Hey @OgLog, great article.

First, I am all for using spontaneous user interactions as input for discovery algorithms, as opposed to NIP77 where the user is asked for a trust judgment that will likely require bootstrapping a new type of user interaction.

Second, I like zaps because they carry a value signal (assuming the sender is not colluding with the recipient).

I think your proposed algorithm is part of a more general class of algorithms that I've described in my
Navigating the Social Graph https://pippellia.com/pippellia/Social+Graph/Navigating+the+social+graph

Here is how I see it:

- The input is some social graph whose edges represent some form of social relationship, such as who follows who, or, as in your case, who zapped who and how much.

- some propagation rule is applied (in your case, it's direct propagation)

- calculate a certain set of candidates

- apply a sorting rule to sort the candidates (this can be pagerank, local pagerank, trustrank, zap weight...) 

 Hey @pippellia! 

I agree that a key here is spontaneous user action. Zaps are a great candidate, because it doesn't feel like work; it feels like a thank you, and we already see people doing it. And within a network with zap trails as a primary content curation technique, it would feel like an upvote as well (though local, not global) . This is why I call the zap network a "self-cultivating" neural net - we don't have to solve the problem of "who creates the graph", because users are already clearly primed to engage in the actions that generate the data the algorithm needs.

By the way, I resonate a lot with your writing on "There is no global", and love the framing that the idea of "global truth" is as dead today as God was dead in the 19th century.

The last part of your piece is a bit too math-heavy for me to follow easily, but your bullet pointed description fits.