GPG PGP S/Mime vulnerability

[Clay Haapala]

This makes so much common sense to me that I must write in support of it
-- a digital signature is akin to a public notary's signature -- I'm not
saying equal to that, in today's environment, but akin to it.  As I
write this, I realize I don't know how much non US-citizens (or
non-Minnesotans, for that matter) know about being a notary public... 
I'm sure there are many local governmental equivalents of public
witnesses to the signing of private documents.

Anyway, Ben's point is well taken.  Sign a document iff it has
information in it that you want to "own".  I can't think of a reason
anyone would want such information to be ambiguous for any of the
objective content.

As email systems work today, to casually repackage a signed email inside
your own envelope and remail it would cause SMTP headers to be created
that are likely out-of-time with the timestamp incorporated with the
digital signature, but that doesn't preclude some clever spoofing of
headers through MTAs.

"Ben . Wise" wrote:
> 
> Folks,
> 
> The solution to this is the same as with physical
> signatures: do not sign ambiguous documents!
> 
> You would never sign a document saying "I sell my house to you",
> you would only sign something like "I, Joe Smith, sell my
> house, 123 Main St., Nowhere PA, to Jane Doe for $150,000,
> effective 1/1/2001".
> 
> BTW, no one with any common sense would trust such an
> ambiguous document as "You're FIRED!", even if it was
> signed. There is no evidence at all that it was addressed
> to the recipient - it's just too ambiguous.
> 
> Similary with the "I sell my house to you": no court would
> even consider it - it's just too ambiguous.
> 
> With the best technology, we still need common sense
> on both ends of the communication channel.
> 
> -----Original Message-----
> From: Guy Van Sanden
> To: GnuPG Users
> Sent: 8/3/2001 6:47 AM
> Subject: GPG PGP S/Mime vulnerability
> 
> I've read through Don Davis' whitepaper about the disadvanteges of the
> current sign (and encrypt) features in all common standards to do so.
> 
> His basic reasoning (and I've tried it, it works!) is:
> I send a signed message to someone stating "you're fired".  He gets
> angry and decides to get even with another collegue...
> Using SMPT he puts my address in the from header, then pastes the
> entire source of my signed message to him in the body (including the
> signatures), and sends it of to someone else...
> 
> That last person opens a message, which he thinks comes from me, and
> trusts the contents because the signature is verified!
> 
> More info is over here:
> http://world.std.com/~dtd/sign_encrypt/sign_encrypt7.ps
> 
> Kind regards
> 
> Guy

-- 
Clay Haapala clay@haapi.mn.org
"Climate is what you expect.  Weather is what you get."
   -- from the sayings of Lazarus Long