Ubercyberlawprof Larry Lessig has written a damning-with-faint-praise piece for Red Herring on Microsoft's new digital rights management (ahem, trusted computing platform) "Palladium" (Anti-trusting Microsoft). The article has created a bit of a minor tempest. Copyfight has been following the discussion via Ernie the Attorney (Lessig "the world's most important pessimist" is quasi-optimistic about Microsoft?), Will Cox (E the A // Comments), Frank Field (2002 September 11 Links) and the response via Larry's own Lessig Blog (an optimist? I wish). Many, many more good links are available in another Copyfight follow up (Lessig on Palladium: A "Token" Defense, Part III).
The main point of the argument seems to be that some digital rights management systems are better than others. In this case, Lessig argues, Palladium may be better at protecting the end-to-end principles of the Internet than digital rights management enforced in the network itself. He is not arguing that some forms of DRM are preferable to no DRM at all. Nevertheless, I have some difficulties with Lessig's piece:
Palladium, Network Copy Control and the End-to-End Argument
As Edward Felten explains, implementing digital rights management in the network itself is extremely unlikely ("Network-Based" Copy Protection). For example, one means of controlling the network would require every node in the network to check every packet to determine whether the packet was infringing or not. Such a system could easily be flummoxed by using the simplest encryption. Of course, the network might require that the sending application provide certain assurances that it wasn't encrypting. However, it would not be too difficult for a rogue application to provide a simulated version of the proper assurances if the "ends" of the network are free to do what they want.
Modern P2P networks permit trading of all file types, not just music files. Back in its heyday, however, Napster permitted trading of only MP3s. If you wanted to search for videos or images, or even music file types other than MP3s via Napster you were out of luck. So, some enterprising programmers created a program that made other file types appear to be MP3s. It was called "Wrapster", because it "wrapped" other files types in MP3 packaging. Although an implemented version of network copy control would include all sorts of encryption tricks, some form of "Wrapster" would eventually be developed that would permit copyrighted works to be transferred across a network with embedded copyright protection.
In order to prevent this, you would have to control the "ends" of the network, where the applications reside. (Aside: Currently applications reside locally on your PC's hard drive. This does not have to be the case. Your PC could be a dumb terminal with the applications actually residing on a central server which is easily controlled. This is how cable companies intend their systems to work. Rather than download a movie for later viewing to a smart cable box, video-on-demand would have you stream the video from the cable companies' centralized servers. Hollywood would love nothing more than to ban the PC in favor of dumb terminals.) You can never have effective network control of the type imagined without controlling the ends. The choice then, is not whether to have "ends" control or "network" control, but whether to have "ends" control alone, or "ends" control and "network" control.
Lessig's response is essentially that Congress might foolishly mandate network DRM anyway (where in the network?). First, as Felten points out, current proposed laws seek to regulate the ends (i.e., "digital media devices"), not the network (Network Centric DRM). (Aside: I've also always thought that Holling's CBDTPA is a red herring. It is so draconian that it can't be serious, but will scare the opposition into "compromises" in name only.) Second, even if someone was so foolish as to introduce a bill that required protection only in the network and not in the ends, it would never have a chance of passing. Hollywood would be less than enthusiastic about such a bill since it wouldn't really help them for the reasons set out above, and the telecoms would fight like starving, cornered tigers against it. You see, implementing DRM in the network would mean huge additional expenses for the telecoms themselves without helping them one bit.
Will Palladium Shut Hollywood Down in Washington?
Seth Finkelstein makes the seemingly related point that Hollywood is not likely to be satisfied with only "ends" control in two posts (DRM and object control vs network control) and (The "end-to-end"s versus the means of DRM). I agree with some caveats. Finkelstein correctly argues that Hollywood will not likely be satisfied with controlling only DRM, but will demand to control P2P technologies as well in order to ensure that the Madsters and KaZaAs of the world don't facilitate the sharing of copyrighted content ripped straight from the audio, for example. I agree with this. However, it is important to emphasize that control of the one facilitates control of the other. My caveat is that, in this case, "network" control and "end" control are nearly the same thing.
