Re:Not going to work.... (Score 1)

You cannot eliminate covert channels and doing so may be more negative than positive in my opinion.

For every covert channel removed one is created (maybe that's even been proved or provable). Since my German is rather poor I'm educated guessing at this situation: (1) plaint text (2) stego text (3) modified stego text (4) modified plain text. So (3) is transmitted and decoded and we wanted (2) to be decoded. First is decoder modification. Is it impossible to convert (3) to (2) or (3) to (1)? Certainly not given we have enough redundancy in the hidden signal. Why not stego your text into the background, foreground, midground (and maybe another "ground" for safe keeping) then take a weighted average. This is all rather trivial.

A second approach, would anticipate the change in (2) to (3) and modify your encoder so that a decrypt of (3) equals (1) not a decrypt of (2) equals (1). For any cryptographer this would be a trivial exercise. So in concluding this not near to a solution it's what we call "snake oil cryptography" and unless a PROVEN algorithm is released it should be entirely ignored.

Also, the above only scratches the surface after half a minutes thought. Imagine what somewhat with incentive could come up with? Concerning the negatives to convert channel elimination: a least now we no one will be sending those hidden pictures of Tibetan monks and children being beaten by the Chinese authorities, I feel safer already. For example.