🚀 go-pugleaf

 In <1992Oct7.224601.8915@qualcomm.com> karn@qualcom.qualcomm.com
 (Phil Karn) writes:

>In article <17939.Sep2300.39.3792@virtualnews.nyu.edu>, brnstnd@nyu.edu
>(D. J. Bernstein) writes:
>|> In article <1992Sep20.195136.8642@cactus.org> ritter@cactus.org
>|>(Terry Ritter) writes:
>|> >  As far as I can tell, the whole point of cryptography is to be
>|> >  able to restrict information to only those who are intended to
>|> >  have it.
>|>
>|> That description is far too broad. You cannot singlehandedly restrict
>|> information to a chosen recipient, because he can in turn give the
>|> information to someone else. You pose an impossible problem; is it a
>|> surprise that there are no solutions?

>Indeed. I've been making this point regarding anti-piracy devices for
>some time. Even if Videocipher, for example, had not been broken there
>would have been no way to prevent someone from taping the output of a
>legitimate decoder and selling it or giving it away. The same is
>undoubtedly true for software packages, at least those running on
>standard general purpose machines. If you can't trust the authorized
>user, then all bets are off.

 The rest of my original comment was:

    "When a public key "appears," it may really be a key
    which was generated inside a spoofing node.  When one responds
    to that key, one's response may be deciphered in the spoofing
    node, then re-enciphered to the ultimate recipient.

    "Note that the recipient does, in fact, get the message.  One
    can communicate in cipher.  The problem is that the spoofing
    node gets to read all the communications.  So unless we actually
    *intended* that there be a spoofing node, and that they should
    read our messages, I think the cryptography has failed."


 The reality that someone privy to a secret can betray it is
 something most of us learned the hard way, before third grade,
 and may have re-learned many times since.  This is not news.

 What *is* news is the idea that anyone would equate betrayal at
 the other end with the possibility that a spoofer could *also*
 be reading the mail and distributing it.

 Expressed perhaps a bit more clearly:  The *whole point* of
 cryptography is to deliver information, *unexposed*, to the far
 end.  Then, if the secret *is* exposed, at least you know it was
 them (or you).  The ability to identify and eliminate channels
 of exposure is a major part of security.  (No previous part of
 this thread has concerned the broadcast distribution of secure
 information.)

 If a spoofer *is* reading the mail (something well within the
 range of a cracker, but which could easily be prevented by the
 user), the system is *not* pretty good cryptography, it is not
 even good cryptography, it is just *failed* cryptography.  Instead
 of a fancy two-key cipher with a strong one-key data engine, it
 might as well have been a modest homophonic substitution, or a
 stream cipher with a little 32-bit LCG or LFSR.  But this would
 *not* be real cryptography, it would be *toy* cryptography.

 ---
 Terry Ritter   ritter@cactus.org