From: "Phillip M. Hallam-Baker" <hallam@ai.mit.edu>
Date: Tue, 22 Jul 1997 06:05:38 +0800
To: "cypherpunks" <cypherpunks@toad.com>
Subject: Re: Verisign gets export approval
Message-ID: <199707212144.RAA15989@life.ai.mit.edu>
MIME-Version: 1.0
Content-Type: text/plain



I think you have to see the whole gameplan to understand what is going
on.

The issue is the "exportable" versions of Navigator and Internet
Explorer. Presumably Microsoft wants to release as strong a general
crypto solution as they can.

So if you are a merchant in the US selling to a European customer you
want to use 128bit encryption if it is available. The issue to consider
is not the 128bit certs, it is the client base. To access that base of
128bit "exportable" clients you have to have the magic certificate
whether or not you are in the US. Sameer's server is irrelevant, the
certificate is used to "throw the switch" on the client, not the server.

This only looks like it hands Uncle Sam a magic key if you consider the
single case. The problem is what the response of the other users of the
foreign certificates is should the government actually put its key to
use. The revocation must be a public act for it to be effective.

Certainly I can't imagine the US government asking Airbus industries to
hand over their accounts data so they can pass it straight on to Boeing.
I can't see it affecting any of the similar industrial espionage risks
that concern the companies likely to be allowed to use the system.

This is not a personal privacy issue, it is an individual talking to
large corporation who is not going to resist a lawful subpoena in any
case.

I don't think getting someone else to act as a CA would work either.
Whoever approved the scheme at the NSA would have insisted on the
clients requiring a cert directly signed off a root controled by a party
that had agreed to enforce the USG terms.

It would be much easier to circumvent the restrictions by replacing the
https transport component in Internet Explorer. The API is designed to
allow new protocols to be slotted in. It should not be a big hassle to
introduce a full and unrestricted SSL implementation.

All Verisign is doing is selling a cert that states that the holder is
trusted by the USG to operate on terms that it considers acceptable to
use 128bit encryption. If the USG no longer trusts the holder in that
manner the attribute is false and the cert should therefore be revoked.
The problem for the USG is that if the scheme succeeds it creates a
demand for 128 bit encryption that makes the revocation mechanism too
dangerous to use.

Freeh is essentially operating under the assumption that he can persuade
the Europeans to accept his proposals if he has more time to explain
them. I suspect that the opposite may be the case and that the
resistance to Freeh is informed about the ecconomic power that the
situation hands the US.


     Phill


Phillip M. Hallam-Baker
Visiting Scientist
MIT Laboratory for Artificial Intelligence.
Email: hallam@ai.mit.edu
http://www.ai.mit.edu/people/hallam/hallam.html
http://www.hallam.demon.co.uk/