Nate Anderson reports that Microsoft has patented a new version of digital watermarks for music in a system that is undetectable and survives "the most common file manipulations" to a standard that meets the RIAA test. The watermark basically tells the viewer who originated the download so that DRM is no longer needed to protect copyrighted material, a modest improvement for buyers of music. If he passed it on, he violated the copyright and can be prosecuted link here.

As I read this story, I wondered whether watermarks could be used for other purposes involving restricted material. This suggest that it has great potential for the violation of privacy. For example, does it allow Big Brother to learn what you are watching or doing on the internet, or video or e-mail? That is the gist of David Lazarus' piece in the LATimes recounting how cable and phone companies which are sending out new privacy terms that would allow them to do just about whatever they wanted with their records of whom you write to or what you watch link here.

He reports that you can opt out of this invasion of privacy, but the means varies from company to company and you may have to do it in writing--it is up to the individual to act. Earlier this month, a federal judge shot down a section of the USA Patriot Act that allowed the government warrantless access to telecom companies' databases, but there is nothing in the privacy policy that would limit the company from giving the information to the government if it decided to, as the phone companies did earlier before the Patriot Act.

Lazarus seems to think that the information will be sold to companies to adjust their advertising, change their programing, or for other commercial purposes. He seems to have missed the possibility that the government might be a recipient.

Incidentally, Lazarus reports the data retention policies are generally far longer than the three-year rule followed by Google after a lot of protest.

Even if this audio watermark is as durable as claimed, it is actually a development with unsurprising consequences.

It is very difficult to see how it assists copyright enforcement.

Much copyright infringement occurs in the digital domain. We've long had technology to insert digital signatures into digital files. So, why does having the ability to do the same in the analogue domain make any difference?

Don't forget: there is no security through obscurity.

And also don't forget, the RIAA has never needed any evidence of copying to sue people so far.

Of course, if publisher and player manufacturers collude, they can produce less flexible music and devices, but then that's fine. Musicians and their fans who aren't into such masochism can publish their own flexible music and use more useful devices.

The more channel and platform owners tighten their grip upon their customers, the more they will end up with customers who do not feel their liberty constrained or privacy violated.

Naturally, there's a small fraction of human beings into such perversions as liberty and privacy. People the members of RIAA are no doubt happy to see take their custom elsewhere.

After I first wrote this post on watermarks, I came across an article that provides a description of how another patented version works link here. That patent is for "a monitoring service that scans the Internet, consuming content as it goes. The system downloads audio, video and images, and then scans them for watermarks. If it finds a watermark it recognizes, the system then contacts that mark's registered owner and informs them of the discovery."

"For the system to work, players at multiple levels would need to get involved. Broadcasters would need to add identifying watermarks to their broadcast, in cooperation with copyright holders, and both parties would need to register their watermarks with the system. Then, in the event that a user capped a broadcast and uploaded it online, the scanner system would eventually find it and report its location online. Yet the system is not designed to hop on P2P networks or private file sharing hubs, but instead crawls public web sites in search of watermarked material. As such, this "solution" is more geared towards sites like YouTube and less towards casual piracy, which rarely involves posting things to a web site."

However, we will need the fine print on the Microsoft patent to know how it works.

You don't need to know how it works. Either it doesn't and you can ignore it, or it does, and you can ignore it.

There is some kind of FUD here that makes people think that one can taint a copyrighted work as one might unavoidably cover photocopier paper with one's fingerprints or taint it with one's DNA.

Things are different in the digital domain. While we may enjoy digital metaphors for their familiarity, the difficulties we have with the material analogues are not necessarily replicated in the digital realm.

We should have learnt this lesson by now in the futile attempt to pretend copyright is sensibly applied in the digital domain.

After some thought, I imagine that a different watermark will be added for every customer, so the original pirate can be identified by the watermark itself.

Availability of the content with that watermark is then proof (or strong indication) that the customer associated with that watermark is responsible for the piracy.

If this is how it works, it should not take any self-respecting hacker more than 15 minutes to figure out how to mangle the watermark enough that identification is impossible.

For a digital watermark to be a perfect metaphor for a paper watermark, it must be:

1. Illegal and extremely difficult for anyone apart from the crown to insert.
2. Impossible to remove without damaging the work.
3. Relatively easy for anyone to recognise (given good light).
4. MISSING FROM COPIES

This is not something you'll find in the digital domain.

1. If the crown can insert mark C, everyone else can insert mark X (and probably C too)
2. If mark C can't be removed, then nor can any other.
3. If everyone can recognise C, then either C+X+Y+Z is illegible or everyone can recognise C, X, Y, Z, etc.
4. No-one's mark will be missing from copies.

This of course assumes that 'everyone' has precisely the same materials as the crown. If the crown has exclusive access to the multitrack digital master, then everyone else would find it difficult to produce the same variety of unique composites. However, it seems, that the signature insertion mechanism can be incorporated into any copying device such that the device adds its own signature to the copies it produces - which effectively means that anyone can insert any signature they fancy (given reverse engineering).

Now, even if at the mastering studio a unique work is recorded onto each of umpteen million CDs, all that can be determined by the diffusion of each via sharing is the ultimate provenance of copies. One cannot actually determine WHO performed the copying. Last time I checked, failing to secure a CD against copying (or even theft) was not illegal. Moreover, it has not yet become illegal to sell CDs (after three owners, who can tell who is the culprit for leaking copies?).

And what happens if 10 CDs are brought together in order to create a composite? Even if the audio is adjusted in the frequency domain, a composite can also be made in the frequency domain.

The Field of Boliauns is always a useful reminder about how readily people can be undone by assumptions regarding identity.

At the end of the day what would most likely happen is that the ability to encode a fairly durable signature in an audio stream would be considered a gimmick and then simply added as yet another option in CD burning software, e.g. "Insert secret audio signature: Y/N?"

It might just catch on as a way whereby musicians could reward their best promoters (one doesn't need to prove who copied anything in order to count how many copies have a particular signature), e.g. "Publish secret signature to Google MusicSurvey Y/N?"

It would miserably fail as evidence of who was culpable for copying. Copyright infringement is based on evidence of who copied, not upon evidence that a bootlegger's uploaded files were ultimately obtained from an unwitting punter's unsecured laptop at a cybercafe.

All that's happened here is that the RIAA have specced a particular technology, and MS have developed it for them. RIAA did not ask for a means of identifying copyright infringers (nor even of plugging the analogue hole), and MS has not given it to them.

When RIAA implements it - and I hope they do. It will only help them and everyone else recognise just how ridiculous copyright is.