Kragen writes:

I was thinking of steganography as being in two stages: first, you encrypt, (possibly with the identity transformation) then, you embed your encrypted message in your medium of transmission. My previous message was describing requirements for a strong encryption algorithm, quite apart from the actual embedding. I stand by my statements: the purpose of steganography is to make it difficult or impossible for an interloper to determine that enciphered data are being transferred. Thus, embedding a magic number in the file defeats the purpose completely. (As opposed to "slightly reducing security.")

True. I was refering to cryptographic security instead of steganographic security when I said "slightly reducing security." Sorry about the miscommunication. For steganographic purposes, there should be no _constant_ magic numbers or CRCs taken _after_ the encryption (and visible in the ciphertext). All magic numbers and CRCs should be embedded _before_ encryption and checked _after_ decryption when you want the ciphertext to look purely random. This way you can have _both_ cryptographic convenience and random looking ciphertext ready for steganographic hiding.

I think that designing a program to embed this apparently random bitstream in an innocent-looking file is a different and much harder problem.

Definitely. Such a program is also very likely to drastically inflate the message, depending on the definition of "innocent-looking" and the characteristics of the channel or storage medium used. For example, a message could be concealed in the number of blank characters after each line of text from an recipe book, but someone might even get suspicious about a sudden interest in cooking among cypherpunks. :) By the way, I heard a rumor from a telephone company employee who I met (face to face) who is in a position to know that a U. S. company was using DES to communicate proprietary information between one of its facilities in Japan and an office in the USA. They got a letter from the Japanese parliament asking them why they were sending encrypted data. Perhaps there is more to the question of steganography than purely academic interest... ----------------------------------------------------------------------------- Mike Johnson | Opinions expressed herein are mine, and come with no mikej@exabyte.com | warranty, expressed or implied. PGP key on request. -----------------------------------------------------------------------------