I'm going to let other people comment on the other aspects of this, but one part I want to speak to is this:
"everything we wanted in terms of an open source protocol, suddenly became a company with VCs and a board"
🧵
Jack Dorsey's problem with Bluesky is...moderation. And the absence of Nazis.
"they started asking Jay and the team for moderation tools, and to kick people off. And unfortunately they followed through with it. That was the second moment I though, uh, nope."
www.piratewires.com/p/interview-...
I've been working in open-source decentralization since 2012, when I got involved with Secure Scuttlebutt. I then started the Beaker Browser with the Dat protocol.
I have an extensive history with community driven open-source without any kind of company driving it. I know the ins and outs deeply.
I'm also familiar with the "pure protocol" concept, the notion that you can build the technology as something for everyone, and then gets adopted by various companies and rockets off to success
We did that multiple times, and it's how Nostr operates
Let's start with the pure protocol thing.
That's never how this works. The Web didn't work this way. TBL wrote the specs, and then shipped WorldWideWeb-dot-app
You need an actual product, and actual product-thinking, to drive development. You need to be thinking about users. You can't isolate it.
The original plan was that Twitter would be that first client, and I would've *loved* that, but Elon killed that straight dead. That entire company was frozen by the prolonged acquisition, and the agreement quickly ended when Elon took over. It was never going to happen.
Now, imagine if we had dug in our heels and said "We're just the protocol guys. We don't ship apps."
Where would this network be? Where would the *protocol* be? It would've been a terribly risky bet to expect other companies to reach the market while downstream of us
My experience with the fully community-driven approach is this, every time:
Devs have a blast rapidly iterating in a community
A bunch of projects come out, none of which work exactly right
The technology keeps missing key features
Users never come
Everybody gets super frustrated
The devs abandon
Me and the bois have been thinking that you'd needed to shift even faster away from being a protocol when the whole Twitter/Elon/X thing happened. There is no protocol without the app really rocking. And we are not there yet even today.
See also: Mastodon
People make a lot of good criticisms of Eugen and the things he's done in pursuit of a product focus with a roadmap and targets and whatnot, but no one would be talking about AP outside listservs. Also probably helped provide a shaping contrast in the development of AT/Bluesky
If you're working on a protocol, giving "other products/companies" second mover advantage isn't strictly a bad thing. It's the tradeoff for them seeing that something has a good foundation/community to build off of.
It's just a weird ouroboros when that protocol is about...community building!
That's one of the reasons Unreal Engine was much better than Unity (back before Unity shot themselves on the foot): Epic got their own games to ensure the engine was usable by developers, whereas Unity didn't have anything to test the tools suitability.
Lack of product-thinking is my gripe with some old-fashioned fediverse devs, many are extremely talented and forward thinking, but one of their bios go like this: "Created fediverse technology long before it became cool. Still building the most innovative fediverse servers you never heard of." 🤦🤦🤦
It's okay to not have the energy to spare in "marketing" your free work as a fleshed-out "app". It's NOT okay to take pride in being unmarketable to the masses. That's just juvenile hipsterism. If you take yourself seriously as someone who wants to make the world a better place, be reachable.
Actually I think you would't need one product but at least two, just to be sure the protocol can handle different use cases and different points of view. I'm sure other client's devs give you good feedback. :)
Client #1 will drive client #2, client #3, etc… faster as people react to the “well I don’t like X about client #1” and they iterate on it.
People staring at a blank page “well I guess I start from scratch” slows progress way down.
As a veteran software dev, I have the expertise to say you are exactly right about this. I'm struggling to express how I feel about "protocol without an implementation" in a way that would properly convey to non-software people just how useless this concept is.
Oh, I've got one:
A protocol without a product is like a writing a recipe for a dish that you've never tasted, using ingredients that may or may not exist.
There's a reason several standards organizations emphasize the need for a working implementation and not just a draft for a protocol before they'll pick it up for standardization
There were also other enabling techs that came along about that time, like onboard modems and color screens. I remember learning how to disconnect my dad's serial cable from the family printer to hook it to the external modem and fire up Prodigy in glorious 16 colors... 😂🤓👴
It just became EASIER.
100%. We don’t encourage engineering for 100x when we are at 1x, but that’s broadly applicable advice.
Build the product to the current market, and begin sub-dividing what you think of as the product as the value and complexity reach natural boundaries (usually around orders of magnitude).
“Why won’t people use the API I made,” that has no reasonable way of using it, is something I have sadly said to myself. I don’t use it either, cool as I think it is.
Further, TCP/IP was stood up by ARPANET and there were existing implementations of it well before it became the protocol of a mostly open platform.
If you want to see how the pure protocol shakes out, see if you can connect to something using the OSI 7 Layer protocol, which was a thing.
Sadly they used the framework for the IPv6 protocol, which has a lot to do with why 26 years after the protocol was created, in the face of ridiculously inconvenient address space exhaustion, IPv6 is configured on almost every workstation and is barely used.
Apologies for spam liking this entire thread, but I think you boiled it down to the key - "Checks and balances, not anarchy" More and more it seems like those balances are equated with censorship, which is not the driving force here. Very insightful, thank you as always for sharing Paul.
There is no way to do open source, protocol-based social media that will satisfy everyone.
There's no way to do it that will be safe for everyone.
What we - and that's the devs and us - are building comes closer than any rational person could have predicted.
Let's keep making that real.
I love your online presence and have a lot of trust in your good intentions, and your expertise in the subject matter. But it must be said: these names are hilarious.
Tbh, it’s the premier evidence that you are just a bunch of Lovely People ™ and care more about the thing than the short term perception of the thing. Sláinte 🥃
The fact that jack doesn’t see the need for formal governance on an open source project says a lot about his lack of experience with community building