Hacker News Reader

Comment by Retr0id | Hacker News Reader

I'm not sure why you're getting downvoted here. In any other context, "paste in the private key that your identity is irrevocably bound to" would get you laughed out of the room. It seems to be the default nostr UX nonetheless, and that's worthy of note.

I am aware of efforts to improve on this though, for example https://snort.social/e/note1crl44xk24yc2ym5xlyyfjdeumxueyguz... - essentially the equivalent of a custodial wallet, if I'm understanding correctly.

View on HN

He’s getting downvoted because that’s a solved problem in the wider crypto-space. To log in you sign a message using your hardware wallet.

Nostr hasn’t grown up to that yet, and pointing out that pasting private keys is a bad idea is fair game.

But implying that public private key cryptography can’t be used to log into web apps is just silly given that value worth billions is being moved around daily using such web apps.

There's a reason why social logins, magic links and SMS login codes are so ubiquitous now - end users want 0 friction getting into their accounts. This increases friction beyond what is currently common practice. No, the public doesn't want nor care about using a key to log in.

Although i have heard of SQRL [1] which purports to be a frictionless way of securely logging in with public private key encryption. I've never seen it discussed here though so I'm unsure if there's significant downside

[1]: https://www.grc.com/sqrl/sqrl.htm

This sounds a lot like passkeys https://developers.google.com/identity/passkeys - I'd be interested to see a more thorough comparison of the two.

Both of these do a challenge-response style authentication with a particular website, and wouldn't really work as part of a decentralized system.

You could use the same signing key ordinarily used to sign authentication challenges to sign nostr notes, but then you're back to square 1 really.

Edit: slightly better passkey info here https://developers.yubico.com/Passkeys/How_passkeys_work.htm...

SQRL is indeed pretty similar to passkeys. It was just invented before passkeys (before WebAuthn even) and designed to work without requiring any new web standards or changes to the existing web browsers at the time.

That greater compatibility came with some UX trade-offs though. Now that passkeys exist and are widely supported by web browsers there's really no need for SQRL anymore; passkeys are a far more polished version of the same concept.

>end users want 0 friction getting into their accounts.

Why not offer both, the easy one for normal users and an option to use the more secure option.

Why not have the secure option also be easy so normal users will use it?

Does public key necessarily have to be high friction?

Couldn't there be a few-step UI similar to "Sign in with Google/Facebook/etc"?

For People who are already familiar with metamask, "Sign in with Ethereum" is so easy that makes people wish it was universal. It is even easy to generate new identities on demand: generate a new address/public key is literally a one-button operation.

The only thing that bugs me about is that is actually all your identities are still tied to same master passphrase, so if that gets compromised all of your identities get revealed.

You should try logging into a “web3” app to see what I’m talking about. No need to buy crypto or anything, of course. Just for the science.

A short guide, assuming a mobile device:

1. Install MetaMask, next, next, finish to create a wallet.

2. Open app.uniswap.org on your favorite browser. See if you can figure out how to ‘connect’.

You can of course use the MetaMask built-in browser, but that’d be cheating.

I’ve actually seen this used for non-crypto sites. I think… it was fetlife?

If you already have MetaMask/Phantom extensions installed it’s easier than email verification.

I am glad Zach posted here also, I think nostr devs need some reality pull from "non-crypto normies". To be clear it is (hopefully) becoming more common practice to use a nip-07 web-extension for signing (hardware or software) or other methods such as NIP-46. To be clear though, a user on nostr will HAVE to interact with cryptography in some way as every message on the network is signed.

Im working on https://github.com/VnUgE/NVault as an option for more paranoid users that want a self hosted networked approach. But there are others listed here https://github.com/nostr-protocol/nips/blob/master/07.md

Finally, I wholly dislike the practice of offering an option of entering an nsec. Use a signing extension!

the wider crypto space still has no empathy for normies. if you can't somehow recover or rotate your password, it's a nonstarter by virtue of being a worse UX

Reply on news.ycombinator.com