straycat on Nostr: Personalized trust metrics definitely fits into the “no solutions, only ...

Personalized trust metrics definitely fits into the “no solutions, only tradeoffs” category. If every metric is always updated at request time and never stale, we limit not only number of requests per unit time but also the complexity of scores. Suppose I want to generate a baseline “real user, not a bot” score using follows mutes, reports, zaps, and content interactions. I want to use that to curate a list of nostr devs. And I want the list of nostr devs to curate a list of NIPs. That is a daisy-chain of 3 scores. Suppose I want to chain together more than 3 scores using a complex configuration? It becomes impractical to recalculate every score from scratch every time it is used. Somewhere along the way, personalized trust metrics will have to be cached, with the frequency of updates based on priority and availability of resources.