Problem with faked-system-time option

Amano Corunga amanoc at
Thu Jun 16 02:29:44 CEST 2011

On Tue, 14 Jun 2011 22:26:57 -0400, Robert J. Hansen wrote:

>On 6/14/11 7:46 PM, Jerome Baum wrote:
>> Err, I have to apologize if I misunderstand, not being a native
>> speaker, but based on I
>> understand that you're saying something like this?
>> "You're ignore the fact that X"
>More like "the original poster is ignoring..."  I am emphatically in
>agreement with your general point, which is that social problems demand
>social answers.

This discussion is more and more moving away from my initial problem
of not revealing the exact time when signing a piece of data in case I
just want to acknowledge its origin (or receipt) but no further
information, which is far from trying to invoke any temporal
implications with intent to defraud.

Here's a mundane real world example, a scenario where an exact
signature timestamp means a tremendous danger, in your words a social
problem arising from a technical deficit, which in my opinion has to
be addressed (to forestall suspicion I must start by mentioning that
I'm running my own small business with a few employees like Bob or
Alice ;-)):

Bob always carries a USB drive containing his mailing system with him,
during leisure time as well as at work.  Now, with a periodically
changing volume of work there's idleness from time to time at his job.
In those slack periods he has the chance to deal with his
correspondence.  And being a diligent person he's used to signing and
encrypting his mail.  But now he's toast.  One of the messages he
wrote at work was accidentally forwarded to his boss, who saw the
signature, which was made on company time.  It was up to him to
provide evidence that that timestamp was wrong, which he couldn't.  He
lost his job, his existence.

How could that threatening risk have been avoided (disregarding
kicking one's heels instead of always aiming at being productive)? Bob
wasn't granted access to the computer's system time, which furthermore
got synchronized with an NTP server in regular intervals.

And now about Alice.  Being in a simular situation she uses her own
netbook, for privacy reasons adjusts clock time back and forth with
every signature she creates, and, to indicate that the timestamp of
the signature is artificial, chooses the key's exact creation
timestamp for all her signatures as well, as by setting it to '0' or
any other time before key creation she would risk the rejection of
those potentially invalid signatures, and by using a later time she
might be accused of fraud.  Key revocation is of no relevance, as in
this respect the time of message reception by the addressee is an
adequate substitute for the lack of a valid signature timestamp. Alice
only wishes to get around the annoying system time manipulations she
has to repeat again and again.

Do you think Alice's approach is or may become an acceptable strategy,
not breaking existing conventions, and getting the chance to become
supported by kind of a '--no-signature-timestamp' GnuPG option?



More information about the Gnupg-users mailing list