CloudFlare’s Desperate New Strategy to Protect Pirate Sites

Cross-posted from the Center for the Protection of Intellectual Property (CPIP) blog.

CloudFlare has earned a somewhat dubious reputation in the online world. Website owners can set up CloudFlare in just a few minutes, gaining the performance, security, and privacy benefits the service provides. Traffic routed through CloudFlare’s global content delivery network is cached for faster delivery times and protected from numerous online threats. Pirate sites have flocked to the service because it hides their true identities from copyright owners by default. And it probably doesn’t hurt that CloudFlare CEO Matthew Prince thinks that “censoring the Internet” is “creepy,” even “under a court order.”

Prince practices what he preaches, and CloudFlare has been all-too-ready to lend a helping hand to even the most notorious pirates. When The Pirate Bay rose from the ashes in early 2015, CloudFlare provided the site with services that helped manage its massive server loads. CloudFlare’s encryption technology even made it easy for users in the UK to circumvent the High Court’s ban ordering ISPs to block the pirate site. Amazingly, The Pirate Bay is now back in the United States, using its original Virginia-based domain and benefiting from CloudFlare’s robust services to make its criminal enterprise run smoothly worldwide.

Of course, the only reason CloudFlare can get away with supporting the world’s most-visited torrent site is because the DMCA is such a mess. Courts have set the bar so high that CloudFlare wouldn’t likely be found to have red flag knowledge of the massive amounts of infringement it certainly knows its service enables for globally-infamous criminal infringers like The Pirate Bay. Rather than taking the high road and refusing to work with obvious pirate sites, CloudFlare lawyers up when pushed and denies the supportive role that its service provides.

We saw this last year in the Grooveshark case. After the original Grooveshark site was found liable for willful infringement and agreed to shut down, copycat sites sprung up at different top-level domains such as and The plaintiffs obtained a temporary restraining order against the copycats, which registrars Namecheap and Dynadot promptly complied with by disabling some of the domains. But when the plaintiffs asked CloudFlare to stop providing services to some of the other copycats, they were met with firm resistance. The plaintiffs had to turn to the court for an order clarifying that the injunction against the copycat sites prevented CloudFlare from providing them services.

With the backing of the Electronic Frontier Foundation, CloudFlare put up a big fight. It denied that it was in “active concert or participation” with the copycats, which under Rule 65 would have made it bound by their existing injunction. CloudFlare argued that its services were merely passive and that the domains would still remain accessible even if its services were cut off. The district court rejected CloudFlare’s self-serving arguments, noting that it was in fact aiding and abetting the copycat sites by operating their authoritative domain name servers and optimizing their traffic worldwide. Since CloudFlare had actual notice of the injunction and was in “active concert or participation” with the enjoined copycats, it was also bound by their injunction under Rule 65.

Hit with what must have been the eye-opening reality that, under penalty of contempt, it couldn’t knowingly help its enjoined customer engage in the very wrong the court had ordered it to stop committing, one might think that CloudFlare would have become more respectful of court orders involving its customers. However, as recent developments in the MP3Skull case show, CloudFlare has decided to again take the low road in shirking its responsibility to the court. And its argument here as to why it’s beyond the court’s reach is even more desperate than before.

In April of 2015, several record label plaintiffs sued MP3Skull for copyright infringement, easily obtaining a default judgment when the defendants failed to respond to the suit. Earlier this year, the plaintiffs were granted a permanent injunction, which the defendants quickly flouted by setting up shop under several different top-level domains. Naturally, the common denominator of these multiple MP3Skull sites was that they used CloudFlare. The plaintiffs’ lawyers sent a copy of the injunction against the pirate sites to CloudFlare, asking it to honor the injunction and stop supplying services to the enjoined domains. But, as with Grooveshark, CloudFlare again refused to comply.

The record label plaintiffs have now gone back to the district court, filing a motion requesting clarification that CloudFlare is bound by the injunction against the MP3Skull sites. They argue that the “law is clear that CloudFlare’s continued provision of services to Defendants, with full knowledge of this Court’s Order, renders CloudFlare ‘in active concert or participation’ with Defendants,” and they point to the opinion in the Grooveshark case in support. According to the plaintiffs, the only issue is whether CloudFlare is aiding and abetting the enjoined defendants by providing them services.

