Nostr Web Client

I'm a fan of how the architecture scales, and the indie-dev community is doing some cool stuff, but I'm slowly starting to suspect that Jack Dorsey may have been right all along.

Nuh 6mo ago 💬 2

The architecture doesn't scale better than ActivityPub unless you are fine with centralisation... for usecase that doesn't tolerate centralisation and opt for small worlds instead, message passing is just as good and more likely to survive long term.

If anything, a mix between message passing and Nostr relays make more sense and is more reliable and scalable.

The problem with Nostr is that it didn't get the datastore layer and message passing right from the get go, and the bet on Websocket and the weird API makes it harder for the Outbox model to be even as good as WebDav.

Reply to this note

Please Login to reply.

Discussion

Garbage nsec 6mo ago 💬 1

Agree that small worlds are where it's at these days.

This is unrelated but before you mentioned an idea about urbit names on Bitcoin as an alternative to that new Spaces deal, maybe pointing to Pkarr. Sounded interesting, did you manage to thresh that out? Also Fiatjaf's rationale for a spacechain to keep things in Bitcoin-land without a competing altcoin made some sense to me, but no idea if that angle has been showing progress these days or not.

Nuh 6mo ago

My idea was simple, basically put your pkarr in an OP_RETURN and use the order of the utxo since genesis block as the ID.

It works, and because it is immutable there is no need to spam Bitcoin to get good names.

But I generally believe we should assume that Bitcoin data will be pruned. And stop pretending that it is a database.

Yes realistically speaking someone will always have an archive of all OP_RETURNs, but there won't be many, and if anything we might be at peak number of archival nodes.

So just from a principal point of view I gave up on this idea.

Spacechain doesn't change this dynamic (I do support spacechains to hopefully extend Bitcoin with smarter opcodes), you can't just pretend that data will be available by lots of people forever... this doesn't scale so any blockchain that doesn't invest in being prune friendly will die, aka be pruned entirely.

Data availability forever for free for many people is just never going to work.

Orgs should keep using ICANN, and fight for it's integrity because seriously it is the best that can happen.

Niel Liesmons 6mo ago 💬 2

What's message passing?

Garbage nsec 6mo ago

Yeeting json into the ether.

Nuh 6mo ago 💬 1

Message passing is the paradigm of organising your architecture as actors sending messages to each other, as opposed to say a central hub that everyone writes to and reads from.

Email is Message passing for example, whereas Nostr is a shared heap of data.

Message passing doesn't scale because you need to send n^2 messages (each email server will need to send a message for each other email server mentioned in a thread).

Shared heaps scale because instead of the (many to many) that results in an n^2 messages, you have a (many to one to many) so you need 2*n messages (from each client to the relay, and from the relay to each client).

Of course that ignores another important downside of message passing which is lack of global view, which henders statistics and engagement signals etc... but that is another topic, although it also relevant because if `n` is so small, it is cheaper to do this aggregation and counting on client side.

So it all comes down to how much scale do you need, and there is always a threshold after which you are doomed to be centralised because ...costs and incentives don't match any more for more than few services that can monetize somehow.

Niel Liesmons 6mo ago

Thanks Yeah, that's why community #interop is how I see this thing scale.

With centralization happening at the niche community server level.

And the publishers being able to target stuff to multiple communities.