hodlbod on Nostr: I'm not exactly following the algorithm as described in that blog post, but it seems ...

I'm not exactly following the algorithm as described in that blog post, but it seems like "influence" adds two new variables to Pip's "direct propagation" algorithm for calculating wot: weight and context? We've already talked a good amount about context, and how hard a problem it will be to find vocabularies that make sense in many situations, but putting that aside for the moment, would a reasonable calculation of influence look something like the below?

- I trust Alice at a weight of 0.9 to recommend content related to `context`. Everyone else I weight at 0.5 by default.
- Alice has recommended Bob at 0.9 in `context` and Charles at 0.1. I have 2 other recommendations in my social graph, at 0.5 for both Bob and Charles.
- Scaling recommendations via multiplication gives me .81 + .25 + .25, a total `input` of 1.31 for Bob, and .09 + .25 + .25, a total `input` of .59 for Charles.
- This then gets scaled based on `alpha` etc to get `certainty`.

This is all fine and valuable, if you can get users to provide good data about how they weight recommendations, but it doesn't really solve the wot bootstrapping problem, because someone (hopefully Alice) still has to discover Bob in the first place. That depends on a different mechanism, either the Transposition WoT algorithm, or other things like content promotion, proof of work keys, etc.