Mnemonic seed vs nsec

[alexgleason]

I think mnemonic seed phrases are a better UX than nsecs. They are much easier to remember, and I think they're less intimidating to users. On the other hand, most clients don't use them, so they have less compatibility. What do you think?

[vitorpamplona]

We don't use them because NIP-06 words are used as a seed to generate the nsec. It's impossible to get words out of an existing nsec. And that means everyone must create a new private key and migrate all of their stuff just to get words. It sucks.

It would be nice if we could all agree in a function that maps our nsec into a sequence of words. Then we can easily accept words for our current users.

[alexgleason]

Oh! That's a good point.

I'm not sure it's right to pretend an nsec is the root key when it's supposed to be at a specific derivation path (NIP-06).

Instead we could focus on letting new users create mnemonic seeds.

[fiatjaf]

We could come up with a way to turn nsecs into mnemonic seeds.

[staab]

Where's NVK when you need him

[fiatjaf]

Why do you say that?

[staab]

He has a vendetta against QR codes, because even if you're in a bunker underground there's a satellite that has line-of-sight to your phone screen

[fiatjaf]

Just ignore that.

[vitorpamplona]

Why don't we just use the BIP-39 wordlist and encode the 32 byte of the nsec in 24 words with a given prefix?

1. Convert hex/nsec into byte array
2. Split byte array in 24 sections of 11 bits (pad with zeros in the last section).
3. Convert each 11-bit section into a positive number
4. The positive number maps to the line in the BIP-39 word list.

[fiatjaf]

Remember that mnemonic phrases in English make no sense to people that don't speak English.