Steganography
Interesting discussion on /. about steganography (well, an interesting ACM article and a moronic discussion - but that’s the usual case for slashdot). Thoughts: steganography in the time domain is not mentioned. For instance, port knocking hides information in the timing between packets. How about the frequency with which one checks one’s email? Times at which one answers one’s instant messages, modulo two seconds? Transmission rates of streaming video?
Second thought: the current stego methods seem to be defeated by statistical analysis - basically, there is noise where there should be no noise. Perhaps a better implementation might be to hide in the noise. An encrypted stream should have no statistically discernable characteristics from white noise. One then looks for white noise in the spacial or frequency domain of an image/video/audio file, and changes the noise signal to one’s encrypted message (with the intensity of the stego message being proportional to the intensity of the image’s original noise pattern).