Planning on releasing it as a service first :)
Experimenting with localized rankings and implicit trust. Real time for these algorithms is too costly, so I am aiming for near-real time (30m-1h) + real time postprocessing
$5000/1M requests is absurd though
> Experimenting with localized rankings and implicit trust
which algos?
> $5000/1M requests is absurd though
I don't think so. For example with RankProfiles you can rank up-to 1000 pubkeys in a single request.
Is it not 1 credit per pubkey?
no, 1 credit per request if using Global Pagerank, 10 if using Personalized Pagerank.
The pricing is probably going to change btw
I would like to experiment building a different kind of client using only nostr.land and vertex (and my vibe coding stack)
If using NL, you will not need Vertex I think, as there are filters already in place
I don't mean for spam but for sorting profiles by rank, etc.
Do you think NL could realistically be used as a single relay while finding most events? Right now clients with outbox don't give a shit and connect to a million relays anyway
I want a proof of concept of a very fast and coherent (paid) client, something like the idea of Primal but without a caching server
Makes sense. Though I could probably expose some of the internal trust metrics as well
For finding most events, after NFDB2.1, which is coming very soon.
Interesting, keep me posted
I might drop everything I am doing to do a beta version of NG2
Personally not in a rush
I want the cool graphs.
Did this 2 years ago and it was interesting
It was only 10K pubkeys.
I used a 150-dimensional embedding back then. Probably need way more now
Are you going to deliver the graphs as TikZ code or something, like we discussed?
Alexandria has a search bar, you know. We can do NSL there, too.
You'd get more money, of you got our relays going. Just saying.
