3 completely bad reasons to write your own Nostr-like protocol that isn't Nostr, from https://news.ycombinator.com/item?id=36259930:

Nostr is pretty good and I was inspired by its idea. However, I didn’t like a few things and that’s why I decided to build my own protocol.

1. Nostr relies on Schnorr scheme for signing the data. While from my research, the patent on it has expired, I also read some other details on how there’s other pieces of it which are still patented. I am not a legal expert, so my understanding on this might be wrong and it might be in the clear.

2. Schnorr is fairly new and not used widely yet. It’s not even built natively in Crypto Subtle which comes built-in in all browsers.

Based on this alone, I couldn’t use Nostr. My implementation uses ECDSA P-384 keys which can be generated using the browser built in crypto subtle library:

https://developer.mozilla.org/en-US/docs/Web/API/EcKeyGenPar...

So this allows one less third party library to rely on the client side.

3. The 3rd reason (and this was one of my biggest reasons) was that Nostr runs on websockets. I didn’t like that at all. Servers are already limited on the number of sockets they can handle. Plus it seemed like unnecessary complexity when vast majority of the developers already know how to use REST api.

So instead of websockets, mine uses a simple REST api with only 2 endpoints: one for creation of records and other for searching for records based on filters.