Re: Access to Storage and Communication Keys

[Marc Horowitz <marc@cygnus.com>]

Bill Stewart <stewarts@ix.netcom.com> writes:

>> Having argued that point vociferously in the past, I'm now going to
>> waffle on the issue - while the business need is for access to
>> stored data, this may often include stored messages received from
>> a communication system in encrypted form.  Either the User Interface
>> needs to make it convenient to store the decrypted message,
>> or else the user will store the message in encrypted form -
>> which means there may be a business need for Proper Authority Access later.

To me, mail encryption is not communications encryption.  The mail
message is encrypted, just like a file might be.  Then those encrypted
bits are sent over the net.  It is precisely because I have access to
the ciphertext as a separate entity that this is not communications
encryption.

This is in contrast to ssh, kerberized telnet, IPsec, etc., where once
the communications has happened, I either have the cleartext bits
(example: scp), or nothing but a memory in my head (example: telnet).
In this situation, private escrow of keys is useless, unless I'm also
escrowing the ciphertext.  Nobody I know archives their cyphertext
data flows.  Anybody know of a contradiction?

The *only* reason to escrow communications keys is to spy on people;
there is never an opportunity for data loss here.

Note that this also means that private key recovery (intra-corporate,
for example) is consistent with perfect forward secrecy, since the
former is never useful for communications, and the latter only is.

This doesn't fix the potential problems with email, but it does let
you continue to argue vociferously and with a clear conscience against
communications key escrow in any form.

		Marc