CloudFlare opposes the motion, though it noticeably doesn’t deny that it’s in “active concert or participation” with the enjoined defendants. Instead, CloudFlare argues that, since this is a copyright case, any injunction against it must comply with the DMCA:

Section 512(j) prescribes specific standards and procedures for injunctions against service providers like CloudFlare in copyright cases. It places strict limits on injunctions against eligible service providers. 17 U.S.C. § 512(j)(1). It specifies criteria that courts “shall consider” when evaluating a request for injunctive relief against a service provider. 17 U.S.C. § 512(j)(2). And it requires that a service provider have notice and an opportunity to appear, before a party may bind it with an injunction. 17 U.S.C. § 512(j)(3). Plaintiffs ignored those requirements entirely.

The gist of CloudFlare’s argument is that Section 512(j) controls injunctions against service providers like itself, notwithstanding the fact that Rule 65 binds those in “active concert or participation” with an enjoined party. In other words, CloudFlare says that the DMCA gives service providers unique immunity from having to obey court-issued injunctions under the Federal Rules—a remarkable claim requiring remarkable proof. And the case law cited to back up this claim? None. Zip. Nada. CloudFlare fails to produce one single cite showing that any injunctive-relief statute, whether copyright or otherwise, has ever been deemed to preempt the longstanding rule that it’s contempt of court to aid and abet an enjoined defendant. The desperation is palpable.

The reason the DMCA doesn’t apply to CloudFlare is obvious. Section 512(j) states that it “shall apply in the case of any application for an injunction under section 502 against a service provider” that qualifies for the safe harbors. CloudFlare goes on for pages about how it’s a service provider that would qualify for the safe harbor defense if given the chance, but all of this misses the point: CloudFlare is not being enjoined. The only service provider being enjoined is MP3Skull—and that injunction was issued under Section 502 without the limitations set forth in Section 512(j) because MP3Skull didn’t even bother to show up and attempt to claim the safe harbors. But the plaintiffs have not sought an injunction against CloudFlare, which they could only do by naming CloudFlare as a party to the suit.

Since CloudFlare itself isn’t being enjoined under Section 502, Section 512(j) provides it no limitations. The issue is simply whether, under the Federal Rules, CloudFlare is bound by the injunction that has already been issued against the MP3Skull sites. Perhaps not wanting to get bench-slapped again on the aiding and abetting question under Rule 65, CloudFlare is taking an even lower road with this desperate new argument that it’s magically immune to court orders against its customers under the Federal Rules. The district court has yet to rule on the plaintiffs’ motion, but my guess is that it will make short work in reminding CloudFlare of the court’s true power to hold aiders and abettors in contempt.

Do As I Say, Not As I Do: Google’s Patent Transparency Hypocrisy

Cross-posted from the Center for the Protection of Intellectual Property (CPIP) blog.

It is common today to hear that it’s simply impossible to search a field of technology to determine whether patents are valid or if there’s even freedom to operate at all. We hear this complaint about the lack of transparency in finding “prior art” in both the patent application process and about existing patents.

The voices have grown so loud that Michelle K. Lee, Director of the Patent Office, has made it a cornerstone of her administration to bring greater transparency to the operations of the USPTO. She laments the “simple fact” that “a lot of material that could help examiners is not readily available, because the organizations retaining that material haven’t realized that making it public would be beneficial.” And she’s been implementing new programs to provide “easy access by patent examiners to prior art” as a “tool to help build a better IP system.”

We hear this complaint about transparency most often from certain segments of the high-tech industry as part of their policy message that the “patent system is broken.” One such prominent tech company is search giant Google. In formal comments submitted to the USPTO, for instance, Google asserts that a fundamental problem undermining the quality of software patents is that a “significant amount of software-related prior art does not exist in common databases of issued patents and published academic literature.” To remedy the situation, Google has encouraged the Patent Office to make use of “third party search tools,” including its own powerful search engines, to locate this prior art.

Google is not shy about why it wants more transparency with prior art. In a 2013 blog post, Google Senior Patent Counsel Suzanne Michel condemned so-called “patent trolls” and argued that the “PTO should improve patent quality” in order to “end the growing troll problem.” In comments from 2014, three Google lawyers told the Patent Office that “poor quality software patents have driven a litigation boom that harms innovation” and that making “software-related prior art accessible” will “make examination in the Office more robust to ensure that valid claims issue.” In comments submitted last May, Michel even proposed that the Patent Office use Google’s own patent search engines for “streamlining searches for relevant prior art” in order to enhance patent quality.

