Anyone know if the Noise Protocol is still being used? Last updates to anything I can find are months to years ago...
Discussion
Wtf is the noise protocol?
After a brief skim of the 65 page PDF, it looks like it’s a bag of crypto messaging solutions without a well defined problem it’s addressing…
Nostr seems to be the opposite: Very specific problem & solution is easier for devs to reason about and integrate.
Or so I guess.
it's a replacement for TLS mostly
also, yeah, holy shit after reading through bluesky docs and an implementation holy shit nostr is so tight and unified and clear
Yeah – but I was more wondering about whether it was under active development anymore or if it's abandoned.
i'm not really sure exactly how much you can add to these protocols, they are based on AEAD and replace AES schemes with chacha/salsa style CSPRNGs
i would describe them as the most recent advances in message encryption protocols, the unofficial successor of TLS
if you read some of the stuff by the cryptographer Luke Champine this work has been emerging for some time and it takes time for implementations and the general dispersion of knowledge and acceptance of the idea that these protocols are secure
and speaking of which, if you've ever seen "safecurves" they claim that secp256k1 is relatively insecure
not heard of any issues with its security with bitcoin which is a really big and obvious target if anyone has found an exploit on an elliptic curve that has been asserted to be weak based on flimsy stuff that seems to me like a mistaken association between the R and K curve types, the K curves are completely deterministic, you start with one prime number and the algorithm cannot be altered, as opposed to the one used in P256 which is aka secp256r1
i dunno, it's the stuff of conspiracy theories if you ask me, being how much spookery is involved
anyway, point is that fundamentally noise doesn't do a lot different to AES AEAD and similar related protocols, i was experimenting with using schnorr signatures as the MAC in place of the HMAC more commonly used and using a CTR mode as this allows random seeking where CBC and CFB modes require the whole message, counter mode you can start anywhere if you know the offset and the IV and secret (i'm interested in this because it allows you to do gossip style protocols where you only examine a section of the message to determine if the rest is worth decrypting)
i started building a bitcoin/nostr oriented crypto suite based on the nyquist library
https://github.com/mleku/noshtr
https://gitlab.com/yawning/nyquist
my scheme is a bit beefed up compared to the more light chacha-poly/blake based, because there is now SIMD versions of xor and sha256
gonna just update its ec library deps as it was what i started playing with in the first week of my vacation last month
whole thing is fully custom, i had already built an AES derived scheme in https://github.com/indra-labs/indra which is partly what i'm doing with this one but i figured on forking off the noise protocol proper with my own crypto suite using some known heavy duty primitives
yeah, if someone makes a SIMD BIP-340 schnorr implementation i'm quite sure this shit is fast but probably a bit heavy for hardware without SIMD, but almost everything has some kind of SIMD now, so you can go completely nuts with 256 bit sizes on everything instead of tippy toeing
yes, i've written a SIMD enabled custom counter mode encryption that uses a 32 byte nonce and secret with a per-message signing key for the MAC that uses bip-340 schnorr signatures