What I'm saying is there's 2 types of clients: pure clients and hybrids.

Pure clients implement Nostr perfectly. Users have total sovereignty. Being totally sovereign comes at a cost. Users have to make decisions about what relays to use, and bear some responsibility for their own UX based on the choices they make. They also have to commit to storing all that data locally, using up bandwidth and battery life, etc.

Hybrid clients rely on some server operated by the developer of the app to serve the data. This means they control the UX. But this way it can always be fast, because there is some middleman who is always online, always collecting and processing data to serve it to you instantly. The downside is they can censor. But as long as you control your key, you can move to a different client. Therefore, the most important aspect of Nostr is that users control their key, and both types of clients need to exist to create a healthy ecosystem.

Reply to this note

Please Login to reply.

Discussion

> Users have to make decisions about what relays to use, and bear some responsibility for their own UX based on the choices they make

This is the current state for most nostr apps.

There is both under the hood and UI/UX work to be done here. There can be a “it just works” mode, and ownership for those who want it.

https://github.com/damus-io/damus/issues/423

For a pure client, outbox is good, it's a big improvement. But what servers can do is completely different. Outbox can tell you where something is located so you can try to get it JIT. That doesn't mean it won't be slow or even work at all. Servers can try all day long to get it, even while you're phone is powered off. All it needs is a moment in the day where the relay is working, so it's available for you instantly when you go online.

Is it possible to have a pure client with outbox but just for the text? for profile images, images, videos..., use the server? So I would be sure that the client doesn't miss notes from the server.

Agreed.

Do you think there is a tradeoff with indexers on the censorship resistance axis?