What can I expect at a Key Signing party?

[Jean-David Beyer]

mike ledoux wrote (in part):
> On Sun, 14 Oct 2001, Jean-David Beyer wrote (in part):
> 
> >If I want to get people to sign my public key, it seems that the best
> >way would be to go to a Key Signing party. What do I bring? What do I
> >get back.
>
[snip]

> What to bring:
> 
>     Photo-ID (passport, drivers licence, etc.).  Too much ID is better
>     than too little, especially if your ID pictures are bad like mine.
> 
>     Several copies of the output of 'gpg --fingerprint your_key_id'.
>     One copy for each person expected at the party, plus a few extras.
> 
>     Information on how to get your public key.  Usually the best
>     way is to put it on one of the public keyservers (pgp.mit.edu
>     is a good one), and let people know where they can get it.
>     Floppy disks are bad--do you know what type of computer each
>     person at the party uses?  Are they likely to trust you to not
>     have some sort of virus on the floppy?
> 
>     *NO COMPUTER*.  It is always a bad idea to use your private key
>     on someone elses hardware (can you sure they don't have key/disk
>     loggers?), so having a computer at the party gains you nothing,
>     and loses quite a bit.

I could not imagine bringing a floppy with my private key anywhere
(other than to my safe deposit box) unless it were someone that I REALLY
TRUSTED; i.e., perhaps one of my sisters. (But since she runs Windows,
while I trust her, I might not trust her machine.)
> 
> The procedure I've used in the past is something like:
> 
>     Everyone receives a copy of everyone elses 'gpg --fingerprint'
>     output, then verifies that the name on the paper matches the
>     provided identification, and that the identification is sufficient
>     to identify the person.  After everyone has verified
>     identification, each individual reads off their name, keyid,
>     and fingerprint, to catch any errors.

If you examine a person's ID and compare it with his printed 'gpg
--fingerprint' and perhaps sign that with a pen, why is it necessary to
read off names, keyids, and fingerprints? What errors would be caught?
> 
>     Everyone leaves the party with a bunch of keyid/fingerprint
>     combos that they've verified the names on.
> 
> After the party, everyone goes home and signs keys according to
> their personal signing policy (everyone has one, but few think about
> it that way).  It is very much a matter of personal preference which
> keys you sign.

What does signing (as contrasted to trusting) mean? It seemed to me that
signing implies that you are convinced the public key whose fingerprint
you have really and truely belongs to the person whose ID you have
checked. I do see a problem here that I do not know how to solve: I know
the person who handed me the slip of paper with the gpg fingerprint on
it is the one whose name is on the passport (or whatever it was) because
the pictures match, say. It is likely, but not certain that the
<ABUser@isp.com> is truely that of the guy on the passport because it
says
Anthony Brian User on the passport. So it might really be his. But it
might just as well belong to Albert Barnaby User and Anthony Brian is
trying to hijack his signature. Perhaps nothing can be done about this.
Now if I personally knew Anthony, perhaps I could be sure that he would
not be creating a signing key to impersonate someone else, but at a Key
Signing, it is quite likely that I know none of the other people.
> 
> If you want *my* signature on your key, we will have to exchange
> encrypted mail using each email address you want signed, so I can
> verify that the Jean-David Beyer I met actually receives mail at the
> address jdbeyer@exit109.com.

So let us say I meet you at a Key Signing and hand you a slip of paper
that says:

GnuPG and PGP key information for Jean-David Beyer.

pub  1024D/9A2FC99A 2001-09-22 Jean-David Beyer (Institute for
Regimented Whimsey) <jdbeyer@jdbeyer.exit109.com>
             Fingerprint: CC3C 3A4D D593 3491 DA41  8FE9 3EDB B65E 9A2F
C99A

The public key for this can be obtained at the following sites,
among others:

http://www.keyserver.net/en/
http://pgpkeys.mit.edu:11371/
http://www.isternet.sk/pgp/
http://certserver.pgp.com/index.html

and perhaps you hand me a similar slip of paper.

Now if I send you an encrypted e-mail using the public I got for you by
looking up mwl+gnupg@alumni.unh.edu on a keyserver, someone will get it;
presumably you, and only you should be able to read it. Is what protects
you from erroneously signing my public key the fact that the email came
to you from jdbeyer@jdbeyer.exit109.com? Certainly the fact that I have
your public key gives you no assurance. Can you really trust the From:
or Return Path: headers? I guess a man in the middle would not be
expected to do this on the off-chance of catching this exchange. Is the
unlikelyhood of this sufficient to induce you to sign my key?

And symmetricly if you want me to sign yours, of course.
> 
> Once you've signed someone's public key, it is good form to email them
> an ASCII-armored copy with your signature so they can distribute it
> as they wish.  It is considered bad form to push someone else's key to
> the keyservers.

It would seem the polite thing to do. I thought it was not possible to
update someone's public key other than your own. I am new at this and
could be wrong about this.
> 
> >So how are these issues usually handled?
> 
> Every group does it differently, but the procedure describe above has
> worked well for me in the past.
> 
Thanks.

-- 
 .~.  Jean-David Beyer           Registered Linux User 85642.
 /V\                             Registered Machine    73926.
/( )\ Shrewsbury, New Jersey     http://counter.li.org 
^^-^^ 2:20pm up 24 days, 21:15, 3 users, load average: 3.08, 3.06, 3.00