Given Google’s stance on the importance of broadly available prior art to help weed out vague patents and neuter the “trolls” that wield them, you’d think that Google would share the same devotion to transparency when it comes to its own patent applications. But it does not. Google has not mentioned in its formal comments and in its public statements that even using its own search engine would fail to disclose a substantial majority of its own patent applications. Unlike the other top-ten patent recipients in the U.S., including many other tech companies, Google keeps most of its own patent applications secret. It does this while at the same time publicly decrying the lack of transparency in the patent system.

The reality is that Google has a patent transparency problem. Not only does Google not allow many of its patent applications to be published early or even after eighteen months, which is the default rule, Google specifically requests that many of its patent applications never be published at all. So while Google says it wants patent applications from around the world to be searchable at the click of a mouse, this apparently does not include its own applications. The numbers here are startling and thus deserve to be made public—in the name of true transparency—for the first time.

Public Disclosure of Patent Applications

Beginning with the American Inventors Protection Act of 1999 (AIPA), the default rule has been that a patent application is published eighteen months after its filing date. The eighteen-month disclosure of the patent application will occur unless an applicant files a formal request that the application not be published at all. An applicant also has the option to obtain early publication in exchange for a fee. Before the AIPA, an application would only be made public if and when the patent was eventually granted. This allowed an applicant to keep her invention a trade secret in case the application was later abandoned or rejected.

The publication of patent applications provides two benefits to the innovation industries, especially given that the waiting time between filing of an application and issuance of the patent or a final rejection by an examiner can take years. First, earlier publication of applications provides notice to third parties that a patent may cover a technology they are considering adopting in their own commercial activities. Second, publication of patent applications expands the field of publicly-available prior art, which can be used to invalidate either other patent applications or already-issued patents themselves. Both of these goals produce better-quality patents and an efficiently-functioning innovation economy.

Separate from the legal mandate to publish patent applications, Google has devoted its own resources to creating greater public access to patents and patent applications. From its Google Patent Search in 2006 and its Prior Art Finder in 2012 to its current Google Patents, Google has parlayed its search expertise into making it simple to find prior art from around the world. Google Patents now includes patent applications “from the USPTO, EPO, JPO, SIPO, WIPO, DPMA, and CIPO,” even translating them into English. It’s this search capability that Google has been encouraging the Patent Office to utilize in the quest to make relevant prior art more accessible.

Given Google’s commitment to patent transparency, one might expect that Google would at least be content to allow default publication of its own applications under the AIPA’s eighteen-month default rule. Perhaps, one might think, Google would even opt for early publication. However, neither appears to be the case; Google instead is a frequent user of the nonpublication option.

Google’s Secrecy vs. Other Top-Ten Patent Recipients

After hearing anecdotal reports indicating that Google was frequently using its option under the AIPA to avoid publishing its patent applications, we decided to investigate further. We looked at the patents Google received in 2014 to see what proportion of its applications was subject to nonpublication requests. To provide context, we also looked at how Google compared to the other top-ten patent recipients in this regard. The results are startling.

Unfortunately, there’s no simple way to tell if a nonpublication request was made when a patent application was filed using the USPTO’s online databases—nonpublication requests are not an available search field. The same appears to be true of subscription databases. The searches therefore have to be done manually, digging through the USPTO’s Public PAIR database to find the application (known in patent parlance as the “file wrapper”) for each individual patent that includes the individual application documents. Those interested in doing this will find startling numbers of patent applications kept secret by Google, both in terms of absolute numbers but also as compared to the other top-ten recipients of U.S. patents.

By way of example as to what one needs to look for, take the last three patents issued to Google in 2014: D720,389; 8,925,106; and 8,924,993.

For the first patent, the application was filed on December 13, 2013, and according to the application data sheet, no request was made to either publish it early or not publish it at all:

Since no such request was made, the application would normally be published eighteen months later or upon issuance of the patent. Indeed, that is what happened in due course—this patent issued just over one year after the application was filed, as it was concurrently published and issued in December of 2014.

For the second patent in our small set of examples, the application was filed on April 20, 2012. In this case, Google requested nonpublication by including a letter requesting that the application not be published:

Google thus opted out of the default eighteen-month publication rule, and the application was not published until the patent issued in December of 2014, some twenty months later.

