.
Let’s call this FreeFeedApp. You’d start off using this FreeFeedApp. It would default to the Bluesky aggregated feed because that’s where the actual people -- i.e. Mark Ruffalo -- are. But your experience on FreeFeedApp is not 1:1 the Bluesky Aggregated Feed, it’s a synthesized feed from FreeFeedApp that includes whatever other aggregated feed you choose. So one day you hear that Mark Ruffalo isn’t allowed anymore and you see that you can get around that by joining the Mark Ruffalo Feed. If you’re using a client -- like FreeFeedApp for iOS -- that presents an aggregated-aggregated feed, then your main timeline doesn’t lose Mark Ruffalo. Your No Mark Ruffalo and Mark Ruffalo Feeds are seamlessly presented together.
That’s a lot. Is that what you want? Because now the request or fear isn’t just about “Mark Ruffalo can’t be accessed anymore” it’s that “you can still access Mark Ruffalo, but depending on the end user application you’re using, Mark Ruffalo may or may not display in your default feed view”.
Another way of looking at this is:
- You have an email client
- You have an account on email service A
- You have an account on email service B
- Your email client presents a unified inbox view aggregating your inboxes from service A and service B
- Mark Ruffalo sends email to your account on email service A but is then spamblocked by that operator and disappears from your email client’s unified inbox view
- You subscribe to Mark Ruffalo using your account on email service B, which has a very strong pro-Mark Ruffalo stance
- Mark Ruffalo re-appears in the unified inbox of your email client
This is easier for a client to implement for your desired service because “all” the developer of your email client has to do is speak imap/pop rather than a particular protocol for each email service.
Which is why being precise about terminology is important: for one, you need to be specific about the layer. You could mandate, for example, that All Social Network Applications Must Present A Unified Inbox Or Feed View and that All Social Network Platform Operators Must Make Their Services Accessible Over ATProto.
Because I think one of the fears is about reach (I am agreeing that data sovereignty is also a legitimate fear! I am not discounting it! That’s why I’m saying “one”!)
And reach is super difficult because it’s a societal issue and this is a great (maybe?) illustration of how governments need to understand how living in a networked society works in order to, well, govern that society.
And this is why part of this comes back to the question/observation I raised in the breakout, which is this: is this because Europe is jealous it doesn’t have control over a popular social network that regular people use?
“I am not appearing in peoples’ inboxes” is a really fucking slippery slope to fall down, and one that can be mitigated in lots of different ways. I suspect that people think “you can’t stop who I subscribe to” means “I decide who appears in my feed and not the platform operator” but people also don’t want that because you want the platform operator to deal with all the shit to do with moderation, which is also why composable moderation is an interesting thing.
But if you really want the experience of “I don’t want to be disappeared from people’s feeds because of the arbitrary whims of a billionaire”, then you can’t avoid paying attention to the behavior of user’s client.
I am going to stop now, because this is nearly three thousand words but could go on for much longer, and am very much itching to talk to more people about it. Perhaps I will do a Hallway Track.
That’s it. I’ve been away. It’s been complicated and difficult. Sorry.
Oh, also: I have availability to help people. Go check out what I do, or have a chat with me.
Best,
Dan