Nostr Web Client

I started developing on nostr because I built a little C library and a signer in a single afternoon. That's all I needed to own my identity and participate in the network via websocket. I didn't need to fight or route around my isp, didnt need to create half a dozen DNS records. I didn't have to cloud proxy to hide my external IP, I was able to open a TLS connection and hope that most other users also receive my note over ssl, I didn't have to spend all week configuring an smtp server, and I didn't have to pray that the remote server would accept my mail and display it to the user.

SMTP is a horrible (and heavily censored) protocol that needs to go. We can't keep stacking junk on top of junk and call it more secure anymore. It has to go. We have better technology. I'm not saying nostr is the silver bullet in coms, but its no where near what SMTP is.

Luxferre 1y ago

"Junk on top of junk" is exactly what JSON on top of WebSockets on top of HTTPS is. At least in terms of wasted CPU cycles and energy watts.

SMTP is just as censored as the underlying infrastructure. Run it e.g. on the Tor layer and you won't have to deal with DNS, DMARC etc. I only gave it as an example of an already established transport that everyone can build upon (and it's not the protocol's fault it has been heavily misused by big players on the clearnet). And more importantly, it's plaintext, human-readable (except MIME extensions for attachments) and can be debugged without all those fancy tools. And this example was given because, from what can be seen from outside, Nostr tries to reinvent the same functionality + mailing list servers' one. Just wrapping everything into more opaque binary layers than it should.

Reply to this note

Please Login to reply.

Discussion

ChipTuner 1y ago

> Run it e.g. on the Tor layer and you won't have to deal with DNS, DMARC etc.

Right, but this limits access for the average user, HTTP connections are far less censored than SMTP traffic in reality at this point in time.

> and it's not the protocol's fault it has been heavily misused by big players on the clearnet.

Solid point. I failed to accept SMTP as a bare protocol and reusing it for client-server, but if we looked at it that way, lets just skip to HTTPS.

> it's plaintext, human-readable (except MIME extensions for attachments) and can be debugged without all those fancy tools

Which is what nostr:npub1wqfzz2p880wq0tumuae9lfwyhs8uz35xd0kr34zrvrwyh3kvrzuskcqsyn has been brainstorming. Get the benefits of HTTP upgrades and decades of heavy development for client-server, server-server and so on.

I'm not married to websockets, but I do think nostr fails in client performance, notifications and so on, without persistent connections, but we need clients of all types to be able to communicate directly on the network, not their own servers, over some other messy protocol.

>Nostr tries to reinvent the same functionality

This is true all over the nostr protocol. However keep in mind, a lot, if not most of us, dislike/distrust "mainstream" big-tech/government funded protocols. I'm not saying its right, or efficient, but then what is Bitcoin?

Luxferre 1y ago

I still doubt you fully got the point, but OK. Let's consider HTTP as a baseline, I have three questions then:

1) Why JSON? HTTP itself offers a wonderful way to transfer metadata, it's called headers. They require much less CPU cycles to parse and shape. They are naturally separated from the request/response body. All specific headers can start with X-Nostr- to not interfere with anything current. Even the digital signature can be put there if it's short enough.

2) Why TLS? Aren't signatures themselves enough to prevent tampering with? Isn't current centralized PKI prone to censorship as well? On top of that, TLS over e.g. Tor isn't necessary at all.

3) Why a single (and not very popular) signature algorithm instead of the wide choice offered by OpenPGP, where we can reuse existing libraries/tools (that have been working all this time even on DOS) for both signing and, if necessary, asymmetrical note encryption (to mitigate number 2 completely)?

P.S. "Just because you're unique, doesn't mean you're useful" Being different from mainstream is not enough to automagically guarantee yo're doing things right.

ChipTuner 1y ago

To be clear I don't wish to bicker, I vowed I would not do that here. I cannot argue for the entire existence of the current state of the protocol. And I cannot refute plenty of your suggestions.

What I can say is, I'd like to see the suggestions you make. Have you make any contributions to the nips repo?

Luxferre 1y ago

Which repo? The one on GitHub? I ceased any activity there a long time ago.

Why are nips on a repo hosted by one of the most censorship-loving companies in the world? And not e.g. on a wiki hosted independently.

I can formalize my proposals into a document when I have time and put it up somewhere if anyone is interested though.

ChipTuner 1y ago

Directly on nostr if you want. If you think your Ideas are worth implementing, like it or not those are the tools. I personally dislike GitHub and only use it as a "backup", but in the real world it's where the most amount of people collaborate on oss code. That's the reality. I encourage you to lead by example, build something and show it off.

https://gitworkshop.dev/repo/nips