Finally, for the third patent, the application was filed on November 10, 2011, and the application data sheet shows that Google requested the application not be published:

Google here again opted out of the default publication rule, and the application was not published until the patent issued in December of 2014—more than three years after the application was filed.

We applied this methodology to a random sample of 100 patents granted to each of the top-ten patent recipients in 2014.

In 2014, Google was one of the top-ten patent recipients, coming in sixth place with 2,649 issued patents:

SOURCES: USPTO PatentsView Database & USPTO Patent Full-Text and Image Database

We randomly sampled 100 patents for each of the top-ten patent recipients for 2014. We reviewed the file wrapper for each to determine the proportion of nonpublication requests in each sample.

Our results revealed that Google is an extreme outlier among top-ten patent recipients with respect to nonpublication requests. Eight of the top-ten patent recipients made zero requests for nonpublication, permitting their patent applications to be published at the eighteen-month deadline. The eighth-ranking patent recipient, Qualcomm, requested that one application not be published. By contrast, Google formally requested that 80 out of 100—a full 80%—of its applications not be published.

The following chart shows these results:

SOURCE: USPTO Public PAIR Database


Based on this result, Google deliberately chooses to keep a vast majority of its patent applications secret (at least it did so in 2014). This secrecy policy for its own patent applications is startling given both Google’s public declarations of the importance of publication of all prior art and its policy advocacy based on this position. It is even more startling when seen in stark contrast to the entirely different policies of the other nine top patent recipients for 2014.

It is possible that 2014 was merely an anomaly, and that patent application data from other years would show a different result. We plan to investigate further. So, stay tuned. But for whatever reason, it appears that Google doesn’t want the majority of its patent applications to be published unless and until its patents finally issue. This preference for secrecy stands in contrast to Google’s own words and official actions.

As one of the top patent recipients in the U.S., you’d think Google would want its applications to be published as quickly as possible. The other top recipients of U.S. patents in 2014 certainly adopt this policy, furthering the goal of the patent system in publicly disclosing new technological innovation as quickly as possible. The fact that Google does otherwise speaks volumes.

Second Circuit Deepens Red Flag Knowledge Circuit Split in Vimeo

Cross-posted from the Center for the Protection of Intellectual Property (CPIP) blog.

The Second Circuit’s recent opinion in Capitol Records v. Vimeo is, to put it mildly, pretty bad. From its convoluted reasoning that copyrights under state law for pre-1972 sound recordings are limited by the DMCA safe harbors, despite the explicit statement in Section 301(c) that “rights or remedies” under state law “shall not be annulled or limited” by the Copyright Act, to its gutting of red flag knowledge by limiting it to the nearly-impossible situation where a service provider actually knows that a specific use of an entire copyrighted work is neither fair nor licensed yet somehow doesn’t also surmise that it’s infringing, it’s hard to see how either result is compelled by the statutes, much less how it was intended by Congress. On the latter point, the Second Circuit in essence has written red flag knowledge out of the statute, reducing the DMCA to a mere notice-and-takedown regime. The reality is that Congress expected red flag knowledge to do far more work, incentivizing service providers to take action in the face of a red flag—even without a notice.

If there’s any good to come from Vimeo, it might only be that the Second Circuit has now deepened the circuit split with the Ninth Circuit in Columbia Pictures v. Fung on two issues related to red flag knowledge. Under the statute, red flag knowledge exists when a service provider is “aware of facts or circumstances from which infringing activity is apparent.” The two circuits are already split on the issue of whether red flag knowledge must pertain to the particular works that are being sued over in the suit. And now with Vimeo, the circuits are split on the issue of whether a service provider can gain red flag knowledge just by looking at an infringing work. The deeper the circuit split, the greater the chance an appeal will make it to the Supreme Court, which would hopefully clean up the current red flag knowledge mess.

In Fung, the defendant, Gary Fung, operated several piracy havens, including isoHunt, TorrentBox, Podtropolis, and eDonkey. The district court found Fung liable for inducement under MGM v. Grokster and denied him safe harbor protection under the DMCA. The district court’s decision came in 2009, two years before the Ninth Circuit first held in UMG v. Shelter Capital that red flag knowledge requires “specific knowledge of particular infringing activity.” It also came two-and-a-half years before the Second Circuit held in Viacom v. YouTube that red flag knowledge is only relevant if it pertains to the works-in-suit. Regardless, since the vast majority of content available on Fung’s sites was copyrighted, including specific content that he himself had downloaded, the district court held that Fung hadn’t even raised a triable issue of fact as to whether he had red flag knowledge. The fact that none of the works he had been sued over were the same as the ones he had been found to have red flag knowledge of was irrelevant.

