Returning to the #FreeOurFeeds (FOF) initiative discussion (for background see links below)...
@pluralistic has a new piece (https://pluralistic.net/2025/01/20/capitalist-unrealism/) that extends his "fire exit" analogy and discusses how it is not corporate ownership, VCs or profit motive alone that causes enshitification. It also requires captive users, and FOF will make it so Bluesky users are not captive. It all sounds good, but it's not realistic because the assumptions behind it are based on vaporware marketing.
more...
Somewhere in the 1980s software companies discovered that they could announce products with fabulous features long before they were developed as a means of getting potential customers to delay purchase decisions for their competitors actual real products. The term 'vaporware' came to refer to these schemes. Over the years vaporware has evolved to be used for all sorts of clever market manipulations and you could say that most of Silicon Valley now runs on vaporware.
more...
Sometimes vaporware products actually get made with a small subset of the promised features, but more often they just continue to get delayed serving their market manipulative purpose. One thing about vaporware products is that there is no shortage of complex and detailed descriptions of what they WILL DO.
more...
Enter Bluesky and the AT protocol. We are told it is a simple matter to make Bluesky distributed. All we need to do is make another 'relay'. Seems easy, right. But Bluesky hasn't done this yet because [insert reason here]. Enter FOF, who figures, well Bluesky won't build a second relay with their technology, we will do it!
Do you see the problem here? Maybe, building complex undeveloped, unproven, untested relays is a huge engineering challenge. Maybe it's impossible.
more...
Certainly seems like it will be impossible without Bluesky's active involvement and support. But, the whole reason for this effort is we imagine a day where Bluesky will be the adversary and something they actively oppose. People running to the fire escapes while Bluesky is pulling up the ladders.
FOF is untenable. It's an engineering nightmare based on vaporware marketing and even if it miraculously got built it is impossible to imagine Bluesky not able to sabotage it.
References
Supporting FOF:
https://freeourfeeds.com/
https://www.technologyreview.com/2025/01/17/1110063/we-need-to-protect-the-protocol-that-runs-bluesky/ (@elipariser)
https://pluralistic.net/2025/01/14/contesting-popularity/#everybody-samba (@pluralistic)
https://pluralistic.net/2025/01/20/capitalist-unrealism/ (@pluralistic)
Critique of FOF:
https://mastodon.online/@mastodonmigration/113829076223876008
https://notes.ghed.in/posts/2025/bluesky-free-our-feeds-mastodon/ (@cebedo)
https://malici.ous.computer/@shellsharks/statuses/01JHN5A9844W3SMPE2A39K0V4J (@shellsharks)
Background:
https://dustycloud.org/blog/how-decentralized-is-bluesky/ (@cwebber)
https://dustycloud.org/blog/re-re-bluesky-decentralization/ (@cwebber)
@mastodonmigration @pluralistic
There are multiple relays currently in production, somewhere between 5 and 10. It costs lower than 50 dollars per month to run your own non-archival relay, with the cheapest I've seen is someone running it for 23 dollars per month (single consumer)
there are also alternative software implementations of the relay, cerulea is one done in rust for example
relay is not even the centralising aspect of atproto, other aspects are
"non-archival relay" is doing a lot of work here.
@mastodonmigration @pluralistic
it really is not, and it is very clear that you dont know how atproto works, and what the actual risks to the system are
atproto has clear failure points, and relays are very much not one of them lol
Love this kind of techmansplaining. You are correct, no idea how atproto works, and don't care. Not being specific in the use of the term 'relay'. Have no idea what that is, and don't want to know. Not getting drawn down into the marketing lingo.
What would be convincing that it could be distributed is if it were actually, you know, distributed.
Not interested in tech vaporware buzzword salad or theoretical papers. Where do you sign up for this distributed network?
@laurenshof @mastodonmigration @pluralistic In the context of FreeOurFeeds, I think the discussion is less about "is Bluesky decentralized", or "can Bluesky be decentralized", it's "why is FoF asking for 4mil (up to 30 mil) to stand up a single 'other Bluesky' ". I understand some of that money is earmarked for "governance stuff" and for the developer community it hopes to support, but it raises the question of "how expensive is it to run Bluesky", just talking operations, not active development, not salaries, etc... If it's TOO expensive to run a single Bluesky, then it'll be too expensive to create enough federated Bluesky instances to make it truly defederated in a way that would truly free us from billionaires. If the FoF experiment can prove that instances of Bluesky can be stood up and run by non-corporate interests, then maybe they're onto something, otherwise, Bluesky isn't likely to free us from anything. I say this not as some Fedi zealot, but as someone who wants to see folks free'd from centralized, billionaire-controlled media in whatever way will work. If it's Bluesky, then GREAT! I think this is a healthy amount of skepticism given what's at stake, and the countless other times we've seen social platforms become...well, what they've become.
Sad the tone has to turn nasty here too . I think we're all just trying to learn and educate.
@shellsharks @laurenshof @pluralistic
Mostly agree, but 30 years of software development management experience screams when someone says that it is easy to do something big and complicated and you don't necessarily have access to all the inner workings or know the quality of the code. People generally way underestimate what it takes to get big production software working, and if you are on the outside trying to do something with someone else's interface, good luck.
I also agree with @shellsharks.
5M–30M funding would be more effective if invested in improving the Fediverse, e.g. adding composable moderation, instead of trying to spin up one other Bluesky relay.
Tbf, Bluesky is bigger than the Fediverse. So one can argue that you help more people investing there.
