Hacker News Reader

Comment by hipaulshi | Hacker News Reader

So.. Anyone wants to do a PGP key signing party in San Francisco? I asked on HN before but no one seems to be interested back then. I hope some meetup organizer can pick this up.

View on HN

Here's the survey to see if people are interested in the meetup. Please register if you are interested.

https://docs.google.com/forms/d/19p-OY-Q5VKtVD2BVt8QVHMCcBO9...

Unfortunately only 2 guys actually signed up so far. So I am pretty sure now PGP is not as popular as it seems in terms of verification purpose. :/

"Hello! I'm Fish!"

"Hi. I'm Jesus!"

"Wow, Jesus? Nice to meet you, can I have your public key?"

I don't trust any of you :)

This highlights a huge problem of the PGP/GPG web of trust: It mostly solves the wrong problem.

I don't care whether a Linux kernel release is signed by a guy who has government-issued ID that says "Linus Torvalds" on it. I care whether it is signed by the guy who started the whole thing more than 20 years ago.

There would be value in a modified web of trust where, when you receive an e-mail from hipaulshi, you can be certain that it's the same person you replied to on HN. Who cares what her (or his) real name is? In a sense, what you really want for most internet-centric communication is something like certificate pinning.

Of course things are different when you're talking about people who you know in person. But for those you don't have to organize key signing parties.

And even for people I know in the real world, I would be happy with a solution that just pins the very first public key it sees as long as nothing obviously suspicious is happening. This suffices for 99.9% of use cases. [0]

[0] No, this is not sufficient if you know that a government is out to get you. But if it helps to move us into an encryption-by-default setting, it is still a win for the overall ecosystem.

You need authoritative, external real name because without it, all you can conclude is that if you are being MITMed, it's by the same entity MITMed all your other connections with that person.

The alternative would be to confirm the fingerprint of the public key out of band, but lots of entities aren't already authenticated to each other (i.e. I can authenticate my best friend but not Amazon) and don't have a safe/practical way to check each other's public keys.

In practice, 99% of use cases do not have to worry about a MitM attack that is successful 100% of the time. The probability of a MitM attack remaining successful for all your interactions (and that's what would be required for it to go undetected) drops exponentially under a lot of (though admittedly not all) attack scenarios.

This is exactly the reasoning behind proposals such as TACK.

Edited to add: Keep in mind that external verification basically never happens, as much as cryptographers want it to happen. In the real world, you have a choice between no verification at all and a pinning-based verification which is weaker than the theoretical ideal, but makes MitM'ing significantly harder and much more likely to be detected for everybody at no user interface cost.

No better way than to do it yourself! Create the meet up and submit it here. I'd be up for it.

Yes, I would definitely be interested

I'm curious, how would you verify an attendee's identity?

government id/picture id checking against pgp keyserver, to make sure you are who say you are. So you have to be physically present at the party with your key and id. that's how the trust network is built.

Yes, because pieces of paper given out by men with guns are the font of all truth.

ADD: Fuck this statist cesspool.

Would you be more comfortable giving your fingerprint or DNA sample to prove your identity?

Side note: maybe there could be a way to irreversibly hash one's fingerprint or DNA sample into an electronic signature

Any hashing function would do it, but you leave your fingerprints and DNA anywhere. Would be easy to get access to it and copy.

that's fine, its a public key :)

I believe proving identity can only be done using a private key.

The problem with that, of course, is that someone's DNA or fingerprint isn't a secret. There's no reason why I couldn't take your fingerprint, embed it into a signature and claim to be you.

And this is part of why authentication and identity are very difficult things to do right, mostly because very few people have thought about what it is they're verifying.

If I publish a public key and say it belongs to me, 'Bob Smith', the only practical use that has is that you can verify that a future message signed by 'Bob Smith' was signed by someone with access to the same private key as the guy who originally published the public key. Any assumption about who 'Bob Smith' actually is, and who that corresponds to in the real world (what other identities do they assert), and also that 'Bob Smith' is a single entity, are simply assumptions.

It's impossible to pin a human down to a single, guaranteed verifiable, non impersonatable and non revocable identity. 'Documents issued by men with guns' isn't foolproof, but we use it as a trust anchor mostly because everyone else does, and we don't have much alternative.

The thing about a fingerprint or even a dna sample, in this use case is that you send your signature ahead of time and verify yourself phyisically at the party. Does your fingerprint hash match your fingerprint? That is more difficult (but not impossible) to spoof. At the end of the day, this discussion stems from notion that government ids are unreliable as a means of verification. Granted, but for what we are trying to achieve in practice is preservation of privacy and data. I was trying to point out that identification of a person - true identification - can be at conflict with our ultimate goal of privacy, since we have to give up a piece of data to prove we are who we say we are, and contemplated finding a way to technically make it happen (in practice) without sacrificing PII.

I see what you mean - you could probably do something like publish your public key and then publish a signed copy of your fingerprint hash. Anyone else could do the same thing, but an imposter wouldn't be able to convincingly produce your fingerprint on demand when physically present. At least, not without a lot of funding and cleverness.

So you are tying the key to fingerprints or other biometric, but how do you tie that to an identity e.g. a name?

Falsehoods programmers believe about DNA:

    1. A person has only one genetic code in their body.
    2. A person's genetic code never changes.

etc. In the spirit of "falsehoods programmers believe about names" and "... about addresses".

Those are solvable problems though. Or are you saying 23andme doesn't work?

23andMe is not going to provide you with an "authoritative" DNA "fingerprint" because how would they tell that this genetic sequence in this chromosome from this cell is legit but this contradictory sequence in this other cell is different.

Obviously noone has the capability to do this, so they present you instead with "most likely" results. Not the level of proof I'd be looking for in a court of law.

No, they aren't. I did not say "A person's tested genetic code never changes", I said exactly what I meant, which is that genetic codes can be changed. Retroviruses do it. Further, on a grand scale, chimeras [1] exist, with human chimeras often entirely unaware of it (while my priors would suggest I am not one, I can't even come close to proving that to you with the rather meagre medical tests I've had done on me), and on a smaller scale, small mutations in cells can easily propagate out into the body over time, even before we ignore the matter of cancer.

Most of the time, this doesn't matter, but if we're talking about taking hashes for security, suddenly it does.

3. A person's DNA is unique

I'm thinking identical twins. Got anything for #4?

Odd. I don't recall anyone at the DMV forcing my license on me at gunpoint.

I think you mean indifferent bureaucrats without guns.

See "The Keysigning Party HOWTO":

http://www.cryptnet.net/fdp/crypto/keysigning_party/en/keysi...

I am very interested.

Reply on news.ycombinator.com