As long as we've got private communications, it will become ever easier to share media over the internet. It only takes one "Pirate_0" to break the copy protection (even if it requires digitizing LCD signals). And if breaking the protection requires substantial capital, the market will oblige.
Two simple examples:
If a license or physical player is needed to create a high-fidelity copy (for instance, by capturing raw video streams from a player), and the output video is watermarked, content providers can revoke the player. Now "free" copies cost players.
If the same technology creates a trail of evidence leading back to "pirate 0", people will start getting sued. Not everyone the RIAA sued was innocent, and there's nothing to say that the DVD legal teams will create the same PR firestorm the RIAA did by abusing the system. Just as importantly, if it takes a huge amount of effort to create a supply chain of unlawful copies, the court case is much more damaging.
Content protection is not really a practice focus for us (though anticheating and kernel protection are), so my thoughts on this are not an exhaustive summary of the space of solutions content providers have. I'm just saying, the content providers aren't always on the wrong side of computer science.
Even if it costs $1k to get each movie release out of protected form, it's cheap for a (dubious) advertiser wishing to get a message out to a huge audience.
And further down the road, I suspect there will be an anonymous market for such content. $1k is only a penny from 100k people.
My point is, "useless" is too strong a word. It isn't silver-bullet simple, but it is a tool that can be made to work. Content providers are already costing Slysoft money; now they just have to figure out how to increase that cost fast enough to push them out of business.
But for getting one-time copies out, there will always be an analog output at the end of the chain. Watermarking allows the distributor to pinpoint the device used, but the only thing they can do with this information is shut down the device. I don't see how this can be an economic threat when player prices must be kept low for consumers, but there are millions of people wanting pirated content (which is easily distributed after extraction).
How hard is it to forge the required identification or bribe a store clerk to not check identification?
If someone's house is broken into, are they going to be liable for piracy through their stolen media player? (Assuming there's a several hours before it can be disabled)
And, with regards to "are they going to be liable", I direct your attention to the track record of the RIAA.
Of course it's hard to speculate the difficulty of such forging in the future. But I would think that $10k would be enough to break some human link in the chain.
"Starting in 2006, many new laptop computers have been sold with a Trusted Platform Module chip built-in... Intel is planning to integrate the TPM capabilities into the southbridge chipset in 2008..." http://en.wikipedia.org/wiki/Trusted_Platform_Module
Or simply unaware, and with little chance of ever finding out:
http://isis.poly.edu/csaw/embedded (and assorted talks at HOPE, Defcon, Shmoocon, etc. on hardware and firmware covert channels)
The analog gap isn't an unsolveable CS problem. Yes, you can't encrypt photons. But you can watermark and trace them. Preventing copies is CS-hard. Preventing covert channels is also CS-hard. Watermarks are a defensive application of covert channels.
Watermarks -> revocation -> economic disincentive to copying transactions.
Watermarks -> evidence -> lawsuit -> economic disincentive to copying at all.
Also, none of these things has to work perfectly. The industry's goal isn't to stop copying. It's to capture the maximum amount of revenue per title.
So far, the BD+ scheme on Blu-Ray disks has had to be bypassed title by title, with Slysoft's breaks being corrected by Macrovision and Slysoft averaging 2.5 weeks on each of the refreshes. Nobody has come up with a scheme that completely defeats BD+. And BD+ is the first system of its kind; subsequent systems will be even more of a bitch to deal with. Add to that the fact that Slysoft has a really smart team that does almost nothing but this.
Take another example: DirecTV. The current DTV smartcards are unbroken, despite a ~7 year track record of breaks leading up to them.
Take another example: iTMS FairPlay. Last time I checked, there still isn't an equivalent break to JHymn, which does a direct decryption of audio files.
Take still another example, of protected media pathways on Windows kernels; currently shipping Intel hardware has chipset-enforceable RSA/SHA/AES locks on kernel integrity, which is something you want anyways, because without it you can't trust your kernel against malware.
And of course, all of this assumes that the goal is to keep people from copying at all. The industry could also solve the revocation problem, so that when copies are detected, players are shut off, accounts are closed, or people are sued. Ed Felten beat early watermarking attempts, but come on, do we really think watermarking is an intractable problem? It's the covert channel problem turned around to favor the defender; the media companies will win it.
I'm not making a moral stand here (although I really think that if you don't want to pay for DRM'd Miley Cyrus tracks, or pay for Miley Cyrus at all, you simply shouldn't give your money to Warner Music Group). But I object to the knee-jerk dogma that copy protection can't work. I think it can.