On appeal, the Ninth Circuit affirmed the district court’s holding that Fung had red flag knowledge as a matter of law. The opinion came out just one week after the same panel of judges issued a superseding opinion in UMG v. Shelter Capital reiterating that red flag knowledge requires “specific knowledge of particular infringing activity.” Importantly, in applying that standard to Fung, the Ninth Circuit did not say that the specific knowledge had to be of the particular works-in-suit. For whatever reason, Fung had failed to argue otherwise. Google even filed an amicus brief supporting the plaintiffs but nonetheless arguing that “the DMCA’s knowledge standards are specific and focus on the particular material that the plaintiff is suing about.” Apparently unaware that this actually helped his case, Fung filed a supplemental brief calling Google’s argument “fallacious.”

In the Ninth Circuit’s opinion, even though red flag knowledge had to relate to particular infringing activity, that activity did not have to involve the particular works-in-suit. Moreover, the Ninth Circuit held that the “material in question was sufficiently current and well-known that it would have been objectively obvious to a reasonable person” that it was “both copyrighted and not licensed to random members of the public.” Since Fung failed to expeditiously remove the particular material of which he had red flag knowledge, he lost his safe harbor protection across the board. Thus, the Ninth Circuit in Fung held that: (1) red flag knowledge that strips a service provider of its entire safe harbor protection does not have to pertain to the particular works-in-suit, and (2) material can be so “current and well-known” that its infringing nature would be “objectively obvious to a reasonable person.”

The Second Circuit in Vimeo parted ways with the Ninth Circuit on these two holdings. Since the “evidence was not shown to relate to any of the videos at issue in this suit,” the Second Circuit held that it was “insufficient to justify a finding of red flag knowledge . . . as to those specific videos.” The Second Circuit thus applied the red flag knowledge standard on a work-by-work basis, in direct contrast to the Ninth Circuit in Fung. Also, the Second Circuit held that “the mere fact that a video contains all or substantially all of a piece of recognizable, or even famous, copyrighted music” and was “viewed in its entirety” by an “employee of a service provider” was not enough “to sustain the copyright owner’s burden of showing red flag knowledge.” The court added that even “an employee who was a copyright expert cannot be expected to know when use of a copyrighted song has been licensed.” So while the Ninth Circuit said it would have been objectively obvious to Fung that particular works were infringing, the Second Circuit in Vimeo set the bar far higher.

Curiously, the Second Circuit in Vimeo didn’t even mention Fung, despite the fact that it was deepening the circuit split with the Ninth Circuit. One wonders whether the omission was intentional. Either way, the circuit split has only gotten deeper. While in the Ninth Circuit an infringement can be so obvious that a court can find that a service provider had red flag knowledge without even sending it to a jury, the Second Circuit says that courts can’t let a jury decide whether a service provider had red flag knowledge even with the most obvious of infringements. And while in the Ninth Circuit a service provider loses its entire safe harbor for failing to remove an obvious infringement that it hasn’t been sued over, the Second Circuit says that red flag knowledge has to be determined on a work-by-work basis for only the works-in-suit. Given this growing divide between the Second and Ninth Circuits, it seems like only a matter of time before the Supreme Court will weigh in on the red flag knowledge standard. And if the Court does finally weigh in, one hopes it will put common sense back into the DMCA.

How Strong Patents Make Wealthy Nations

Cross-posted from the Center for the Protection of Intellectual Property (CPIP) blog.

By Devlin Hartline & Kevin Madigan

How did the world’s wealthiest nations grow rich? The answer, according to Professor Stephen Haber of Stanford University, is that “they had well-developed systems of private property.” In Patents and the Wealth of Nations, recently published in the CPIP Conference issue of the George Mason Law Review, Haber explains the connection: Property rights beget trade, trade begets specialization, specialization begets productivity, and productivity begets wealth. Without a foundation of strong property rights, economic development suffers. But does the same hold true for intellectual property, particularly patents? Referencing economic history and econometric analysis, Haber shows that strong patents do indeed make wealthy nations.

Before diving into the history and analysis, Haber tackles the common misconception that patents are different than other types of property because they are monopolies: “It is not, as some IP critics maintain, a grant of monopoly. Rather, it is a temporary property right to something that did not exist before that can be sold, licensed, or traded.” The simple reason for this, Haber notes, is that a patent grants a monopoly only if there are truly no substitutes, but this is almost never the case. Usually, there are many substitutes and the patent owner has no market power. And the “fact that patents are property rights means that they can serve as the basis for the web of contracts that permits individuals and firms to specialize in what they do best.”

Turning back to his claim that strong patents make wealthy nations, Haber presents data showing the relationship between the strength of enforceable patent rights and the level of economic development across several different countries. The results are remarkably clear: “there are no wealthy countries with weak patent rights, and there are no poor countries with strong patent rights.” The following figure shows how GDP per capita increases as patent rights get stronger:

Haber - Figure 1

Of course, while it’s clear that patent strength and GDP per capita are related, it’s possible that the causality runs the other way. That is, how do we know that an increase in GDP per capita doesn’t foster an environment where patents tend to be stronger? This is where the evidence from economic historians and econometric analysts comes into play. Exploring what economic history has to tell us about the impact of patent laws on innovation, Haber asks whether the Industrial Revolution was bolstered by the British patent system and whether the United States emerged as a high-income industrial economy because of the U.S. patent system.

To the first question, Haber notes that the consensus among historians is that “from at least the latter half of the eighteenth century, the patent system promoted the inventive activity associated with the Industrial Revolution.” He then cites the recent book by Sean Bottomley that carefully shows how “many of the changes to Britain’s patent laws and their enforcement—the requirement for detailed specifications, patents conceived as property rights, the emergence of patent agents—all preceded, rather than followed, the onset of industrialization.” Haber also cites a research paper by Petra Moser, which finds that countries in the nineteenth century with weak patent systems trailed both Britain and the United States in technological development.

Moving to the United States, Haber notes that three generations of economic historians have agreed that just after it gained independence, the country’s strong patent system played a pivotal role in fomenting the remarkable industrial developments that soon followed. After pointing out that the United States was the first country to call for a patent system in its Constitution, Haber compares the GDP per capita for the United States, Britain, and Brazil from 1700 to 1913. The following figure shows just how quickly the agrarian American colonies caught up with, and ultimately surpassed, Britain in GDP per capita, while the GDP per capita of Brazil, a country that became independent at about the same time but had no patent system, stagnated:

Haber - Figure 2

As the figure shows, the GDP per capita in the United States and Brazil were less than half that of Britain in 1700, and by 1913, the United States had overtaken Britain as both countries left Brazil far behind. Noting that “there is uniformity of views among economic historians that the U.S. patent system played a large role” in this success, Haber provides specifics examples of improvements upon the British patent system that contributed to it, including broad access to property rights in technology through low fees and a routine and impersonal application process under the Patent Act of 1790. He goes on to highlight the importance of major reforms to the U.S. patent system introduced in the Patent Act of 1836, including the examination process that “reduced concerns third parties might have had about a patent’s novelty, thereby facilitating the evolution of a market for patented technologies.”

The second half of the nineteenth century saw the development of an active market for inventions in the United States, leading to the emergence of a class of specialized and independent inventors as well as patent brokers, patent agents, and patent attorneys, who would connect the inventors with manufacturers looking to buy or license new technologies. While some of these intermediaries were derided, much like the “patent trolls” of the twenty-first century, as “patent sharks,” Haber contends that this market for inventions played a critical role in the emergence of new industrial technologies and centers: “[O]ne would be hard pressed to make the case that patents in the nineteenth century, or the intermediaries who represented their inventors, did anything but facilitate the rapid development of American manufacturing.”

Haber then shifts his focus to econometric analysis, examining the different ways that economic scholars research the relationship between patent rights and economic progress in different countries over a period of time. He stresses that accurate econometric estimation of causal relationships is a relatively young area of inquiry requiring considerable care. He uses the example of a widely-cited study by Josh Lerner, which looks at “whether the strengthening of patents affects the rate of change of innovation in an economy within a two-year window after a patent reform.” Haber points out that many changes neither begin nor end so quickly. With laser technology, for example, “follow-on innovations” have developed “over decades, not two-year windows,” and Lerner’s study thus discounts much innovation.

