Comment on Science News article re: Rivest's Chaffing and Winnow ing

Pete Klammer (pklammer@racom.com)
Wed, 6 May 1998 13:53:02 -0600

• Next message: Wei Dai: "Re: fast modular multiplication inverse"
• Previous message: Jim Thompson: "'rules'? for commercial postings"

re: "Hiding secret data in plain view" by Ivars Peterson;
Science News, Vol. 153; May 2, 1998; pg. 286.

  Ronald L. Rivest has published a new technique called "chaffing and
winnowing" which allows communication of secret messages without traditional
encryption. This is accomplished by larding the intended message with
packets of gibberish, and differentially tagging all packets with message
authentication codes (MACs) so that the only the intended recipient,
possessing the correct MAC keys, can pick out (winnow) the gibberish (chaff)
from the intended message (wheat!).

  We say "without traditional encryption" because the intended message,
although broken up into fragments, is not scrambled: neither are any of its
characters substituted nor are they rearranged in sequence. However, one
may very well argue that chaffing and winnowing,
viewed large, is nonetheless cryptography: sender inserts message and key
into black box at one end, producing unintelligible output; recipient
inserts this output and key into black box and gets message at other end --
ist das nicht ein encryption system?

  But Rivest has achieved something of much greater sublime impact than just
another encryption method, regardless of how practical chaffing and
winnowing may be. Much more profound, he has demonstrated a "reducto ad
absurdum," that is, a proof of futility, of any attempt by authorities such
as the U.S. government to prohibit secret communication. The mere
description of chaffing and winnowing is a "thought experiment," like
Einstein's imaginary trips on lightspeed trains; it is a philosophical
demonstration that legal restrictions of secrecy are doomed to fail, because
the "gibberish" can easily be chosen and arranged to mask the secret message
and look as innocuous as you like.

  You and I could just as easily prearrange a system of winks and nods to
clue each other when we were truthtelling and when we were lying, and then
carry on a conversation entirely in the open but useless to anyone without
the "key" to know which of our words were significant and which were filler
-- could the government outlaw that? [Rivest acknowledges another another
correspondent's analogy with baseball coach signals.] Between winks and
nods, and chaffing and winnowing, is just matters of degree. The only way
to prevent secret communication is to prohibit communication at all. Hence,
freedom of speech ultimately equates to right to secrecy!

• Next message: Wei Dai: "Re: fast modular multiplication inverse"
• Previous message: Jim Thompson: "'rules'? for commercial postings"