Hacker News Reader

The tying of identity to one’s home instance is IMHO a fatal flaw. Absolutely fundamental error in a decentralized system, making it effectively not decentralized.

It’s understandable in ancient protocols like email where storage was at such a premium that universal replication was out and cryptography was primitive. It’s not forgivable today.

I am ignorant of AT — does it have this problem? I know that Nostr doesn’t and it’s always struck me as technically superior. Problem is there is nothing on there but Bitcoiners and all the topics adjacent to that subculture.

Nostr sadly doesn't scale. IMO it's a better system for decentralized account identity lookup but not great for content delivery. It needs something else for the content part.

ATproto allows data to be hosted off-site but account lookup goes through the Bluesky owned centralized infra. Just my hunch but maybe its "federation" aims is just a sugarcoated version of "it's a carbon copy of late 2010s Twitter microservices, but we're building it on public IP with intentionally minimal authentication".

Hmmm... if Bluesky owns identity then it's just another centralized SaaS play which I guess is to be expected.

There is zero mechanism for the funding or promotion of anything that's not a lock-in play or a data play (or both).

I didn't realize Nostr had such scaling problems but I think it makes sense now that I consider how it's a client-server system with a network of servers. Making all traffic go through it that way is going to cause scaling issues or require scale-up of infrastructure that will break decentralization. AFAIK they intentionally passed on P2P because "it's hard," which is true, but it's also how you don't pay for bandwidth.

IPv6 has enough penetration now that you could probably get away with easy mode P2P where IPv6 is required. You still have to hole punch there but it works about 100% of the time because no port remapping. (Even the few areas where V6 NAT is deployed, it's usually 1:1 NAT without port remap.) If you don't have V6 you get a slower experience because you have to relay.

All kinds of innovations of the network stack would be easier when IPv6 has that penetration. I saw a very cool vid by Brett Sheffield of Librecast [0] titled "Privacy and Decentralization with Multicast" [1] (btw, it is hosted on a decentralized PeerTube instance) and it was an eye-opener for me, as the average tech person not deeply into this stack and taking the one we have for granted (mostly).

[0] https://librecast.net

[1] https://spectra.video/w/9cBGzMceGAjVfw4eFV78D2

I've wondered if this might not be a reason for some of the slow rolling. It might reduce the all-important role of cloud and centralized services in facilitating connectivity, which is almost mandatory in IPv4 world due to the existence of symmetric NAT.

Yes, I had the same feeling. There's still a massive amount of money sloshing around in cloud vendor market to ensure people remain glued to their services. And then there are a range of new technologies that are all like dark clouds threatening this digital cloud playground. Local-first, P2P networking, generic sync protocols, Wasm-everywhere, etc. where a paradigm shift in computing away from both cloud and web browser hegemony is possible, and these become optional choices instead of 'where it all happens'.

There's much more interesting innovation waiting for adoption on that lower part of the internet stack I suppose. As someone for whom that's a too specialist area I would love to have more overview of what are the promising technologies and upcoming standards to place early bets on.

The other day by accident I found out about Named Data Networking networking [0] via a paper [1] "Exploring the Design of Collaborative Applications via the Lens of NDN Workspace", and saw that NDN still sees active development after many years, so I wondered about the extent the technology still is considered promising for mass adoption today.

[0] https://named-data.net/

[1] https://arxiv.org/html/2407.15234v1

Your referring to the ID registry (PLC) which is intended to be moved to a separate org.

Why doesn’t nostr scale?

AT protocol currently supports two different schemes for base identifiers based on w3c's DID system. The DID:PLC scheme is centralized, but not inherently tied to any one host. It is currently hosted and run by the Bluesky PBC, but they want to spin it off to help protect against themselves turning evil. This system lets users change their handle and move to a new personal data server seamlessly.

There is also DID:Web. This one has the downside that you need to continue to control the domain name in question indefinitely, and it can be argued that the domain name system is still a form of centralization. Like PLC users can theoretically change handles to another domain name with this scheme (but must contrinue to control the original domain name). Users can freely move to another personal data server.

AT Protocol can add new DID schemes in the future to avoid these downsides, with the caveat that users cannot change from DID type to a different one seamlessly, and adding new DID types may potentially require updates by multiple other parts of the ecosystem.

> And the third one is it takes them forever to ship updates. Bluesky is so much faster moving when it comes to implementing new features, but Mastodon ships even slower than Twitter which is an "achievement" in itself.

Mastodon is a non-profit with a handfull of engineers. How can you compare their resources to something like Bluesky or even Twitter, that has thousands of engineers, is beyond me.

Bluesky also has but a handful of engineers

But was initially started internally by Twitter with millions of dollars of funding and since being split out has taken several million dollars in outside funding from VCs. Which does help grease the wheels somewhat, no?

Yeah, probably.

I don't know how many engineers work at Bluesky, but my guess is that their yearly budget is at least 10 times of Mastodon.

> Mastodon is a non-profit with a handfull of engineers. How can you compare their resources to something like Bluesky or even Twitter, that has thousands of engineers, is beyond me

Put another way, they were structured in a way that doesn't allow them to compete.

so... it's IRC all over again. I wonder why we need a new protocol for that