>From: "Esteban Gutierrez-Moguel" <[EMAIL PROTECTED]>
>> a solution that problem could be a cipher where a key (K1) decrypts the
>> ciphertext to the real text and a key (K2) decrypts the ciphertext to a
>> meaningful text, but not the real one. In that way if the police requires
>> the key you can provide K2 and nothing is lost.
Systems like this have been discussed occasionally, but nobody's got a good
one.
Problems include:
- need twice as much cyphertext to store the real plaintext and the cover
plaintext
- software that does this encryption/decryption tends to be obvious about it,
so if they find the software, they'll look for the hidden message.
- Doing this without obviousness in the decryptionware is much harder -
the cover text tends to be gibberish, and what you need is plausible
deniability.
At 06:51 PM 7/20/00 -0400, Marcel Popescu wrote:
>Yes, I think that this, Rivest's "chaff", and steganography, might be used
>among Britain's libertarians... and terrorists, btw. [As usual, restrictive
>laws will affect law-abiding citizens. Now I started preaching to the
>choir... <g>]
Chaffing is fun, because it does technically not use encryption.
Doesn't mean the authorities won't have a judge order you to fork over the
keys
even though they're just authentication keys, not secrecy keys.
It might be interesting to mix either chaffing or multiple-key with stego -
e.g. the secret message is hidden in the low-order bits of the red pixels,
and the cover message in the low-order bits of the green pixels,
and you pick red/green/blue as a decryption option.
Or the winnowing stream tells you whether the real message is in red,
green, or blue,
though it's probably much more effective to use secret-key encryption on
the stegobits
and let the cover traffic be the encrypted picture itself, like the badly
retouched picture of
Jack Straw saying "Wassuuuup!" or the crypsterised MP3 Metallica copyright
violation.
Thanks!
Bill
Bill Stewart, [EMAIL PROTECTED]
PGP Fingerprint D454 E202 CBC8 40BF 3C85 B884 0ABE 4639
