> But if an attacker puts his e-mail address on a key he claims to be
> mine, he won't get my mail sent to (or encrypted to) him.

If someone somehow gets that key, reads your name in the ID and relies on that 
name he might sent mail intented for you to the attacker's email address, that 
might even pretty much look like yours email address. 


>But the e-mail access control check *does* protect
>against the attack scenario where at the time of keysigning, Eve does
>*not* have access to Bob's inbox.

Yes, but the fingerprint check already protects against that, so why do we need 
another check?

>> The reason why the email address is in the user ID is for convenience >>(so 
>> everybody knows where to send emails) and makes sure keys can be >>easily 
>> found on the keyserver. Apart from that it enables user to >>distinguished 
>> between keys of persons with the same name.

>This is pretty critical in some contexts.  E-mail is a (mostly) unique,
>global identifier.  "John Smith" is not.

What do you mean with critical? 

"John Smith <[email protected]>" is quite global and quite unique, although I don't 
check the email address before signing. 
1. John tells me [email protected]. 
2. I believe him he has access to [email protected] (see former email).
3. I find keys on the server by looking for [email protected].
4. I choose "John Smith <[email protected]>", because I know his name. 
5. I make a fingerprint check on the phone (I know his voice). 
6. I sign the key. 
7. I upload the signed key to the keyserver.

If there is a clever attacker he might issue a key with the very same ID. 
People then looking for John's key will be presented the following list:

"John Smith <[email protected]>" (signed by me)
"John Smith <[email protected]>" 

If they don't know me they can simply do their own fingerprintcheck with John, 
otherwise they will take the signed key.

Thanks for your answers, I know I'm asking unorthodox questions, but I pretty 
much feel I'm right and the conventional procedure is partly unnecessary and 
thus hard to understand and difficult to use. 

Best regards, 
Jan

_______________________________________________
Gnupg-users mailing list
[email protected]
http://lists.gnupg.org/mailman/listinfo/gnupg-users

Reply via email to