Dissolving Innovation in Meltwater: Copyright and Online Search
Internet search engines serve an ever-more-pressing need. They rely heavily on leeway under copyright law; in the United States, that primarily means fair use. A 2013 decision, Associated Press v. Meltwater, advances an unsupported, highly limiting interpretation of fair use as it applies to search. Of particular concern, the court artificially constrains which services meet the legally favorable standard of having a transformative purpose. The ruling also provides too much protection for highly factual works. This article offers a detailed critique of the court's analysis and a discussion of the ruling's broader context and significance for news producers and search firms.
The amount of information available online is astounding and still growing at an exponential rate. Cisco estimates Internet user downloads in 2013 totaled 513 exabytes (EB), or 513 quintillion (1018) bytes.1 By comparison, “a digital library of all books ever written in any language” is about 0.4 EB of data, and “a transcript of all words ever spoken” would take up about 5 EB of space.2 Users are expected to download 1,457 EB worldwide by 2018.3 This volume of data is beyond comprehension, and the volumes transmitted in future decades will surely dwarf these numbers. Making sensible use of even a tiny sliver of this data requires increasingly powerful search technologies. Primarily by being the most popular search engine, Google has become the second largest company in the world4 and the most visited website,5 obvious indicators of the immense value users and investors place in search.
Many copyright holders, however, have much cooler feelings about search engines, for several reasons. Most obviously, the more data a search engine can help locate, the more attractive it is to end users—and the more likely it is to link to infringing content. Further, even the process of creating a searchable database of the web runs up against the problems of making unauthorized copies of online works in order to be able to search the content more rapidly and deliver relevant excerpts. Finally, copyright holders who had become accustomed to selling lucrative analog bundles of works resent seeing those collections unbundled into less-lucrative, smaller-scale transactions. Whereas record labels primarily blame digital singles retailers like iTunes and now music streaming services for this unbundling,6 newspapers can largely blame search engines.
Driven by a fear that their markets are being usurped and their right to demand licensing fees is being outpaced by technology, several plaintiffs have gone to court to try to use copyright to slow down search engines’ use of their content.7 While other causes of action such as “hot news” have been attempted against news aggregators,8 it is still primarily copyright law that determines how far search engines can go in helping users identify, sort, and retrieve relevant content.
In Associated Press v. Meltwater,9 the tussle over copyright and content aggregation comes to a head. Meltwater Holdings, a news monitoring service, scrapes online content based on users’ search parameters. Believing that this business competes unfairly by taking an excessive amount of copyrighted content from the Associated Press (AP) licensee news websites, the AP sued, alleging copyright infringement and related causes of action. Meltwater raised several affirmative defenses, the most important being fair use. In March of 2013, Southern District of New York Judge Denise L. Cote ruled overwhelmingly for AP. This ruling is “the first time a district court has ruled against a news aggregator by denying a fair use defense.”10 While the ruling features extensive analysis, the decision is misguided on several important counts, with implications reaching far beyond the parties involved.
This paper offers a detailed critique of the fair use analysis in the Meltwater ruling. After a very brief discussion of the facts of the case, I give a quick overview of the statutory and case law around fair use, followed by four sections discussing the Meltwater ruling's failings in the application of the four fair use factors, with special emphasis on the first and second factors. I then bring these factors together to show why the court should have ruled for Meltwater, followed by some concluding thoughts that include fears about the ruling's significance.
