[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GPG fingerprints

Tim Haynes wrote/napisał[a]/schrieb:
> Wade Richards <wrichard@direct.ca> writes:
> > >            A five minute explanation of the principle of a
> > >man-in-the-middle attack, followed by a swift bat upside the head with a
> > >copy of "Applied Cryptography" seemed to do the trick, and he sheepishly
> > >removed it.
> > 
> > I think that many people put their fingerprint in their e-mail signature
> > to exploit the Internet's archiving capability. If I e-mail you my public
> > key, you should not pay attention to the fingerprint in the signature of
> > that e-mail. However, you can go to dejanews.com, or the debian mailing
> > list archives, or your own "saved mail" folder, and notice that every
> > single message from me has the same GPG fingerprint, even the messages
> > that are months or years old. From that, you can develop a degree of
> > trust.
> Yes. A zero-trust sense of trust.
> The whole point of having a fingerprint is to be able to compare it out of
> band - eg you send me your public key, I phone you back and you have to dig
> out the fingerprint which I compare from the public key, which is totally
> defeated if someone else can dig it out of deja/google!


Anyone who gets hold of a public key can check what fingerprint it has.
There are public keyservers. There are public keys on the w3. Key
fingerprint never was meant to be a secret.

> If you want to develop a sense of trust, then the most trust you can have
> is that `this poster' is the same as `that poster', because their messages
> both validate against the same key ID (*not* fingerprint).
> Unless I'm well mistaken, of course... But I'd never trust a key whose
> fingerprint had turned up in public before.

I believe you are mistaken. Publishing fingerprint is a (weak) way to defeat
MITM attacks. If someone constattly uses a key with a known fingerprint
sudden change of fingerprint may may suggest MITM. Note: your method of
comparing a fingerprint is weak. Fingerprint comaprition is a two way
protocol. If Bob is to sign Alice's key he should read first group of
fingerprint, then Alice should read the second, then Bob the third, etc.
This ensures at least that Bob and Alice are talking about the same public

C _-=-_ H| Janusz A. Urbanowicz | ALEX3-RIPE | SF-F Framling |         |   *  	
 ; (_O : +-------------------------------------------------------------+ --+~|	
 ! &~) ? | Płynąć chcę na Wschód, za Suez, gdzie jest dobrem każde zło | l_|/	
A ~-=-~ O| Gdzie przykazań brak dziesięciu, a pić można aż po dno;     |   |   

Reply to: