Your Feeds Are Already Free

The Free Our Feeds initiative (hereafter, “FoF”) has been seeing a lot of discussion over the past few days. Cory Doctorow signed on and criticized opponents on Mastodon as impractical, puritan scolds. Overall, the initiative purports to be a way to keep Bluesky from turning evil under the control of a maniacal billionaire. They want to raise $30M to, among other things, run a second Bluesky relay, under their control, to free Bluesky users from reliance on a single entity.

I’ve been pretty vocal in my belief that this initiative is a bad idea, as have others. And, I may well be a “scold” for harshing someone’s mellow, or whatever, but my opposition to FoF is not in any way puritanical. I don’t have a real problem with Bluesky or people choosing to use Bluesky. If they had focused on being a federated service from day one, I’d probably invest more time in using the service and have more faith in its longevity and usefulness. I think they mostly do a good job of running a Twitter-like service and there are certainly some interesting aspects to AT-Protocol. Their personal moderation tools, for instance, are lightyears ahead of what’s available on Mastodon, at least for my needs. My objection to FoF is imminently practical: it cannot work technically and it is exceedingly unlikely to work socially, under the stated conditions. And, there’s already an alternative that works today that meets the stated requirements of FoF: Mastodon.

Massive cognitive dissonance on the purpose of Free Our Feeds

Much of the pro-FoF discussion on Mastodon has focused on Free Our Feeds as providing a sort of emergency exit, allowing users to port out their content and social graph in the event Theoretical Evil Bluesky (hereafter, “TEB”,) materializes. FoF proponents quickly dismiss concern that FoF is diverting money and effort away from an already-extant open standard (Activity Pub). Its all about protecting the Bluesky user base, not about harming Activity Pub.

While the emergency exit angle may be one aim, even the primary concern of some FoF-sponsors, it can hardly be said to be the only, or even, primary goal of the initiative. In FoF’s own words¹:

The “emergency exit” would, I suppose, be primarily under the second bullet point. But, make no mistake, the backers of this initiative want to develop and expand AT Protocol’s usage. That development effort could go into making Activity Pub stronger. As one example, Mastodon’s current filtering and moderation tools aren’t as good as Bluesky’s. What would it take to bring them up to par? Would it cost $30M? It is hard to imagine that being the case.

The stated aim is to build a social network resilient to billionaire control, but that already exists with the Fediverse. Why build up a currently-proprietary system when an open one that meets your stated requirements already exists? Why not devote effort into making the already-open system better?

Others have pointed out the lack of specificity in the structure and governance of FoF; I’ll not get in to that here. For one thing, I think a certain lack of detail is understandable and to be expected in a brand new initiative, and second, those issues just aren’t relevant to my thesis.

This cannot work technically

The other common refrain over at Mastodon amongst the pro-FoF crowd is that this is a social, not technical effort. Setting the handwaving about a technical effort (participating in a standards committee and building parallel server infrastructure) not really being technical, I actually almost agree.

Almost.

There is one technical detail that I think matters quite a lot: this can’t work technically, in the sense that the infrastructure and even standards work being proposed cannot be relied upon to give users an escape hatch or really constrain TEB in any way.

Today, Bluesky controls the Bluesky app, the “relay” which is a kind of central mixing server in the AT Protocol, and almost all of the data stores. It is also true that, today, a user can setup their own Personal Data Server which will host that user’s content. The Bluesky relay will pull data from that server and show it to any and all Bluesky users - ones with their own PDS, and users on the Bluesky server. There’s a thin veneer of federation that already exists with Bluesky.

But, two other things are also true: almost nobody has done that, so the vast majority of the Bluesky’s 20M active users are on the Bluesky data server and there is no technical mechanism – and there can be no technical mechanism – to force Bluesky today/TEB tomorrow to continue to support third-party PDS instances or to interoperate with any third-party infrastructure. Remember: they control the app, the view, the relay, and almost all of the data stores. Regardless of any changes in the underlying AT Protocol, they could make their relay (or their app) ignore data from third-party PDS instances. They could refuse to allow people to port their social graph out. Short of taking control of the app away from TEB, there’s really nothing anyone could do, technically speaking, to force them to interoperate.

This is exceedingly unlikely to work socially.

There are, of course, legal mechanisms that could force Bluesky to interoperate in good faith: the government could pass a law mandating they do so. But, relevant government agencies do not currently exercise this control over Bluesky, so its hypothetical. Such a hypothetical government could also be imagined to pass a law forbiding whatever abhorrent behavior would motivate users to want out of the TEB walled garden in the first place. Which would obviate the need for the third-party infrastructure. If we’re just making scenarios up, why not imagine that Bluesky gives $1M to each of their users as a sign of good will, or establishes peace on earth, and good will toward men?

Regulators in the United States have shown no interest in mandating interoperability. The EU, of course, has the Digital Markets Act, but Bluesky is a California-based Public Benefit Corporation and TEB could just leave the EU if the going got tough. Perhaps EU-based users would be able to get their data out in time.

There’s also a network effect (in the social sense of the word) that could have worked in the past, but is unlikely to work now: Bluesky could have started out as a federated social network where it didn’t control all of the users. If only, say, 50% of the users were on Bluesky’s infrastructure and the remaining users were distributed amongst third-party servers (as is the case with Mastodon,) Bluesky would have to think twice about defederating, less they lose access to a bunch of content that keeps their users using the service. This is the one tiny sliver of hope mechanism that has an infintesimal chance of working: if FoF could get a substantial portion of the current Bluesky user base on their servers, while current Bluesky is in a cooperating mood, that could complicate TEB’s future decision to defederate.

But, how many users are going to proactively move to FoF’s server? Remember, Bluesky already has 20M users. Will FoF convince 5 - 10M of the existing userbase to sign up? Will they capture a significant portion of the new signups? Why would a prospective Bluesky user opt in to FoF’s network instead? Bluesky has an enormous first-mover advantage in that they captured a huge portion of the userbase that left Twitter. What event is going to precipitate a move to FoF?

Users are likely to find this “fire exit” welded shut as soon as they need it.

If you believe - as I do - that most users aren’t going to proactively migrate off of Bluesky’s infrastructure, then what of the emergency/fire exit argument?

Viewed as a threat model, the FoF proposal attempts to solve for one of the most unusual scenarios I’ve ever encountered. It hypothecates a specific, identifiable threat by a malicious actor, then predicates its response on specific good behavior on the part of that actor that would be contrary to the actor’s goals and interests.

Let’s stipulate for a moment that FoF raises their $30M, or enough of it to begin operating, and that Bluesky is an enthusiastic supporter of their efforts.

What happens, x number of years in the future if Bluesky becomes TEB and users need this exit? Will they have enough time to migrate off once it becomes obvious that Bluesky has become TEB? Will they be able to get out with their data and social graph in time? I find that unlikely. It isn’t hard to imagine that the first thing TEB does is turn off the federation bridge and stop users from being able to export their data from TEB’s data store.

This mechanism is only going to work if FoF can attract enough users to make TEB think twice about defederating (again, how will they do this?) or if TEB just leaves the federation bridge in place after deciding to start abusing their users. Why would TEB not abuse them a little more?

Your feeds are already free

What’s an end-user to do about this? If you are concerned about a billionaire or other malignant agent taking over your social network and turning it into a hellhole, you don’t need to wait for FoF to build anything. You can join Mastodon today. It works. It’s federated. No single entity can control it. It isn’t perfect, not by any means, but if decentralized control is important to you in a micro-blogging platform, it delivers today.