Facts of the Case
Meltwater Holdings is a news monitoring service with “offices in 27 countries and 20,000 companies as customers.”11 For substantial fees, the service will help clients—primarily communication professionals—to monitor a set of news outlets for one or more sets of terms at the user's direction. The company uses web crawlers to search and archive these web outlets to create a searchable database. A user may perform an ad hoc search for stories on a given topic, or she may create a set of customized, recurring searches, with results e-mailed on a regular basis. Within these e-mails, each search result generally includes the headline, a link to the article, source information, and two excerpts. “The first excerpt consists of up to 300 characters … from the opening text of the article or lede. The second excerpt is shorter and is known as the ‘Hit Sentence.’ It is approximately 140 characters … ‘surrounding a single, algorithmically chosen appearance of one of the customer's matched search keywords.’”12
Except for the length of the excerpts, both the ad hoc search and e-mail digest functions are also features of Google News.13 Meltwater also provides additional services, such as allowing users to create and save longer excerpts or complete articles by visiting the target websites. Users can also perform analytics on these searches to glean quantitative insights about search results. The final feature is a newsletter and newsfeed service, where, for “an additional fee, Meltwater News assists its subscribers in creating their own newsletters [based on content] that has been saved in a subscriber's archive folder.… Alternatively, subscribers can elect to have their search results incorporated into a Newsfeed on their internal or external website.”14
The AP, founded in 1846 and one of the most visible brand names in journalism, is “a not-for-profit news cooperative owned by its American newspaper and broadcast members” with staff operating “in more than 280 locations worldwide.”15 Due in no small part to its dominant market share in the US news market, a large portion of Meltwater's news results are AP stories. This dominant market position also allows AP to sell services to third parties that include delivery of similar excerpts or of complete articles. It is these licensing fees that are most directly endangered by services like Meltwater, thus leading AP to sue. The case centers on issues of copyright, and the most important question is whether Meltwater's use of AP's articles in this way constitutes fair use.
The undisputed facts of the case set up a decision that, in essence, hinges on a single legal question: Does Meltwater's use of AP's articles qualify as a fair use? Meltwater conceded that AP is the copyright holder in the disputed articles and that Meltwater took nontrivial amounts from these articles in its news alerts. Meltwater contends, however, that their use of these portions is permissible under the affirmative defense of fair use. The relevant statute reads, in part, “the fair use of a copyrighted work … for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright.” It then sets out four nonexclusive factors to determine what counts as a fair use:
the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
the nature of the copyrighted work;
the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
the effect of the use upon the potential market for or value of the copyrighted work.16
All of the factors, and any other relevant considerations, are to be weighed together under an “equitable rule of reason,”17 with “the results weighed together, in light of the purposes of copyright.”18 The statute, part of the Copyright Act of 1976,19 was intended to encapsulate the case law to that point, and case law has continued to flesh out the meaning of each of the four factors in the decades since. The application of relevant case law is thus a particularly labor-intensive task when, as in this case, the lack of an exact analog leaves some room for picking and choosing which of a panoply of cases to invoke and how best to apply them to the present conflict.
In each of the following four sections, I summarize and respond to the ruling, on each successive fair use factor. I then apply the four factors in arguing for a different conclusion.
First Factor: Purpose of Use
The first factor is generally interpreted along two tracks: whether the use of the copyrighted work is transformative and whether it is commercial.20 The court's reasoning is easy enough to summarize and may seem sound, but there are at least three reasons readers should find it deeply unsatisfying. The decision's attempt to differentiate from relevant case law is strained and incomplete. Further, the court's analysis of click-through rates is too important to the decision and also shows a limited understanding of how Internet traffic flows actually work. Finally, the court's analysis on transformativeness relies on an equivocation between individual articles and AP's own searchable database of news.
The Meltwater Court on Purpose of Use
On whether a use is transformative, the court emphasizes that “a ‘use … that merely repackages or republishes the original is unlikely to be deemed a fair use’ and a ‘change of format, though useful’ is not transformative.”21 The court also looks to “the percentage of the allegedly infringing work that is made up of the copyrighted work since this offers some indication of whether the defendant's use of the ‘original materials has been sufficiently transformative.’”22 This selection of the case law leaves some room for application to the present case, but it leads naturally to the court's conclusion.
The other component of the first factor, whether a use is commercial, is not as easy a determination as it sounds. “The fact that a given use is profit-driven is not the focus of the commerciality inquiry. Instead, the ‘crux of the profit/nonprofit distinction is … whether the user stands to profit from the exploitation of the copyrighted material without paying the customary price.’”23 Under this interpretation, this element of the first factor is about whether the copier is attempting to extract some of the same commercial value that makes the original work valuable. This puts even more pressure on the question of transformativeness; unless a use is substantially transformative, it is likely to be viewed as an exploitation of the commercial value of the original work, thus leaning strongly against a finding of fair use.
