EFF and Study Authors Refute PFFs ‘Junk Science’ Claims

The Progress & Freedom Foundation (PFF) had some choice words expressed by a study on DMCA notices received by Google. While the posting was very pointed, the authors of the study Thomas Sydnor accused of being “junk science”, as well as the EFF, gave ZeroPaid a response to the many accusations made by the author of the blog posting

Note: This is an article I wrote that was published elsewhere first. It has been republished here for archival purposes

The posting is known for being supporters of the RIAA, the MPAA and other organizations that represent a point of view that copyright should be more restrictive.

The PFF Argument

Sydnor wrote, “The Internet can contribute significantly to the quality and quantity of debate about important questions of public policy. But–particularly when copyrights are at issue–it can also amplify and perpetuate the worst nonsense spouted by the least competent employees or agents of even generally well-intended entities.”

Sydnor discussed a document by Google during a hearing in New Zealand to discuss copyright related issues. He cites a small footnote as well as a comment by a representative which was traced back to a summary of a study regarding DMCA notices and the chilling effects of the DMCA. The claim was, according to Sydnor, that 37% of the DMCA notices issued to Google were problematic.

Sydnor responded, saying, “This claim is nonsense: That 37% figure derived from an egregious misreading of an absurdly aggressive “summary” of a “study” that was, itself, so biased that it could find even a takedown notice directed at inarguably infringing conduct to be “substantively flawed.””

Sydnor said, directed at the summary, that “not only does the Urban/Quilter summary repeatedly contradict the actual study, it also repeatedly spouts blatant nonsense. […] In short, the Urban/Quilter “summary” of the Chilling Effects study distorts law and the study itself so badly that even foreign lawyers only dimly aware of the basics of U.S. law probably should have realized that this summary was unreliable.”

Sydnor then says, “In conclusion, the Google comments to New Zealand were not only dead wrong, they were also hyper-technical at best, and self-destructive at worst. Nevertheless, the usual cheerleaders-for-piracy will surely claim that these were mere technical errors because the Chilling Effects study itself would support claims that 31% of DMCA notices are “substantively flawed.” See Chilling Effects at 667.

Don’t bother, piracy cheerleaders. Even had Google correctly reported its results, the entire Chilling Effects study would still be useless junk science.”

Sydnor concluded, “The “scholarship” of the copyright-hating Free Culture Movement has become so absurdly biased against creators and copyright enforcement that refuting it is often even less challenging than shooting fish in a barrel. After all, fish in a barrel do not exacerbate their predicament by obliviously pinning “Shoot-Me” signs upon their own scaly backsides.”

Fred Von Lohmann from the EFF Responds

“In my view,” Fred Von Lohmann, Senior Staff Attorney for the Electronic Frontier Foundation (EFF) replied, “this is essentially name-calling. The study speaks for itself — Google simply cited it as one example in a much lengthier document.”

“As for the potential copyright liability of search engines,” Lohmann told Zeropaid, “Mr. Sydnor overstates his point.”

Lohmann explained some of the case law behind the DMCA, “Most copyright experts agree that providing hypertext links to infringing material (whether in a search engine or otherwise) can give rise to copyright infringement liability where the link provider knows or should know that the linked material is infringing (Tom doesn’t even cite the most important case on this, Perfect 10 v. Amazon.com). But search engines that qualify for the DMCA safe harbors cannot be hit with monetary liability for it. There is lots of debate about the precise contours of the DMCA safe harbors (notice-and-takedown, removing links when it has actual or “red flag” knowledge, etc), but no one seriously debates that the DMCA safe harbors were intended to include search engines (Yahoo was a chief architect).”

“How offshore infringements interact with copyright liability for search engines and the DMCA safe harbors is a complicated subject, and one that has received almost no attention in the courts. There’s certainly nothing approaching legal consensus about it, despite some comments by the judge in the recent Isohunt case (cited by Sydnor).” Lohmann added.

Sydnor did say that the Chilling Effects Study was biased because of the samples given for one. Lohmann responded to that criticism.