Looking at studies that utilize a “very long time dimension,” Haber cites one finding that “there is a significant positive effect of patent laws on innovation rates” and another finding that “patent intensive industries in countries that improve the strength of patents experience faster growth in value added than less patent-intensive industries in those same countries.” Haber praises a recent study by Jihong Zhang, Ding Du, and Walter G. Park, who “not only find that there is a positive relationship between the strength of enforceable patent rights and innovation in developed economies, but that that relationship holds for underdeveloped economies as well.”

In sum, Haber states that “there is a critical mass of multi-country studies” that leads to two conclusions:

First, there is a causal relationship between the strength of patent rights and innovation. Second, this relationship is non-linear: there are threshold effects such that stronger patent rights positively impact innovation once a society has already reached some critical level of economic development. The reason for the non-linearity probably resides in the fact that innovation is not just a product of the strength of patent rights, but of other features of societies, which are necessary complements, that tend to be absent at low levels of economic development.

Finally, Haber looks at whether the innovation landscape of the twenty-first century is somehow so different that the lessons from economic history and econometric analysis no longer apply. In particular, he questions whether the emergence of patent licensing firms, sometimes called “patent assertion entities” or “PAEs,” and the alleged strategic behavior of “patent holdup” with standard-essential patents (SEPs) are really new features of the U.S. patent system that might hinder innovation. Haber concludes that the evidence shows that neither PAEs nor patent holdup is hindering innovation. In fact, there’s little reason to think that patent holdup even exists.

Haber takes on the recent study by James Bessen and Michael Meurer, which claims that PAEs are a new phenomenon that “constitute a direct tax on innovation” to the tune of “$29 billion per year.” This claim has been rebutted, Haber notes, by scholars such as B. Zorina Khan, whose recent study shows that many great inventors of the nineteenth century were themselves PAEs. Haber further cites the recent paper by David L. Schwartz and Jay P. Kesan that carefully demonstrates fundamental problems with Bessen and Meurer’s methodology, including selection bias, the conflation of “costs” with “transfers,” the lack of a benchmark for comparison, and the failure to even consider the benefits of PAE activity.

Turning to patent holdup, Haber points out that products have long been comprised of numerous patented innovations, and he cites a recent paper by Adam Mossoff showing that there’s nothing “new about firms whose sole source of revenue comes from the licensing of essential patents.” As to evidence that innovation is hindered by patent holdup, Haber notes that the “theoretical literature” says it’s possible, but the “evidence in support of this theory, however, is largely anecdotal.” Haber then cites his recent study with Alexander Galetovic and Ross Levine, which looks at the “extensive economics literature on the measurement of productivity growth” and shows that “SEP holders” are not able “to negotiate excessive royalty payments” as predicted by the patent holdup theory.

In conclusion, Haber acknowledges that while “no single piece of evidence” should “be viewed as dispositive,” it’s certainly quite “telling that the weight of evidence from two very different bodies of scholarship, employing very different approaches to evidence—one based on mastering the facts of history, the other based on statistical modeling—yield the same answer: there is a causal relationship between strong patents and innovation.” Haber then challenges the naysayers to make their case: “Evidence and reason therefore suggest that the burden of proof falls on those who claim that patents frustrate innovation.” Given the copious evidence showing that strong patents make wealthy nations, the IP critics have their work cut out for them.

For a PDF version of this post, please click here.

Capitol Records v. Vimeo: Courts Should Stop Coddling Bad Actors in Copyright Cases

Here’s a brief excerpt of my new post that was published on IPWatchdog.

Here’s where we are after Capitol Records v. Vimeo: A service provider can encourage its users to infringe on a massive scale, and so long as the infringement it encourages isn’t the specific infringement it gets sued for, it wins on the safe harbor defense at summary judgment. This is so even if there’s copious evidence that its employees viewed and interacted with the specific infringing material at issue. No jury will ever get to weigh all of the evidence and decide whether the infringement is obvious. At the same time, any proactive steps taken by the service provider will potentially open it up to liability for having actual knowledge, so the incentive is to do as little as possible to proactively “detect and deal” with piracy. This is not at all what Congress intended. It lets bad faith service providers trample the rights of copyright owners with impunity.

To read the rest of this post, please visit IPWatchdog.