On the first question, the court is emphatic that Meltwater's use is not transformative. The ruling holds:
Neither the purpose nor use of the Meltwater News Reports, nor its excerpts from the Registered Articles in the News Reports, is transformative. Meltwater uses its computer programs to automatically capture and republish designated segments of text from news articles, without adding any commentary or insight in its News Reports. Meltwater copies AP content in order to make money directly from the undiluted use of the copyrighted material; this is the central feature of its business model and not an incidental consequence of the use to which it puts the copyrighted material.24
This is a remarkably curt dismissal of Meltwater's claim to transformativeness, given the facts of the case and prior case law.
Transformativeness and Search Engines
The first fair use standard “turns primarily on whether, and to what extent, the challenged use is transformative. The use must be productive and must employ the quoted matter in a different manner or for a different purpose from the original.”25 In turn, whether a defendant meets this standard has probably become the most important component of fair use analysis in the era since Campbell v. Acuff-Rose.26 As Matthew D. Bunker describes post-Campbell jurisprudence:
Although the transformative inquiry is primarily associated with the first factor (purpose and character of the use), it also has implications for other statutory factors. For example, to the extent that a work is nontransformative, some courts, following dicta in Campbell, have reasoned that market harm under the fourth factor can be presumed or at least more easily inferred. This sort of presumption is warranted, several courts have noted, because a nontransformative use of a work can easily act as a substitute for that work.27
Conversely, transformative uses tend to be ruled as fair uses,28 not least because that transformativeness infuses courts’ readings of the other factors.
Much of the argumentation in this case is, regrettably, about whether Meltwater counts as a real search engine. The defendants in particular apparently relied on this singular claim to argue that their service is transformative,29 a strategy that is fairly described as a misstep. The Meltwater service has many similarities and some differences with what an ordinary computer user would describe as a search engine; whether these mean that it is, in the end, something that can be defined as a search engine is not legally dispositive. The question before the court is, in very large part, whether the use is transformative—whether the purpose for a Meltwater user is different than a typical visitor to a newspaper website. Here, the answer is that yes, it is a plainly different purpose. A newspaper website visitor is going there to consume the news, whereas a Meltwater subscriber is monitoring the news for whether and how specific stories get covered. News website visitors typically know less about the issue than the reporter, and they rarely have any direct knowledge about the behind-the-scenes public relations jockeying that went into it. A Meltwater subscriber generally knows more about the issue than the reporter, and for the stories that she primarily cares about, she herself or her team was probably involved in the behind-the-scenes public relations jockeying. It is roughly analogous to a college student reading an assigned book on the one hand, and a professor skimming a textbook to evaluate it for possible course adoption—one is an act of consumption, the other is an act of research and evaluation.30
The defense and the court would have been better off if they had looked less closely at the question of whether Meltwater is a search engine, and more closely at the lessons about transformativeness available in the case law. The defense's appeal to the case law on search engines is, of course, understandable. That case law is highly favorable to a defendant, Meltwater's service shares many features with even a standard search engine, and despite some potential differences, the service is still quite transformative in much the same way. This is why automated search engines are widely regarded as generally being a transformative use, a point the court concedes.31 This is a point on which Kelly v. Arriba Soft.32 and Perfect 10 v. Amazon33 are quite clear. The Kelly court held that “improving access to information on the Internet”34 justified the copying and reproduction of entire images, albeit with substantially reduced resolution. The court held that this is necessary in order to help users identify whether the image in question is the one they are seeking. Even though the thumbnail may meet the needs of some users35—removing the need for those users to click on the full-size image—the purpose of helping users find these images is a substantially different purpose than the expressive function of the images.
The Perfect 10 court is even clearer. On whether search engines provide the sort of added social value associated with transformativeness, the court writes, “a search engine provides social benefit by incorporating an original work into a new work, namely, an electronic reference tool. Indeed, a search engine may be more transformative than a parody because a search engine provides an entirely new use for the original work, while a parody typically has the same entertainment purpose as the original work.”36 At least since the 1994 Campbell decision,37 parody has been held up as a canonic example of transformative uses, making the Perfect 10 court's language even more striking.