Gnutella and eDonkey2000 are networks, but to the ISP network backbone they look just like any other application - they use IP and that's that as far as the ISP is concerned. That's a funny thing about computers. What is network at one logical layer is application at another and data at yet another layer. To your PC, both BearShare and a media player look like applications. If Palladium can control how a media player functions, then it can control how a filesharing application functions.
Now, Palladium will not out-of-the-box prevent you from running your favorite filesharing application, but the latent capability will be there. Once Palladium and its hardware dongle (aka "Security Support Component") are part and parcel of a sufficient number of PCs and associated embedded devices, it would be a very short step for Congress (or - it's possible - the courts) to mandate that filesharing applications take advantage of such "trusted components." Moreover, even more devious possibilities exist. For example, while Palladium will not stop your computer from running Gnucleus, it might be possible for Palladium to prevent you from running your media application if Gnucleus was present on your computer. Such a system would work thus:
In Palladium, all data can be digitally signed with a "nub" and the computer's particular personal ID. The "nub" is the particular designation for a specific operating system version. So, for example, Hollywood could sign a movie file so that it will only play on your computer when it is running Windows XP SP2. If you changed computers, the file wouldn't run. If you upgraded to Windows XP SP3, the file would not run. Thus, if you wanted to play the movie on your new computer or upgraded operating system, you would have to renew your license with Hollywood. Gee, I wonder if Hollywood would put any conditions on such approval? The technical possibilities for increased Hollywood control are mind boggling and will keep amoral software engineers well-employed.
Conversely, Hollywood could also ensure that you played your movie on operating systems of which they approved. Want to play on Linux? Sorry, too bad, Hollywood doesn't support the Linux nub. When Hollywood refuses to provide their files legitimately unless Microsoft provides a crippled operating system that doesn't permit consumers to run certain applications, what will Microsoft say? Hint: (Top Ten Benefits of MS WinXP Media Edition - LawMeme Style). This also would be something that Hollings could hang his legislative hat on - "We must mandate crippled OSs, else Hollywood will whither and die."
Is Palladium "better" than some alternatives proposed by Hollywood's partisans? Sure, but the difference is more akin to that between the electric chair and lethal injection; either way fair use gets a death sentence.
This difference might be important if Lessig is correct when he writes that, "by increasing trust at the ends of the network, Palladium would weaken an argument that Hollywood now pushes: that Congress regulate every machine on the Internet to protect Hollywood's content." However, I don't think that advantage is going to be crucial. First, it is hard to see how consumers "voluntarily" adopting Microsoft's DRM mechanism is necessarily an improvement over Congressional mandate. After all, what Congress is proposing is essentially the same thing. The bills introduced don't target routers, but consumer devices. Second, the success of Palladium will weaken one of Hollywood's arguments, only to provide them with new ground upon which to argue. As mentioned above, what is more likely to pass? The draconian and technically silly CBTDPA, or a bill mandating software "take advantage" of Microsoft's Palladium architecture? Once "trusted systems" are widely distributed, it is a lot easier to force people to use them.
As he explained in his blog, Lessig is "arguing that we need constantly to think about how different technologies affect the policy of the net." This is true, but does Palladium really preserve the principles of end-to-end? Sure, technologically, Palladium permits continued innovation at the ends of the network, but architecture is more than technology. The architecture of PCs is also open technologically, but that hasn't resulted in a lot of operating system competition (even taking into account the biggest antitrust lawsuit launched in the last 15 years).
A False Distinction Between Copy Protection and DRM
One of Lessig's key points in damning Microsoft with faint praise is his distinction between pure copy protection systems and trusted platform DRM - "Trusted platforms could enable a different kind of DRM--indeed, I would say, a "better DRM"--one less damaging to innovation and more supportive of content competition." Lessig frames the distinction thus:
To see the point, distinguish between DRM systems that control copying (copy-protection systems) and DRM systems that control who can do what with a particular copy ("token" systems that Palladium would enable). Copy-protection systems regulate whether machine X can copy content Y. Token systems regulate whether, and how, machine X is allowed to use content Y.
