Oddbean

i'm torn on encrypting IoT data with the same scheme that NIP4 uses. There are all the metadata issues but the metadata leaks are likely less consequential than with social media clients. however, this doesn't mean it's ok. there are plenty of situations where it could be v. bad. it's likely that if you're doing iot stuff, you use a different nsec for social media clients, and the devices have semi transient nsecs too. i've implemented nip04 encryption in the arduino libs. giftwrapping is a good intermediate solution until we have a better NIP. i'll add nip59 support to the lib when i can.

in terms of cryptographic strength against all known attacks nope there was nothing gained by complicating things and all it has done is further undermine the usability of messaging on nostr and the protocol specified in nip-44 is overly complicated, and for the situation where you are encrypting short messages, the use of a chacha20 CSPRNG versus using AES-CBC does not substantially change the brute force vectors they both: - use a strong CSPRNG function for generating the message nonces (aes vs chacha20) - use secp256k1 x-only ECDH for deriving the secret - use a strong HMAC there was no technical or security-wise upgrade... only additional complexity idk who sold the "audit" bullshit to those who funded the creation of NIP-44 but they got robbed