Hacker News Reader

Comment by pfraze | Hacker News Reader

The bluesky team worked previously on IPFS and Dat/Hypercore and Secure Scuttlebutt. Whyrusleeping - one of the core authors of IPFS - has been an active technical advisor from the start. ATProto is basically those p2p & content addressed techs moved into the server stack. None of us are bullish on client side p2p for large scale publishing or social applications, and we spent 10 years each doing that.

View on HN

I'd love to hear what the Bluesky team thinks of Nostr. It seems pretty damning to me that a founder of Bluesky is now talking up the latter.

Dorsey didn't like the idea of moderation, which sure, probably isn't necessary for a while billionaire guy to feel safe.

Nostr is tied to the crypto coin crowd, which is largely off-putting to most people

Dorsey is certainly focused on that realm

nostr's protocol is also fairly unstructured compared to atproto. ive researched bluesky, farcaster, and nostr, and their cultural origins have definitely impacted how they have approached their designs.

There is no predicting what approaches will win, but those 3 are the current viable options I think.

FYI I am part of the crypto crowd myself, but I am a SWE looking at these systems, not a CT gambler, lol.

Oh, so what about the origins of Bluesky by the same founder(s) as Twitter ?

That alone gave me pause, as Twitter ended up as one of the most despicable platforms we have to suffer today (and yes, those issues are inherent in its nature, and long predate Musk buying it). But I guess that they might have learned from their mistakes this time ??

Dorsey funded Bluesky and not much else.

One can authenticate message integrity on the client side and the server side, it doesn't have to be a trade-off. The same is true for encryption and decryption.

P2P is hard, but that does not mean undesirable.

Feel free to pick up where we left off, but just know that you're in for a lot of pain on key sync, device pairing, unreliable data availability, poor connection establishment latency, high end-user device resource usage, and very small data indexes which make it nearly impossible to produce even moderately-sized social networks.

I think you're looking at it the wrong way. It's not just hard, but the user experience with it is just terrible. The masses will not use it. If the masses will not use something that requires network effects to be successful, there's no point. These problems may be solvable, but I think I'd trust the opinions of people who have worked on it for years and decided to do something else.

Reply on news.ycombinator.com