From: jim@bilbo.suite.com (Jim Miller)
Date: Sun, 13 Feb 94 13:01:14 PST
To: cypherpunks@toad.com
Subject: Re: a protocol (that doesn't work)
Message-ID: <9402132053.AA19346@bilbo.suite.com>
MIME-Version: 1.0
Content-Type: text/plain



Did you ever wish there was an "unmail" command?

I realized about halfway home that the protocol I described not only  
didn't work, but demonstrated to the world my lack of understanding  
of the man-in-the-middle attack against Diffie-Hellman.  Oh well, I  
guess I'll keep my day job a little longer...

At least I now have a better understanding of just how hard it is to  
foil man-in-the-middle attacks.

>From reading the description of the Interlock Protocol, I saw that it  
is possible to arrange things so the man in the middle has to do a  
lot more work.  It may be that Robert Cain has come up with a  
protocol that increases the work necessary to maintain a  
man-in-the-middle attack to the point where the attack becomes  
impractical, although not impossible, in theory.  However, I think  
that is this becomes the case, an attacker would simple cut Bob  
completely out of the picture and change the man-in-the-middle attack  
to a %100 spoof of Bob.  Since Alice and Bob have never met and don't  
share any secrets, how would Alice be able to tell the difference  
between the real Bob, and Mallet completely spoofing Bob?  In the  
abstract, I don't see any way.

There may be some real-world situations where Alice can tell the  
difference between Bob and pseudo-Bob.  It depends on the situation  
and what assumptions Alice makes about a properly behaved Bob.  If  
pseudo-Bob doesn't behave the way Alice expects real-Bob to behave,  
then Alice could get suspicious.  But now we've existed the realm of  
cryptography and enter the realm of human relations.  Of course,  
there's still a lot of money to be made offering imperfect solutions  
that are good enough for some people.


Jim_"still learning"_Miller@suite.com