Is DRM working?: how could we tell?

Abstract

The success or failure of digital rights management is often taken in legal circles to be a technological question: has a particular scheme already been cracked? How broadly is protected content being redistributed? Can any scheme provide absolute security for content? By these measures, DRM, at least in its most visible applications, has been a failure, as has its legal bulwark, the Digital Millennium Copyright Act. Most widely available schemes are cracked within a few years of release. And due to the nature of the Internet, breaking a scheme once means it is broken everywhere. Under these conditions, absolute security is both required and impossible. This is the so-called "Darknet" hypothesis, first described at the ACM-DRM workshop nine years ago. But the success or failure of DRM and anticircumvention policy generally is also a legal question, or more properly, a question about how law and technology interact with society. Assessing DRM's success therefore requires first determining its place in a copyright landscape that is undergoing a fundamental transformation. That transformation can be described simply as a disappearance of gates. Copyright relies on a world that makes copying without permission costly and difficult. That is, it relies on natural choke points at which access to content can be traded for money. Those natural choke points, or "gates," are disappearing. DRM is an attempt to reestablish a sort of gate, and its success or failure in any given application depends on how well it mimics the real-world gates it is replacing. And that is primarily a social question, not a technological one. Furthermore, it indicates a different set of threats to DRM schemes, and to the policy embodied in the DMCA: to the extent such schemes visibly interfere with common uses, their viability as replacement gates is diminished. The gravest threat to DRM schemes may come not from a particular sophisticated attack, but rather from a dissipation of the illusion of naturalness.