On Whether Meltwater Is a “Real” Search Engine
The court differentiates from Kelly and Perfect 10 in a few ways. Before exploring these, though, it is worth a moment to clarify a bit about the court's writing here. Throughout, Cote implicitly defines “search engine” in a way that would require a search engine to be legitimate under copyright law. For instance, she writes, “Nor has [Meltwater] offered a comparison between the click-through rate for any single News Report and the rate for a single Google News search or any other search for news conducted through a recognized Internet search engine.”38 This implicit, but consistent, use of the term is confusing. A search engine, according to the ordinary meaning of the term, is “A program that searches for and identifies items in a database that correspond to keywords or characters specified by the user, used especially for finding particular sites on the World Wide Web.”39
Crucially, the term “search engine” does not entail any meaningful insight into a site's compliance with copyright. For instance, one reviewer writing specifically for those seeking infringing copies of works online identifies the BitTorrent search website Torrentz as “a very simple search engine,”40 and Torrentz describes itself as “a free, fast and powerful meta-search engine combining results from dozens of search engines.”41 These engines are specifically designed for and targeted at infringing users. Additionally, whether or not something is a search engine does not hinge on the type of results retrieved. Retrieving even entire works as a result of certain searches would not preclude that a service is also a search engine. Google Books is still a search engine, even when one is looking up Victorian novels and can thus read the full text. Thus, the Meltwater ruling makes much more sense if the reader imputes “legitimate” in front of the term “search engine” in most instances. This is a provisional recommendation, however; broader use of this implicit definition would be inadvisable and confusing.42
One attempt by the court to differentiate from Kelly and Perfect 10 is based on the following reasoning: “By emphasizing the small size and low resolution of the thumbnails, the Ninth Circuit relied on the fact that the thumbnails could not substitute adequately for the copyrighted works. If Meltwater captured and displayed the complete text of copyrighted news stories, it could no longer attempt to defend its business model as simply a search engine that aims to direct readers to the underlying story.”43 This is a straw opponent, to say the least. The company did not display entire articles, so the analogy is patently unfair.
Meltwater reproduced portions that are roughly analogous to the reduced-resolution results returned by image search engines: An accurate representation of what is contained therein, and enough to convey a great deal about what is contained therein, but not a total substitute for the original from which it is taken. Google's practice of downloading an entire image, then showing search results containing the reduced-resolution version, is roughly for image search what Meltwater was doing to news stories: reproducing enough of a news story to know what it is about, but not the entire news story. The goal is obviously not to substitute for the entire news article, but to help users identify and track uses of keywords across articles. To consume the entire article, the user still needed to click through to the original source. The analogy to a thumbnail image in a search result is therefore valid, and the remaining question is whether the amount taken and the substitution effects are substantial enough to negate this change in purpose.
The court also tries to differentiate based on two more specific features that are different from the image search engines in the prior cases. One is that that Meltwater searches “are run against a defined list of content providers” instead of the entire Internet.44 The other distinction is that Meltwater charges subscription fees: “Unlike the searches in Perfect 10 and Kelly, Meltwater's searches are not publicly available.”45 As one commentator observes, “Both distinctions are singularly unpersuasive: the court's distinctions would require all search engines to support themselves with advertising and would prohibit vertical search entirely.”46 The standard prohibiting a “defined list” of search targets is transparently silly, even within the terms of the court's reasoning. After all, Google News only searches news sites, and the Meltwater court describes Google News as “a recognized Internet search engine”47 and concedes that Google News is engaged in fair use.
