Yes, to keep people from copying a movie or a book requires more-than-totalitarian control over people. But that's from a special property of noninteractive works, that one viewing is all there is to them. To copy-protect a program, in principle you could deliver it only to tamper-proof machines that self-destruct before giving up the code. You can't infer the program from its behavior, in general, unless P=NP (though of course you can do well enough for many programs; but black-box reverse engineering is more like creative work than scanning/recording). And releasing an app only for a particular platform is not a human-rights violation, at least not on the scale needed to suppress cameras and Turing machines.

That would take engineering beyond the current state of the art, but I know of no proof it can't be done, and at least one person I respect seems to think it can: http://e-drexler.com/d/06/00/EOC/EOC_Chapter_11.html#section...

So "DRM is impossible" strikes me as short-sighted. DRM is bad for humans, impossible for the media that Big Media currently cares the most about, and currently unable to keep anything from escaping into the free world.

You do bring up some interesting points. Restricting a program to physically secured hardware along with software protection would definitely make it extremely difficult to get the program or information outside of the way that the maker wants you to. Although I am not entirely convinced that it would be impossible. The issue is that it still must be a deterministic system. And this means that with enough money, expertise, time, precision someone could produce a situation in which they would be able extract the code and edit it. Now in most cases the amount of resources and effort that would be required to do so would probably not be worth it which would mean that it wouldn't be worth it to pirate, but i would be hesitant to say that it would be impossible to. Any deterministic system can be compromised the key is to make compromising it so difficult/expensive/time consuming that compromising it is not worth it.

Now from a practical stand point I don't know how well this would work in a consumer market, but can certainly see it being used in areas where protection of information is extremely important. It just seems a little to costly to implement on a wide scale to me, and I think consumers would see it as too invasive and resist it, so if there were alternatives in the marketplace I think a lot of people would opt for the alternative.

I broadly agree, but claims of impossibility face a burden of proof; and you can see iPads as cheap crude approximations to sealed systems, and they're effective enough that AFAIK most developers needn't worry about pirating. And I'm not sure but I get the impression that successive models of iDevices tend to take longer to find a jailbreak for.

See my post to another one of your comments as to why as far as I can see it is impossible to make a system that is completely unbreakable. But I definitely think you could create a system that is so difficult to break that in practical applications it would cost more to break then what is to be gained. The solution you suggested is likely a very good example of this. I'm not saying your proposed solution isn't a good solution to the problem or that it wouldn't work in the real world(if security is the only concern). I'm simply saying I don't there is such a thing as an unbreakable system, as long as it is deterministic.

>in principle you could deliver it only to tamper-proof machines that self-destruct before giving up the code.

"Tamper-proof" is impossible. That's why they call it "tamper-resistant" -- someone with enough resources is going to be able to tamper with it. In theory you may be able to make tampering expensive, but now you're just back to playing the cat and mouse game, and people start showing up with timing attacks etc. that aren't as expensive as you might have hoped.

Exactly. Anything that is deterministic can be compromised, even if that is something physical. The only thing you can do is make something so difficult/long/expensive to compromise that it't not worth it. This is the theory behind encryption, given enough time and/or computing power and encryption can be broken by brute force but making it require so much time and/or hardware to crack that the information would be useless by the time it is broken. You can't make a deterministic system unbreakable, only harder and harder to replicate the situation needed to break it.

I guess you're bringing up determinism with the idea that if you can extract some information from a probe even if the system self-destructs, then given enough diverse probes you can extract everything. But the designer could refuse to send you an unlimited number of copies.

I linked above to someone sketching a strategy to distribute limited DRMed designs worth trillions of dollars to unlock. He may be wrong, but it's enough to make me think the current state of the arms race need not hold forever.

If the system has a state where the hardware reliably doesn't self-destruct, then that state can be replicated. From there probing can handle extracting the information needed and presumably some sort of interpreting of the information would be required to make it workable. Now I think this process has to be possible for any system (a reliable deterministic state has to be able to be replicated unless I am really missing something) but there a lot of ways to make replicating this process very difficult and unlikely to the point where I can't imagine anyone actually doing it, but it still is theoretically possible for someone to break it.