trust your corporation for keyowner identification?
Paul R. Ramer
free10pro at gmail.com
Tue Nov 5 09:40:11 CET 2013
Leo Gaspard <ekleog at gmail.com> wrote:
>> You are right. Decryption is sufficient to demonstrate control of
>the private key, because if he can decrypt, he can also sign. What I
>said, "decrypt and sign," was redundant.
>Well... I still do not understand why decryption is sufficient to
>control of the private key and not adding a UID (note I'm talking about
>UID's, not unsigned ones, of course).
I don't know how I can explain it any better than I have. I think you are confusing assertion with verification. Unless you can differentiate between the two in this case, I don't think you will see what I am talking about.
The process of certifying someone else's key involves the following:
(1) He claims that a key with n number of UIDs and fingerprint of x is his key.
(2) You verify his identity and compare it with the information in his UID(s).
(3) You send encrypted emails to each email address in his UIDs.
(4) He replies with the decrypted messages that you sent.
(5) If all went well, you certify his key.
In the case that we have been discussing, it is assumed that all of those steps have been followed for the first key. With the second key, only the first two steps, and part of the third, have been followed. And now you are assuming that the second key (being independent from the first) is valid without following through all of those steps simply because you have validated the first key which, according to what you are suggesting, was used to sign the second key.
Simply, assertion is not verification; probability is not certainty. If you didn't verify control of the key (wasn't that the point behind signing someone's key?), then your signature on his key will be baseless.
If on the other hand we were talking about a new UID on the first key, we would just need to verify control of the email address in the new UID if the UID contains the same name as the other UIDs.
No one is going to stop you from signing someone's key if that person sends you an email saying, "Hey, would you sign my other key?" But if I know you sign any keys without following the same thorough verification process each time, don't expect me to assign ownertrust to you.
I guess all I can say is that one should have a key signing policy to let others know how he verifies keys.
There. I said it all over again, just differently (and a whole lot more).
More information about the Gnupg-users