The court's implicit standard—advertiser-supported services are fairer than subscription-based services—is inconsistent with prior case law. At least one appellate court decision deals with the question of fair use in a subscribers-only search service, and the service was found to be engaged in fair use. In the iParadigms case,48 the database at issue contains a compilation of student papers and published writings that make possible the Turnitin plagiarism detection service. The fact pattern in iParadigm would be even better than the Meltwater case for testing the theory that a lack of public accessibility is at least a partial barrier to a finding of fair use. As with Meltwater, Turnitin is only accessible to paying subscribers, but the latter is also de facto limited only to educators—and is thus even less publicly accessible. Yet the Fourth Circuit has no compunction whatsoever identifying this for-profit niche database as a highly transformative use, despite the lack of broad public accessibility. The Meltwater court creates this “publicly available” standard from thin air, directly counter to extant case law, with utterly insufficient elaboration or justification. Little wonder that a subsequent S.D.N.Y. decision had no problem finding fair use for a subscription-based service.49
The focus on subscription fees is perhaps an attempt to invoke the commercial nature of the business. If so, it borders on the silly, given that it is placed alongside a positive evaluation of Google's similar service. This juxtaposition leads another critic of this decision to point out the obvious: “Google is not some nonprofit entity that exists to provide public benefits. It is one of the most successful companies in the world.… [C]ourts should not conclude that because Google does not charge a fee in order to use its search engine, it somehow is not exploiting copyrighted content in the same way as Meltwater is when charging a fee.”50
The iParadigms case also contradicts some of the Meltwater court's other reasoning on transformativeness. The plaintiffs in iParadigms, four students who authored school papers that were added to the database, made another argument that would later be echoed in the Meltwater court's reasoning, only to be rebuffed soundly by the trial and circuit courts. The Fourth Circuit ruling elaborates:
Plaintiffs also argue that iParadigms’ use of their works cannot be transformative because the archiving process does not add anything to the work—Turnitin merely stores the work unaltered and in its entirety. This argument is clearly misguided. The use of a copyrighted work need not alter or augment the work to be transformative in nature. Rather, it can be transformative in function or purpose without altering or actually adding to the original work. iParadigms’ use of plaintiffs’ works had an entirely different function and purpose than the original works; the fact that there was no substantive alteration to the works does not preclude the use from being transformative in nature.51
This is part of a broad trend in fair use jurisprudence; “exact copying plus transformative purpose has a stunningly good fair use record in recent cases. Copies, both large-scale and selective, have been blessed as transformative by courts reasoning that exact copies can still have transformative purposes distinct from the purpose of authorized copies.”52 It is also rather directly in opposition to the Meltwater ruling's implicit point, in the passage quoted earlier, that the service is not transformative by virtue of the service not “adding any commentary or insight in its News Reports … [and making] undiluted use of the copyrighted material.”53 One need not add new content or commentary for a use to be transformative, and higher courts in two other circuits54 had held exactly this, in decisions issued before Meltwater was decided.
The defense likely could have gotten much further by branching out their argument about transformativeness. The argument that Meltwater operates a “search engine” was neither necessary nor (apparently) sufficient to win the day. What matters is whether the service is making transformative use of the news articles, and in this case, an especially relevant point is the research purpose served by the service. As another S.D.N.Y. decision, Fox News v. TVEyes, noted, “Meltwater failed to show that its service was actually used by subscribers for research or to transform the original news story into a factum or datum that told a broader story about the overall news reporting industry.”55 Going forward, similarly positioned defendants should argue that they are providing a tool for research, that subscribers/users are actually using the tool for research, and that such research is a highly transformative purpose. In “future scenarios, what if a news aggregator can show that its commercial consumers only use the snippets for monitoring how frequently it is mentioned in the media and by whom? Is that not a different ‘use’?”56 It is quite different, and that difference does not depend on whether or not the aggregator is a “recognized internet search engine”57 or is “simply a search engine”58 or falls into the category “of search engines, as that term is commonly understood.”59
Too Late for the Defense: Authors Guild v. HathiTrust
Unfortunately for Meltwater (the company), the Authors Guild v. HathiTrust decision60 came out just eleven months after their suit with AP was settled. It is likely the most directly relevant decision the Second Circuit has ever issued—which would have more tightly bound the District Court's hands in recognizing that search is indeed transformative.61 The defendant, HathiTrust, is the nonprofit arm that operates the HathiTrust Digital Library (HDL), which serves as the online search portal for all of the data from the books scanned by Google as part of their Google Books project. The author and publisher plaintiffs sued, alleging copyright infringement, and the court found that the defendants were engaged in fair use. Many of the findings in HathiTrust would have been important here and helpful to Meltwater, but the point most different from the Meltwater court's reasoning may be the court's definitive finding that search is indeed a transformative use.
