call it a media key … it differs from encryption as the index of functions to be determined (for a given application of media binding depending for instance on use, bandwidth, type of signal, frame setting, other schemes to be integrated, etc.) is not directed at the signal in the sense of an encryption …

if you know that the signal will have to be rendered in the clear (phish :: all that i see cant be taken from me) wasting computation on access restriction instead of units of complexity enables you to maintain legacy versions of the content … at lower computational cost … the physical media or network settings can provide embedded or meta-data to affect other aspects of the system.

it’s gotta be my shoes … sorry to be tongue tied on this one

While watermarking can provide a social deterrent to copying, it doesn’t prevent the act itself. As part of an overall software protection scheme, I do think marking has its own place.

bind key material which is signal-specific, to any level of granularity of the content itself. the constraint access-restriction inherent to *any* crypto cipher (the ciphertext does not reveal or leak information concerning the plaintext – security held “only” in the key) can be relaxed … plainly: if the basic structure of the data is a bit, then we are binding more than a bit up to a level of granularity that is consistent with the processing of the media to be “protected” … speed bump the process … do not completely access restrict the data — that is wasted computation (an opinion).

this could be the frames of AAC or MPEG or some other “formatting” but it is simplistic enough to generate and replace keys (a la itunes), by focusing on the encoding itself and relaxing the crypto cipher constraint … this has overlap with winnowing and chaffing and was developed earlier than that work.

another explanation to the very specific goal you are discussing …

if a digital watermark is encoded with key material eg you can differentiate between 2 copies with knowledge of the key material but otherwise the content is perceptually the same … there needs to be at least one bit of difference between the original unmarked content and the marked content …

now, instead of a a watermark key which for purposes of discussion describes how the watermark is encoded into the content you have a key which describes the encoding or binding of the media — it is not as secure as crypto keys in the traditional sense but secure enough to enable upgrades with smaller computational overhead …

we do this at minimal overhead and are able to match a specific piece of content with a specific key — the key is the binding between analog/dsp I/O and “level of complexity” to discourage piracy using devices or software which render the content … the key and content are specific to each other and relate input to output with a measurable amount of complexity …

this is the narrow goal, achieved…

This post was about one very specific security goal — tie some set of bits to its original media. Even with a system like AACS that uses cryptography (NNL key tree), preventing bit-for-bit copying of the disc comes to inserting some piece of data (i.e., 128-bit Volume ID) in a location that can’t be written on recordable media. DVD-Rs that come with the CSS key block area already overwritten with zeros use a similar approach.

So in this very narrow goal (tying bits to their original media), encryption does not get you anything. It ultimately comes down to one of the three categories of media binding I discuss in the article.

for example,

let’s call encryption coding requiring a secret such that the encrypted information can be extracted by those with the secret.
the secret is a key which is a family or functions or index of functions.

the primitives for crypography are based on the computational difficulty of cracking the key. it is a yeah or nay on access to extract the plaintext. but, in an age when such computational complexity would appear to limit enabling a market for content or information that may be subject to the digital copy problem. this is not to say conditional access, white box drm, etc. are not useful.it is to say the competition is for time and attention and may best be served by relaxing primitive to include information that requires access to enable fair pricing (who says what is fair?)

but, a simpler way to do security, in the sense that keys are easy to change and not based on a strict access vis-a-vis encryption but a nuanced use of a key to alter the associated signal in a manner that makes measurement of the complexity or quanta of “security” applied. additionally functions as part of the key may include specific information which may increase or decrease the computational requirements for rendering the content. no new players just key readers … (call em predetermined or media or content keys, whatever)

i think a fair way to observe security options are: proprietary coding (though the perceptual model is easy to replicate as it is based on human observation) – encryption is transport layer only and does not reveal anything about the plaintext — ideally; content extensions or wrapping which are active controls but not integral with the content – they aren’t digital signatures that survive transform conversions; watermarking for plausible deniability/traceability and integrity; format manipulation relies on a key describing how the encoding has been manipulated…

the manipulation can be a measure of security or simply a way to measure how to split the money from transactions.

code protection, by the way, is related in the sense that the code level is interoperating with digitized signals and vice-a-versa. the index of functions relates inputs to outputs in way that can be manipulated for any number of threat scenarios.

hope i am not rambling … would like to hear some thoughts …