In 1998, Congress enacted the Digital Millennium Copyright Act (DMCA) to help prevent widespread copyright infringement online. Congress envisioned a system where copyright owners and internet service providers would work together to combat piracy. Congress’ plan included “safe harbor” provisions that shield service providers from liability if they quickly remove infringing movies, music, or other creative works from their websites after being notified by copyright owners—a process generally known as “notice and takedown.”
After nearly twenty years with the DMCA, the Copyright Office has launched a new study to examine the impact and effectiveness of this system, and voices on both sides of the debate have filed comments expressing their views. For the most part, frustrated copyright owners report that the DMCA has not successfully stemmed the tide of online infringement, which is completely unsurprising to anyone who spends a few minutes online searching for copyrighted works. Unfortunately, some commentators are also pushing for changes that that would make things even more difficult for copyright owners.
To support their position, a surprisingly large number of these commentators cite a recently-released study entitled Notice and Takedown in Everyday Practice, which alleges that nearly 30 percent of takedown requests are “questionable” and makes several policy recommendations that would invariably make it harder for copyright owners to protect their rights online. The problem with this study is that the authors’ suggestions are based on an extremely incomplete and skewed assessment of the notice and takedown system, ignoring the fundamental purpose of the DMCA. And it continues the disappointing narrative that users and service providers are somehow the victims in the grand scheme of things.
Claiming to have produced “the broadest empirical analysis of the DMCA notice and takedown” system to date, the authors provide no data or analysis about whether the system is achieving the critical goal of curbing pervasive online piracy. Their narrow focus on the notice and takedown process, with no mention of the systematic results of that process, reveals a study that is not concerned with improving the notice and takedown system as a whole. The authors seem utterly unconcerned with the incredible frequency of online piracy that necessitates the millions of takedown notices in the first place.
Even assuming that the study’s data is accurate, the samples are severely limited and do not support the authors’ broad conclusions and recommendations. For one thing, the attention-grabbing “30 percent!” headline comes from a sample of takedown notices that consists almost entirely—to the tune of 99.8%—of notices directed at Google Search. At best, this data could tell us something about one company’s experience with one of its products. But it’s certainly not appropriate to draw broad policy recommendations from such a limited sample. Furthermore, as we document in a recent essay, there are numerous reasons to believe that the 30 percent figure is severely overblown.
The authors even admit that the “dominance of Google notices in our dataset limits our ability to draw broader conclusions about the notice ecosystem” and that “the characteristics of these notices cannot be extrapolated to the entire world of notice sending.” Inexplicably, the authors then ignore these limitations and proceed to make policy recommendations that would broadly affect the entire notice and takedown system. Among other things, the authors recommend overhauling the entire statutory damages scheme for copyright infringement, which would apply even in cases where the DMCA is not at issue. The overreach is startling.
The authors want statutory damages reform so that service providers will be more likely to jump into the middle of things when copyright owners send takedown notices. But this is not what Congress had in mind when it passed the DMCA, and the authors otherwise acknowledge that service providers are not well-positioned to make judgments regarding the accuracy of the notices they receive. This is especially true when it comes to fair use, which the authors complain about despite admitting that even they don’t have enough information to actually do the analysis for the uses they flag as being potentially fair.
To justify their policy conclusions, the authors also rely heavily on excerpts from secret survey responses and interviews. Putting aside the obvious problems with making policy recommendations based on unverifiable survey data, the qualitative data they present from the surveys appears woefully restricted and does not include crucial constituencies in the notice and takedown system.
The authors survey twenty-nine service providers, but only six notice senders. Even worse, for the notice senders, they only survey large copyright owners or their representatives, completely excluding smaller copyright owners from their survey data. As a result, the study largely ignores the serious struggle that individual musicians, photographers, filmmakers, and other middle class artists go through in trying to stop the theft of their creative works. It’s simply amazing that, despite this glaring omission, the authors nonetheless make policy recommendations that would have severe consequences for small copyright owners in particular.
The study also strangely dismisses the entire counternotice process, which allows recipients to challenge takedown notices, as ineffective and unused. But apart from limited samples of unidentified survey respondents, the authors offer no empirical data on the frequency or effectiveness of counternotices. The dearth of counternotices in the quantitative study of Google Search is readily explained by the fact that the search engine only includes its own content, and there are no third parties to alert when Google is asked to take something down. But that’s no reason to call the counternotice process a complete failure. One should at least look at how counternotices work when it comes to user-generated content, where there is a third party to contest the takedown notice, before drawing any conclusions.
As the Copyright Office and other stakeholders further explore these issues, we hope they will be wary of one-sided studies that offer broad policy recommendations that go far beyond the underlying data. While Notice and Takedown in Everyday Practice may be an interesting read, it is crucial to understand that it does not provide a foundation to make responsible policy recommendations, and it certainly does not support the authors’ aggressive conclusions and proposals.