von Lohmann: First Post*

Good morning, and thanks for coming. I thought I'd start by trying to summarize the heart of the argument in 500 words. Here goes...

The proponents of the Digital Millennium Copyright Act (DMCA) hoped that technical protection measures (TPMs, aka DRM) designed to control access to and duplication of copyrighted works would restrain infringement and thereby encourage the entertainment industries to make their wares available in the digital world (or, at minimum, protect their incentives from substantial erosion).

The notion was not that TPMs would be invulnerable, but rather that TPMs would be a “speed bump” sufficient to keep “honest people honest.” TPMs would recreate for end-users one important constraint they had experienced in the pre-digital world—if they were going to copy and distribute movies or music, they would have to find an intermediary to assist them. Before the digital age, the intermediaries were copy shops and tape duplicators; in the TPM era, the intermediaries would be vendors of circumvention tools (“black boxes”). Legal enforcement could then focus on this smaller universe of necessary intermediaries.

Has it worked?

In Measuring the DMCA Against the Darknet, I argue that advances in digital distribution technologies during the last 7 years pose a fundamental challenge to the reliance on TPMs and legal prohibitions on circumvention to solve copyright’s “digital dilemma.”

The argument is rooted in the insights set out in the now-infamous Microsoft Darknet paper. That paper sets out three assumptions about the modern digital environment:

1. Any widely distributed object will be available to some fraction of users in a form that permits copying.
2. Users will copy objects if it is possible and interesting to do so.
3. Users are connected by high-bandwidth channels.

These insights, taken together, render obsolete the basic mechanism on which the efficacy of the DMCA’s anti-circumvention regime depends. The first assumption is simply another way of saying that no TPM has yet been developed, nor is one likely to be developed, that is invulnerable against an expert attacker. The remaining assumptions imply that, once compromised by a sophisticated attacker, a TPM is effectively useless at further restricting the widespread redistribution because users have the capability to rapidly duplicate and propagate the now-unprotected work. In other words, in light of modern digital distribution technologies, all it takes is “one leak” to neutralize a TPM entirely. Rightsholders are left no better off (from an enforcement point of view), than they were without a TPM.

Consequently, in light of the the Darknet assumptions, it appears that legislators may have chosen to regulate the wrong thing (namely, TPM circumvention) when enacting the DMCA's anti-circumvention provisions. If so, the error is particularly grievous in light of the mounting evidence that those provisions are inflicting serious collateral damage on other public values, including scientific research, free speech, innovation, fair use and competition. Moreover, it suggests that policy-makers should be exploring different mechanisms to respond to the “digital dilemma,” rather than continuing to press for DMCA-like anti-circumvention laws around the world.

Comments

[Litman]

Fred, if legislators chose to "regulate the wrong thing" in enacting section 1201, what would have been the "right thing" to regulate to advance the putative goals of the DMCA?

[Bruce]

If the DMCA isn't working, how come I can't find any user-friendly DVD-ripping devices at Circuit City? The flaw of the Darknet paper seems to be its assumption that when something is available only on the darknet, that is not itself an effective "speed bump" to massive consumer adoption. By that logic, because there are fences who sell stolen goods, store owners should not bother preventing shoplifting.

In particular, the Darknet paper seems to refer to the "efficiency" of the darknet in two different ways, and while there may be a connection between them, I don't think it can be assumed. On page 2, the "efficiency" of the darknet is described as "measured by aggregate library size and available bandwidth." That seems to be a technological definition of efficiency -- the amount of content the darknet is capable of storing and transmitting. Call that Efficiency-1. Later, on page 12, however, the authors state that "[i]f the darknet is efficient, then content will be rapidly propagated to all interested peers." This is trivially true if by "interested peers" the authors simply mean those who are connected to the darknet. But that doesn't lead to the conclusions of the paper about the difficulty of resisting the darknet unless there are a lot of "interested peers." There seems to be a hidden assumption that if a darknet is efficient in terms of storage and bandwidth, it is also efficient in terms of delivering content to a large portion of those consumers who would be interested in receiving it. Call that Efficiency-2. I do not see how Efficiency-1 leads necessarily to Efficiency-2, at least not with a non-trivial definition of "peers." You seem to make the same jump in your paper on page 7 when you conclude that "all it takes is 'one leak' to neutralize a TPM entirely." The Darknet paper supports this conclusion only if the "interested peers" on page 12 are a large portion of all consumers interested in a particular piece of content. If it is a smaller portion, then isn't it possible that the TPM, Section 1201, and the transaction costs and risks involved in visiting the darknet, are responsible for the difference?

[Fred von Lohmann]

Bruce, you're quite right about the two forms of efficiency. However, it seems empirically pretty clear that "a large portion of those consumers who would be interested in receiving it" do, in fact, have ready access to Darknet channels. First, there are, by most estimates, over 200 million people using public P2P worldwide. P2P file sharing software is easy to find and download, and there is (as yet) no chance of enforcement against you so long as you only download from other global users.

Perhaps more important is the rise of other Darknet channels, most obviously optical disk like CD-R and DVD-R. Even the most unsophisticated users have access to these tools, and can use them to share content easily.

So I think the Microsoft Darknet authors were on safe ground when predicting that a large proportion of interested end-users will continue to have easy access to Darknet channels. In other words, the point is not that everyone will have the expertise to build their own DVD ripper, but rather that everyone will have P2P file sharing software, portable hard drives, and recordable optical media.

[Fred von Lohmann]

(Jessica) Litman,

Your question deserves a longer answer, which I will post later this week. But, at minimum, the following forms of regulation should continue to have some efficacy (setting aside any analysis of their wisdom) in a Darknet world:

1. counter-measures aimed at impeding Darknet channels (spoofs, surveillance, watermarks, filtering, tech mandates on storage and internet technologies, etc);
2. more enforcement against end-users;
3. voluntary collective licensing; and
4. compulsory licensing/levies.

There is also the option of simply competing with free, but that does not, by itself, require additional regulation (though some of the regulatory mechanisms above could potentially assist).

[Bruce]

Thanks, Fred. I agree the number of P2P users making unauthorized copies of copyrighted works is large in absolute terms. But 200 million worldwide seems like a small fraction of all content users. Even if we assume TPMs are useless for 200 million people worldwide -- which I think is a big assumption -- that still leaves a pretty large market.

As for spreading content via recordable media, that seems unlikely to lead to a massive darknet. It seems exactly like copying videocassettes; and as you know, despite the initial fears of the movie industry, it turned out that the hassle and expense of copying videocassettes dampened any negative impact to a manageable level. Why would the impact of copying to optical media be any different?

[Douglas Lay]

> If the DMCA isn't working, how come I can't find any user-friendly DVD-ripping devices at Circuit City?

Yep. Even if you accept that Section 1201 has been a total failure at keeping content off the darknet (I certainly accept this), the impact on so-called "schoolyard copying" also must be considered. I know several people who are avid CD rippers and owners of DVD burners who do not engage in any ripping of copy-protected DVDs. Would they if DVD ripping tools were available at places like Circuit City or (much better yet) bundled with the DVD burner driver software?

Interesting to see that just around the same time Mr. von Lohmann's paper appears, we see the RIAA citing research that shows schoolyard copying is a greater threat to the record industry's bottom line than file sharing.

http://www.siliconvalley.com/mld/siliconvalley/news/editorial/12377388.htm

[bill]

Bruce, you can find them. They have computers, DVD drives, and ripping software, all available for your purchase. They also have machines that you can use to copy DVDs to DVD! What other kind of machine could you want? I can't imagine there's much efficiency in a stand-alone box where DVDs go in and DivXs come out when regular computers do this so well.