A group of Microsoft researchers, including Paul 'Mr Secure PC' England, has delivered a paper which concludes that all efforts to stop content swapping/theft - possibly even including Palladium - are in the long term futile. This message, particularly the bit that dealt with the economics of DRM-enabled versus 'clean' content, must have gone down a storm with the audience.



Which, since you ask, was the Association for Computing Machinery DRM conference.



The paper, which is currently available here, is particularly striking in that it argues its way persuasively through the history, present and future of file sharing, the success or otherwise of 'attacks' (academicspeak for 'lawyers') on it, and concludes that file sharing will triumph.



All the way through, given their employer, you're expecting them to get onto 'and this is why we need Palladium'. But where they end up is more or less with the conclusion that vendors should sell their content unprotected and compete with the 'darknet' on "convenience and low cost rather than additional security."



The full paper is well worth reading, but they arrive at their conclusions via the following route, approximately. The darknet, which is their catch-all term for networks of content-swappers, originates as 'sneakernet', where overlapping groups of friends and colleagues swap stuff, so stuff gets around, but slowly. Attacks (lawyers again) at this stage are difficult because the groups' identities are largely unknown to the content owners.



The arrival of Napster sends the stuff around to far larger groups much faster, but with a fatal disadvantage - Napster, as custodian of the database, is itself vulnerable to attack because the lawyers know where it lives. Gnutella, then?



Now, this bit is interesting, because although in theory there is no central attackable focus point, they argue that the network is showing a tendency to produce such vulnerabilities. Free riders, people who take content without giving back, constitute a large part of the network, whereas contributors tend to be concentrated on smaller numbers of hosts with large amounts of bandwidth and storage space.



Which at the moment tends to mean unauthorised use of company or uni equipment, which again means a small number of hosts vulnerable to legal attack. This will also apply to individuals hosting large amounts of hot stuff, because if you're going to be a warez king, you're going to get made an example of.



A fascinating and plausible argument, this. That however is not the end of it. Growth in consumer access to broadband will make distributed file sharing networks more proof against such attacks, and in the interim a resumption of smaller overlapping groups, together with systems whereby hosting is compulsory for members, will mean the content will still get swapped. And the factor that slows up distribution at the small group phase, lack of a central index, will eventually be overcome by comprehensive distributed indexes.



So stuff will always get around, it'll get around faster and faster, and although you can lawyer the odd Napster out of existence, ultimately lawyering is futile (you guys sure you work for Microsoft?)



OK then, so granted you can't stop it, can you protect it? Watermarking, fingerprinting, DRM, hardware IDs? Watermarking wil always get cracked, and key-based protection will always be overcome by key leakages. They've some words of good cheer for Senator Fritz Hollings, who wants to make it compulsory, here:



"Proposals for systems involving mandatory watermark detection in rendering devices try to impact the effectiveness of the darknet directly by trying to detect and eliminate objects that originated in the darknet. In addition to severe commercial and social problems, these schemes suffer from several technical deficiencies, which, in the presence of an effective darknet, lead to their complete collapse. We conclude that such schemes are doomed to failure."



Fingerprinting (i.e. you don't protect it all but you know who's got it) isn't much use either. It might let you chase down individuals, but then the big boys will just buy several originals, stir them together and screw up the fingerprint. They don't altogether rule out hardware, so from a technical perspective they're only nearly saying the DRM associated with Palladium is pointless, but their final economic argument seems pretty conclusive on this.



"There is evidence that the darknet will continue to exist and provide low cost, high-quality service to a large group of consumers." That makes it a competitor to "legal commerce," and "increased security (e.g. stronger DRM systems) may act as a disincentive to legal commerce."



So for example, you could buy an MP3 file or you could buy a DRM-protected version of the same content. The MP3 file has equal utility to the darknet version, whereas the protected one is worth less, people will vote with their feet, and WMA format is doomed (no, they don't say that specifically, but...)



Vendors should therefore compete on price and convenience. This is a moving target, because it's dependent on the level of the price and the increase in convenience over stealing the stuff instead. So for music, the price of unprotected legal content would have to be quite low, and the convenience quite high, whereas for video the price could be higher, because the darknet hosting bandwidth isn't up to it yet. Think about it being just nicer to not steal than to steal, and how that could be achieved in particular areas of content.



But thinking about it again, maybe in some senses they are arguing in support of Palladium. Palladium itself is not DRM, it is a trust system which can support DRM systems. Now, if content vendors are to compete on price and convenience, they need users to have trust systems that are convenient to use and secure enough for them to pay the price with confidence. Heard of something that's supposed to support this? Maybe Microsoft has the last laugh after all. ®



Related stories:

Of TCPA, Palladium and Werner von Braun