How would you build a p2p twitter clone with GunDB?

Question

GunDB is supposed to support peer-to-peer data access, so I'm trying to better understand how this would work. If I were to build a twitter clone, what would the high-level architecture look like if I wanted each user to store their own tweets on their own server?

Answer 1

Answer by the author of GunDB:

Every tweet is cryptographically signed by the user (in fact, with SEA, all data by that user is signed automatically, so you don't have to worry/do anything!) which means that no matter where it is stored (by the user, by a server, by another user), it cannot be tampered with.

By default with GUN (and you could modify this if you wanted, but you'd be adding a lot of unnecessary complexity, which I don't recommend), the tweets are stored by whoever subscribes to that data. That would mean: (A) the user who made the tweet would store it (B) a server peer, which is subscribed to everything, stores it, (C) and friends/viewers/followers/audience who reads the tweet also stores it.

Realistically, most users probably are using a browser to access their app, so you wouldn't want to assume this is reliable - but you can Electron-ify (or something similar) your app so that users can install it on their desktop, in which case, they would be their own server. Then you (or other users) can also deploy the app to AWS/Heroku/DigitalOcean/whatever and ALSO store data as a backup there (like in the case of B, if you add your S3 credentials, it backs data up to S3 - ideally this would be IPFS instead, or similar), etc.