@mastodonmigration @pluralistic that was my thought too ...if standing up a relay to create an alternative feed takes $30M, then there will only be at most a handful of them; and initially with FOF, even if it succeeds there would only be two. Wherever you have only two to five options, that's not a real, effective "fire exit", because it's trivial for those two to five parties to get together and agree on enshittification for their mutual benefit, to the detriment of users. It just doesn't seem worthwhile to hack fire exits in a corporate owned building, if instead of leading outside, they just lead to the building next door which is also owned by an extractive corporation with no real constraints.
@sspopovich @mastodonmigration @pluralistic If only people who can access $30M can stand up a Bluesky clone, that pretty much means only billionaires can do it. Not a great way to escape oligarchy-controlled social media.
@dgoldsmith @sspopovich @pluralistic
Tough to fault that logic.
Meanwhile did you see the European Digital Rights Initiative (EDRi) report which calls for "substantial" investment in public social media and in particular Mastodon and the Fediverse to counter X and Meta
https://edri.org/our-work/meta-and-x-are-going-rogue-here-is-what-europe-should-do-now/
Would be really good to see Europe get serious about this space.
@mastodonmigration @pluralistic Bluesky has gone all transphobic today. Guess they are out!
@mastodonmigration @pluralistic Why would you expect Bluesky to create a second relay? They already have one.
The entire point about relays is that to keep the ATverse federated, independent relays should exist. It has to be a third party that creates the next relay.
@mastodonmigration @pluralistic As to why there isn't a second relay - well, it's a design flaw that there's no incentive to create one, and that's a legit criticism of AT. But a non-profit creating one to support the network is definitely one way to get a second relay.
@mastodonmigration @pluralistic (The main reason it's not been done "for funsies" is because relays are rather resource intensive. You'd be looking at a fairly major bandwidth and storage bill if you set one up. My view is the entire concept needs to be revisited because of that but... *shrug* Regardless though, a suitably big institution should have no problem hosting one.)
Love your confidence.
Would expect them to build "a second relay' because that would prove it was possible to do so.
Let's be clear, we are using the term 'create a relay' here to mean actually create whatever technology is needed to make the AT Protocol distributed at scale, in a production environment. So far, Bluesky has demonstrated that they have a centralized social media platform that can accommodate 10s of millions of users.
more...
They have not demonstrated that they have a distributed platform that can accommodate multiple nodes in a system that can handle tens of millions of users. This is a much much harder thing to do.
All we have is their description of how the AT Protocol should work to do this and assurances that it can be done. We have no idea what is "under the hood" or how many shortcuts they took to get the centralized system up and running.
more...
And the whole shooting match is bursting at the seams, growing like a weed with engineers doing whatever it takes to keep it up and running while adding new features in response to management demands.
It's chaos everyday over there. And into this environment FOF is supposed to show up and get access to their inner workings and make the whole network distributed. Slim chance.
@mastodonmigration @pluralistic They don't need to build a second relay to prove it's possible to build a relay, no. They just need to build one, which they did.
Unfortunately, that's not how it works.
@mastodonmigration @pluralistic Yes, that's exactly how it works.
Bluesky is an active, up and running, social network that's built upon the three technologies that make up AT. It currently has something in the region of 30 million users, 3.5 million of which are active.
Saying they haven't "proven" that they can build a relay is essentially a ludicrous moon-landing-hoax level conspiracy theory at this point.
No. They have proven they can build a centralized system that services 30 million people. They have not proven that the system can be distributed at scale. That is a complex engineering matter even if the underlying protocol was designed with this intention.
All this talk of 'relays' is just pseudo technical descriptions to make it seem like it is less complicated than it is to actually create such a big distributed networked system.
@mastodonmigration @pluralistic All of the other components (PDSes, App Views, etc) have been replicated in some way and are somehow able to talk to this "only works if it's centralized" relay? That's what you're claiming?
Like I said earlier, there's a clear, logical, reason why there's only one relay: there's no incentive to run one, and they're expensive to run. There is NOTHING exotic and hard about the concept. (1/2)
@mastodonmigration @pluralistic
But sure, it can't be that, it must be that Bluesky are lying and haven't really created a relay at all! The messages are flowing from PDSes to App Views etc via some... uh... magic? A room full of squirrels?
This is just silly. Especially when it'd be easier to build one than to fake building one. (2/2)
So you get the last word. Don't feel like we are making any progress and your tone is becoming sarcastic and hostile, so exiting the conversation.
@mastodonmigration @pluralistic Sorry if I come across that way. I'm just increasingly bewildered. There's huge criticisms one can make of the Relay concept, notably the lack of incentives to run one and the huge amount of resources needed. But you're focusing instead on an assertion instead that makes no sense. Yes, a relay exists!
Incidentally, after reading up on it, I found that you weren't even right claiming Bluesky hadn't stood up a second one. They did just that: https://whtwnd.com/bnewbold.net/3kwzl7tye6u2y
@mastodonmigration @pluralistic But regardless, they do actually have a relay. It's not fake. It's not made up. It's a relay capable of managing the 3M/30M users of Bluesky right now. You can prove it exists by using a third party App View to access a third party PDS.
@mastodonmigration @pluralistic And remember, each Relay manages its own data. It doesn't talk to other Relays, it just finds PDSes and crawls them, and answers queries as it gets them from App Views. So it's not like this is a complicated networking thing that'll suddenly fail if there are 256 relays or anything like that. Each is 100% independent from one another.
@mastodonmigration @pluralistic What's needed is not proof that one can exist, but for indepedent replication of all of Bluesky's components so that more AT protocol people are not using Bluesky's PDSes, App Views, and (indirectly) relays, than are. That'll ensure the network stays safe and open going forward. And that, not "Bluesky doesn't really have a relay", is the real thing to put pressure on here.