This is important according to Lessig because,
The difference can be critical to network design: if a technology could control who used what content, there would be little need to control how many copies of that content lived on the Internet. Peer-to-peer systems, for example, depend upon many copies of the same content living in many different places across the Net. Copy-protection systems defeat this design; token systems that respect the network's end-to-end design need not.
Lessig is providing a very simple model of copy protection, one which doesn't permit any copying whatsoever. He contrasts this with trusted computing DRM (like Palladium) that permits copying, but restricts use. In other words, with Palladium I could give you a copy of my RIAA approved music files, but you would not be able to play them without authorization from the copyright holder. File sharing in such cases would not be a problem and Napster and Hollywood would peacefully coexist (but see Seth Finkelstein's comments above). Indeed, this is the fantasy of "superdistribution."
The problem with Lessig's argument is that the model of copyprotection as "no copying whatsoever" is a strawman. Modern systems already implement variations on the "token" model and they don't require a trusted computing architecture. Indeed, modern copy protection systems aren't really about copying anymore, they are about preventing unauthorized use of copies. Take, for example, CSS (Content Scrambling System), which is used to encrypt DVDs. I can copy the data on a DVD readily enough and pass it around. I can easily make multiple copies of all of the movie files and give them to friends. However, without the proper "token" (the right CSS key) I cannot use the encrypted files. Now CSS isn't terribly sophisticated, but it demonstrates the fact that "copy protection" isn't about preventing the making of copies (which even Hollywood knows is a nonstarter), but of controlling the use of copies. In other words, Digital Rights Management is use control.
This also demonstrates that all use control is a subset of "copy protection." The primary thing that any DRM system must accomplish is that it not allow unauthorized copies to be functional. Although DRM can be very sophisticated and have all sorts of intricate usage rules, at base it starts with a copy protection system. Before you can do interesting things like allowing a file to play only three times, you must ensure that an unauthorized copy will not play even once. Even CSS has interesting additional DRM functionality. CSS is how Hollywood enforced its region encoding schemes as well as why you can't fast forward past the FBI copyright warning on some DVDs (both very interesting usage rules).
Lessig's argument also presumes that Hollywood would prefer banning all copying to a more sophisticated DRM scheme that would permit a flexible set of usage rules. In fact, the opposite is the case. To the extent that different usage rules would permit more finely grained price discrimination, Hollywood would be all for it. Price discrimination is the key to profitability with regard to copyright. Hollywood is still attempting to figure out the business models, but once they do, the ability to craft sophisticated usage rules will be just the thing.
In any case, similar technology to Palladium already exists, so why doesn't Hollywood simply implement it itself? Inter and intra-industry rivalries, of course. Remember SDMI? Microsoft doesn't bring a particularly better technology to the table, they bring distribution. If Microsoft insists that every computer sold be Palladium-compliant, the vast majority of computers will be. If it is important that the "ends" are trusted, then Microsoft's technology has few advantages other than the fact that Microsoft software runs on more "ends" than any other.
Of course, no matter how sophisticated the usage rules, fair use will still be stifled. However, politically, people are likely to be more accepting of more discriminating DRM systems. This is one of the main things that scares me about Palladium. It starts out innocuously enough, but creates a situation similar to that of the boiled frog:
They say that if you put a frog into a pot of boiling water,
it will leap out right away to escape the danger.
But, if you put a frog in a kettle that is filled with water that is cool and pleasant,
and then you gradually heat the kettle until it starts boiling,
the frog will not become aware of the threat until it is too late.
The frog's survival instincts are geared towards detecting sudden changes.
This is especially important with Palladium given how all data is tied to one particular computer and a particular operating system. Any changes or upgrades and you have to renew the license for all your data. This provides the opportunity for Hollywood to slowly turn up the temperature by restricting usage gradually. You buy a DVD and you own it forever. Purchase Palladium-protected content and you control it, at best, until you have to upgrade your computer or OS, at which point you get to renegotiate your license. You may not like it, and you may even want to leave the system eventually, but by then it will be too late, Palladium makes sure you lose everything if you do.
"Better" DRM, indeed.