Is that really the architecture that makes sense? Would it not be more efficient for trust providers just to serve them up? Sorry if I am missing context here. What's the motivation for making everyone store a copy of the whole graph?

Reply to this note

Please Login to reply.

Discussion

It is more efficient to get trust providers to offer them, but it also requires trust in the provider. If the app can compute all of, you wouldn't need anything else.

Isn't there something that can be done cryptographically to make the trust provider's results verifiable (just as trustworthy as if it came out of your own copy of the graph at least)

Or is the idea that you can cache your scores locally and verify less frequently?