“Frankly,” Lohmann told ZeroPaid, “if Sydnor is so concerned about accurate statistics on the number of bogus DMCA takedowns (and there have been many, many documented reports of bogus takedown notices), he should be pushing to copyright owners provide copies of all of their notices to Chilling Effects. Then we’d have a data set that would give us solid insights!”

Summary Authors Also Respond

Jennifer M. Urban, one of the two authors of the Summary, “Efficient Process or “Chilling Effects”? Takedown Notices Under Section 512 of the Digital Millennium Copyright Act”, the Summary Sydnor criticized so heavily also responded.

Urban agreed with Lohmann that Sydnor missed the most relevant case, Google v. Perfect 10. “The Napster, Grokster, and Aimster cases all involved peer-to-peer filesharing software or services as opposed to the basic search engine (Google) we were studying.” Urban told ZeroPaid. She added, “There are quite important differences in both the technology and the actions of the defendants in peer-to-peer cases, as compared to search engines like Google or Yahoo. For example, Grokster was found to intentionally induce infringement by users, Aimster showed “willful blindness” to infringements, and Napster had actual knowledge of specific infringements by users who were using its software to pass information back and forth through its indexing server. The caselaw has changed some since we wrote our paper, but these differences largely still remain.”

We asked Urban to give a brief explanation about her study and she responded, “We set out to review DMCA takedown notices in order to understand what we could about how they were being used, whether they were working for copyright holders and whether they were being misused. Takedown notices are very difficult to study, because for the most part, they are simply private correspondence between copyright holders, service providers and in some cases, alleged infringers, and are not available to researchers. We were fortunate in this instance to have access to all the takedown notices Google had received for a certain time period, so we were able to look at that whole landscape of notices. Certainly there are limitations to the study–for example, because we largely had notices from Google, which were mostly targeted at its search engine, we couldn’t tell anything about notices to webhosts or other service providers. We were, we think, very clear about these limitations in the study. At the same time, because we had all the notices from Google over the time period, there was not a bias in the selection–just a limit on how much we could compare them to other situations.

The finding that roughly a third of the notices were problematic was based on a relatively conservative set of criteria. For example, we counted only the clearest types of possible fair uses. Basically, we counted notices that presented a clear question for a court–situations where you would want some external review rather than automatic takedown. We certainly welcome people to read our study and to let us know their comments.”

Urban also clarified that there was more than the dataset in the Chilling Effects clearing house that was used in the study. “I explained the Google dataset below, which was the bulk of our research project. We also, however, looked at notices that individuals had self-submitted to Chilling Effects. We kept this sets separate from the Google set for almost all purposes and were cautious with it. This is because the self-submitted notices likely do have selection bias–they were submitted by people who presumably had various reasons for doing so, and we had no insight into their motives, etc. For this reason, we separated them out. They did not make any difference, from a statistical perspective, in the overall percentage of notices with substantive flaws–it was 31% with the self-reported notices and 29% without them. Either way, it was about a third.” Urban said.

After Thoughts

It should be noted that discussion on this issue was encouraged. We couldn’t agree more because copyright infringement claims, in the face of a massive push by the major entertainment industries to implement a three strikes law everywhere in the world through agreements such as ACTA, should be carefully examined. After all, in some instances, even in France at one point before the constitutional courts forced France to have judicial oversight on disconnections, a three strikes law has been seen as little more than three accusations and your out. How reliable are copyright infringement claims anyway? What if Yahoo filed three copyright infringement claims against Google? If they were invalid, should Google be disconnected from the internet? What about someone who has personal issues with another person online? If one were to make three fake copyright infringement claims against another individual, would that warrant a disconnection against the accused? What safeguards will there be against potential abuse?

DMCA notices can be a very good way of asking these important questions. Are the DMCA notices legally sound or are they filed by people who have limited knowledge of the applicable laws? Certainly if a third of the DMCA notices received by Google appear to be legally problematic, that’s enough to raise some concern especially when countries are considering implementing a three strikes law.

This is definitely an issue worth discussing.

Drew Wilson on Twitter: @icecube85 and Google+.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: