Web of trust is a consensus problem. Equal in significance to the Byzantine generals problem. But this consensus problem concerns language. So I call it the problem of how to reach linguistic consensus.

This is the problem: how do we get a decentralized community to build our digital tools of communication without reliance on centralized authorities, like w3c committees and repo managers?

If we can solve the problem of linguistic consensus over the spoken word, we should be able to do it in the digital world too.

Solution: the tapestry protocol. The concept graph + the grapevine. An approximation of how we solve the problem in normal life, ported over into the digital realm.

Scaling Dunbar’s number up from 150 to 8 billion.

Reply to this note

Please Login to reply.

Discussion

Interesting. Was Listening to nostr:npub1h8nk2346qezka5cpm8jjh3yl5j88pf4ly2ptu7s6uu55wcfqy0wq36rpev AI Unchained with a guest, HopeTS. He's trying to build what I would describe as an AI-powered truth-filter to help us disseminate fact from fiction by checking politicians' track record of public actions against the bullshit they're deciding to spew in the moment.

Thought there might be overlap.

https://www.patreon.com/hopets

I can envision the utility of the Grapevine would be that it would help people discover the AI-powered tool that you mention. If someone in my grapevine has used the tool, that person can attest to its utility, and can also assign it to relevant categories, like the category of “truth filters”. The categories are themselves managed by my Grapevine. So when I go looking for a tool in that category, I’ll be able to find it.

So the Grapevine is a sorta open-source app store? Ish?

💯 open source. But it needs more work.

I have an open source desktop nostr app which is linked in my bio. Proof of concept that your grapevine can curate the items on a list. But it crashes and needs to be rebuilt with better UI, which is why I have an article with screenshots to show what it does. Curation of a category tree to go along with each list is on the roadmap. I think my existing proof of concept is enough to show that this can also be done. But it’s more than I can code by myself. So, I’m going to need a team to make it happen.

Here’s my AI for deciding if someone is full of shit:

If politician:

Return true

Copyright DO NOT STEAL

Lmao hilarious, but not useful