Depends who you're pitching to. It's concise and might do the trick.
What do you mean with community? Social graph?
Discussion
Correct, community refers to your social graph. The Grapevine allows you to attest your “trust” in some user in some context. Every context is an action and a category. Alice endorses Bob for all actions and all categories; or Charlie to curate content (the action) for wikifreedia (the category). Maybe all wikifreedia articles; or maybe only articles in the topic of bitcoin. Or maybe I endorse you to review code (action) in the categories of BDK and LDK (the categories). Your Grapevine will calculate an influence score for every user, in every category, and will include a measure of certainty: Your Grapevine says that Bob is highly talented in some category, but only 1% certainty bc it’s based on a small number of attestations from users who are all multiple degrees of separation away from you.
Actions and categories are hierarchical, eg: the category of movies has subcategories: dramas, westerns, comedies. If my grapevine endorses you highly to be a critic (the action) of all movies (the category), then the high score for movies will be used for all of its subcategories, unless my grapevine has additional attestations saying something like: your taste in sci-fi is really bad.
And the hierarchies of actions and of categories are each represented as graphs, and each graph is curated by your Grapevine.
How do you see the UX around these "endorsements" for all these separate things? As in, do see a way to avoid having that as an extra action, and thus friction, users have to perform?
Right now I’m thinking that I need to write up a roadmap for how you as a developer can incorporate the Grapevine into your app or platform, and it needs to be broken down into steps:
Step one will be to use the follows list to create a WoT score, something that is already done by wikifreedia and a few other apps.
Step two will be to allow users to create trust attestations formatted according to the Grapevine protocol and in a context that is relevant to the app in question. Context has an action dimension and a category dimension. For example: “Alice endorses Bob to curate wikifreedia content (the action) for all topics (the category).” At this step, only one context will be utilized for the app. This is an extra action and will cause some friction as you mention, but users can ignore it if they don’t want to use it, or wait until they have a good reason to play with it, which might not be until we build out a few more steps.
Step 3 will be to pool together all available attestations data with the follows data for calculation of the WoT score.
Step 4 will be to replace the WoT score with the Influence Score, calculated according to the Grapevine protocol, basically the same way it’s already done using my proof of concept. That will be a lot of work for the developer but the user experience doesn’t have to take a hit because it’s all under the hood.
Step 5 will be to phase out follows data and use only attestations from step 2. There could be a user setting whether to use one or both sources of data, or perhaps it can be managed automatically depending on how much attestation data is available.
Step 6 will be to expand the available contexts by creating new categories. For example: “Alice endorses Bob to curate wikifreedia content in the category of technology (versus sports or entertainment or whatever).” The categories will be defined by the dev team at this point, but contributed by users and curated by the Grapevine in future steps, which will include arranging them into hierarchies.
I could go on. But I’m wondering whether steps 1-4 might work as the foundation of a hackathon.
I should add another step: your Grapevine will make a list of known users, calculate the Influence Score for all users for all contexts, store the data in a .csv or some suitable format which will be part of the Grapevine protocol, and stored as a nostr note. It can be encrypted if desired. The purpose will be to improve performance, since calculating influence scores from scratch will be slow, once the number of users gets large enough.