Post

André Staltz
@staltz.com

More about Bluesky/atproto in contrast to SSB: with SSB we were uncompromising with our decentralization principles, but we also wanted to see how far our technology could go. There were very few radically p2p technologies like us and didn't know its full potential.

4 reposts26 likes

André Staltz
@staltz.com

In hindsight now, SSB's limit was scalability. Just not the kind of scalability you imagine. SSB *can* scale to millions, billion, trillions of users, with no adjustments needed. The problem is that each user's *view* of the network is limited to <=100k users.

Fleeky
@fleeky.bsky.social

Seems like there is a very real need for compression tech research into p2p ? Instead of just purging you could compress or selectively prune. Also backing up a single large feed across several devices hmm

André Staltz
@staltz.com

Each user can only see as much as their local database can contain content. This is (optimistically!) 100k users on a desktop machine, and ~10k users on a mobile phone.

André Staltz
@staltz.com

The scalability limit is also on the "age" of user accounts. The older SSB became, the heavier each account was in storage cost, and thus the worse performance got over time.

André Staltz
@staltz.com

I was working on a successor protocol to SSB called PZP (I still love that name, it's a play on the term P2P because Z looks like 2), where accounts were auto-pruned to delete older content. This had the promise of "account age" scalability, but...

boreq
@0x46.net

after a lot of thinking I think I'm in the "reliable replication protocol but with relays to help p2p connectivity" camp e.g. nostr but with willow's replication protocol so that we are not guessing if we got all the content