Oddbean new post about | logout
 In theory yes, in practice you will need more memory to keep the list of keys while processing. :) 

Definatly less possible on a phone :)  
 definitely? 
 Seriously folks. That’s what the internet is for. We’re currently building out all network mapping and score calculations server side. This will be a lot of calculating, but I’m confident we can keep the user experience crisp and responsive without melting client devices. Even mobile.  
 Yeah, I built this to do all the calculations on the front end and it’s too much. Takes too long to download the data (a few minutes) and too long to calculate all the scores (maybe 30-60 seconds). The best solution is to do them ahead of time on the back end. So that’s the next step in the project. I think what makes the most sense is for a relay (maybe a personal one, maybe a third party service) to do the calculations and store them for you, ready for quick access when they’re needed. Right now, that would mean maybe 30k or 40k pubkeys with a nonzero grapevine WoT score. 
 That would be nice, maybe a DVM?  
 We’ve had some discussion about DVMs although I think it still remains to be seen how best to use them.

Right now the plan is to build a relay kind of like purplepag.es that keeps track of all (non bot) kinds 0, 3, and 10000 notes. For free, a user can have scores calculated and stored for the entire network of 30k-40k pubkeys, and can also export subsets of the list using NIP-51 ranked from top to bottom in groups of 100 to 1000. Then you can play with the lists on other clients like listr, amethyst, coracle, etc. Then for a fee, you can set the NIP-51 lists up for automatic update every week, day, hour, whatever. 

Then the next step will be contextual lists. Example: gather together all NIP-51 lists named “Nostr Devs” and merge them together, but disregard any lists that are not vetted by the grapevine WoT score, which ensures bots won’t creep into the final product.

Now we have a grapevine-curated list of Nostr Devs who can be relied upon to curate more content. Lists of relays or nostr apps, perhaps.

Is there a role for DVMs in all this? Probably, although I’m not sure I’m yet 
familiar enough with DVMs to see how best to use them. 
 We’re working on ways to keep memory requirements trim, even when processing the entire network, by only handling subsets at a time. 🤞🏼