Can you copy over pauls post? Dont have bluesky access
So #[0] posted a blog post about bluesky that ruffled some feathers: https://fiatjaf.com/ab1127fb.html
This triggered a long skeet stream reply from bluesky developer Paul Frazee: https://staging.bsky.app/profile/pfrazee.com/post/3jv72j3fp6g2r
This then caused me to write up some thoughts:
The world of decentralized protocols is gaining momentum, and it's exciting to see projects like Nostr and Bluesky at the forefront. Many of us have dedicated years to developing these protocols, and now they're capturing global interest. I've been tracking various decentralized social media protocols for a long time, and if you're interested, you can find my comprehensive database of open social media protocol projects here: https://airtable.com/shri7e7EHoTi0cEjO
Nostr, at_protocol, and other projects take inspiration from Secure Scuttlebutt, which I had the pleasure of working on alongside talented individuals like @pfrazee and @jay. Nostr is a slightly modified version of Scuttlebutt, while at_protocol represents a more significant reimagining. At_protocol borrows ideas from the IPFS ecosystem and W3C DID standards, while Nostr incorporates concepts from Bitcoin technology (not a blockchain or cryptocurrency project). Both projects have received substantial support from #[1] , who funds them but doesn't control their direction.
Nostr began as a humble side project, growing organically as developers adopted it. In contrast, Bluesky started with significant press and a high-profile search for a team lead, taking years to evolve from an idea to a funded project. Bluesky's community experienced challenges that led to a split, and the original community renamed itself https://dsocialcommons.org/. Nostr, on the other hand, never encountered such issues, with developers contributing to the project independently.
The two projects represent different approaches to open source development: the cathedral model (Bluesky) and the bazaar model (Nostr). Both have seen success, but I must express my disappointment with Bluesky's choice to follow the cathedral model. Despite my frustration, I have great admiration for the team behind Bluesky and the work they're doing. However, Bluesky employees maintain full control over the at_protocol, leaving little room for external contributors.
In contrast, Nostr provides an open platform for contributions, enabling me to create an app (nos.social) and write specifications that are openly debated: https://github.com/nostr-protocol/nips/pull/457. Bluesky's code is indeed open source, but their development process is not. This is reminiscent of how Safari's WebKit or Android operate as open source projects without truly embracing the open source development methodology.
Recently, I expressed concerns about Bluesky as it currently operates as a single unified network. Friends advised me to take a step back and give the team time. I experimented with their Indigo PDS server and found it promising. I believe that the at_protocol will eventually become an open, multi-server protocol. The people behind Bluesky have a long history of working on open protocols and are not developing this technology to create a new, closed ecosystem.
On a personal note, I feel a stronger social connection to Bluesky's early adopter community but appreciate Nostr's openness for contributions. I could potentially create an at_protocol client, but making substantial contributions to the Bluesky protocol seems reserved for employees and select advisors. Therefore, I choose to invest my time and effort in open projects. I firmly believe in Conway's law, which states that the structure of the organization building a technology will shape the technology itself.
I believe @fiatjaf might be overly critical of Bluesky. He had the luxury of working in obscurity without the pressure of media attention while figuring things out. In contrast, Bluesky faces high expectations and the responsibility to "replace Twitter." The stress that the Bluesky team endures while trying to develop their project under the watchful eyes of many likely contributed to their adoption of the cathedral model of open source. I empathize with the challenges they face in this environment.
Much of the internet was built using the bazaar model, consisting of small pieces loosely joined. This approach gave us the web and numerous other systems we use today. Bluesky's design-driven model is more meticulously architected, but it reminds me of Java and XML (no offense intended).
I believe that these networks can interoperate. I already communicate between the fediverse and Nostr daily, and while it's not perfect, it mostly works. I'm optimistic that we'll achieve similar interoperability with at_protocol once the system becomes more open. It is essential to recognize the strengths and weaknesses of both approaches and to appreciate the incredible work and progress made in both projects. As the world of decentralized protocols continues to grow, I remain hopeful for a future that embraces collaboration and openness.
Discussion
Here it is: "so a couple of thoughts on what fiatjaf is saying
I’m gonna start with this: I don’t feel the same way about nostr or activitypub or most other projects. generally speaking, everybody is engaging in tradeoffs and optimizing for different things. which is a good thing
I’m pretty confident in what we’re doing but I won’t pretend I’ve got it all sorted out. there are so many ways I’ve fallen on my face working on this stuff in the past. I have strong opinions but we’re all working in a difficult space
the anger at us starting with a single instance, I don’t know what to tell you. it was either that or we kept cooking in private rather than testing the core of the software with real users. the absolute wrong move would be throwing the doors open on an incomplete protocol
we’re working on the vibe control tech right now: curation and moderation. *that* is protocol. curation and moderation in a social network just is core protocol. you know it is because we all know the nightmare that would follow if we didn’t do that first
you can open tcp sockets between a bunch of computers and say you’ve created a network but all you would get is noise and resource consumption—so we make things like atproto and nostr to structure the communication once those sockets are open
the same applies to people
if we just open connections between people and don’t structure how they communicate then it’s just going to be noise between them. people will be miserable. curation and moderation tools are core protocol as much as solving how things go over the wire
now let’s talk about his thing about the BGS — the aggregators in the network — and whether those are sufficiently decentralized
I agree: aggregator nodes have power. not a small amount.
I once wrote a paper about the dynamics of power in social technology that was so pretentious that I’m ashamed to share it now but it was entirely about how authority models get embedded in technological systems and this is a really good example of that. authority pockets exist all over these things
the problem is that if you view decentralization from the lens of removing authority entirely from the system then you’re asking for anarchy, which means no resource sharing and an overwhelming amount of pairwise relationship
this is exactly what we were doing with secure scuttlebutt and while I am really proud of that project, I think it proves the upper limit of what can be accomplished if you refuse to entrust any authority in another entity. the answer is: very small scale
and small scale interpersonal networking can be really nice for the right things but that’s not really what we at bluesky want from public broadcast social media. we want a global network. we want big world networking
once you’ve accomplished that you can create small groups and communities within it — which is also wonderful, and greatly assisted if there’s a solid public backbone to connect them
well for large scales the laws of physics just come into play. federated queries do not scale. you need high performance, high volume indexes that can satisfy queries about the state of the network. things like how many likes does this post according to hundreds of millions or even billions of users
people who have worked in the ethereum world may be familiar with Infura. same exact thing. you’ve got an entire distributed VM on a blockchain with ethereum but the basic reality is that companies can’t afford to compute the whole dataset so they use infura instead
so yeah, pockets of authority. what do you do about that?
you lean on ways to check that power. external auditability. an open data network that allows competitors. a low friction to switching off the points of authority
you also do your best to divest responsibilities from those potentially powerful entities
one of the reasons it’s so important to have community driven curation and moderation is so that doesn’t get done by these powerful aggregators, because that becomes the mechanism of control over people
what scares me personally is when a profit motive drives the decision about what’s good for people rather than people deciding what’s good for themselves. and everything that we’ve been building for this protocol is about making sure that the capital intensive systems don’t do that
so these are the tradeoffs we made. this is the atproto system and the bluesky approach. and I’m sorry that the nostr folks don’t see eye to eye with us, but I really do hope they find success because it’s good to have more than one network in the world. because hey, I could be wrong”