The court writes, “Turning to the first factor, we conclude that the creation of a full-text searchable database is a quintessentially transformative use.… [T]he result of a word search is different in purpose, character, expression, meaning, and message from the page (and the book) from which it is drawn.”62 Further, “by enabling full-text search, the HDL adds to the original something new with a different purpose and a different character.”63 Even more strikingly, the court rejects the moralistic reasoning that often pollutes first factor analyses, writing, “Contrary to what the district court implied, a use does not become transformative by making an ‘invaluable contribution to the progress of science and cultivation of the arts.’ Added value or utility is not the test: a transformative work is one that serves a new and different function from the original work and is not a substitute for it.”64 If courts were to use this relatively more objective standard, judges’ beliefs about which parties are adding more value to society would less often pollute their rulings; sadly, Meltwater is an especially fine example of such pollution.
The substantial impact of the HathiTrust case, and its implication on later district court cases in the Second Circuit, is not merely hypothetical. Later in 2014, the HathiTrust decision had a substantial impact in a different S.D.N.Y. case, Fox News v. TVEyes.65 In that case, Fox News sued TV news monitoring and research service TVEyes for alleged copyright infringement. In reaching a conclusion that TVEyes was engaged in fair use, Judge Alvin K. Hellerstein draws heavily on HathiTrust, citing the case repeatedly. For instance, he writes, “I find that TVEyes’ search engine together with its display of result clips is transformative, and ‘serves a new and different function from the original work and is not a substitute for it.’ In making this finding, I am guided by the Second Circuit's determination that databases that convert copyrighted works into a research tool to further learning are transformative.”66 This does not mean that Meltwater necessarily would have prevailed. On some of the features of the TVEyes service, Hellerstein found the use to be unfair.67 Still, the HathiTrust standard on transformativeness could have greatly clarified the reasoning in Meltwater, and the Second Circuit case certainly makes for better law. It also would have made a partial or full victory for Meltwater far more likely.
Searches, E-mail Blasts, and Clicks
The ruling sets out to demonstrate that Meltwater's service is not a “real” search tool, and therefore does not qualify as a search engine—and therefore is not transformative. In particular, the court argues, “Meltwater News is neither designed nor operated to improve access to the complete, linked news story.”68 This conclusion is based almost entirely on Meltwater's seemingly low click-through rates, or the frequency with which a subscriber clicks on a given news link. Based on the thirty-three articles at issue, Meltwater submitted evidence that, relative to the number of times the excerpts were sent to subscribers, they were opened just 0.08 percent of the time.69 Out of context, this seems like a shockingly low number and solid support for the conclusion that this is not an information location tool.70
It is only fair to compare the Meltwater click-through rates in relation to other comparable rates, and while it may come as a shock to those who do not track strategic e-mail figures, even one click per thousand links sent is not only reasonable but normal. The most direct analog is automated Google News alerts, sent to subscribers by e-mail based on search terms.71 The court does not report on these click-through rates, relying instead on an amicus brief that reports 56 percent of visitors to the Google News webpage click on at least one link.72 This is an embarrassingly bad comparison—while the defense counsel surely should have torn it to pieces, it is also silly that the court included it without caveat or critique.
As any experienced user of both the web and e-mail subscriptions can attest, a visitor who proactively visits a website is orders of magnitude more likely to click on an available link than somebody sorting through her inbox. Nearly every serious professional is a (voluntary) subscriber to very many e-mail lists that contain professionally relevant news, and the vast majority of the links contained therein are surely going unclicked. While I could not locate any published research on click-through rates for noncommercial e-mails, these rates are surely very low—in large part because of the over-filled nature of knowledge workers’ inboxes. One study found that an average knowledge worker receives 52 e-mails per day and writes 31 per day, taking up 19 percent of the average workers’ computing time.73 Another study found a mean of 57 e-mails received per day, as well as an eye-popping 1,551 e-mails in the average user's inbox at the day's end.74 Of course, many of these e-mails are genuinely mission-critical, and these workers also need to do many other tasks besides pruning their inboxes—many of these also mission critical. Therefore, “I should monitor that; I'll subscribe to the e-mails” is very different from “I should read that automatically generated e-mail alert now, instead of any of the dozens of other work tasks I could be doing.”
The court's comparison is made even less fair by equivocating a click on any link on the Google News webpage with a click on any specific link in a Meltwater e-mail. At any one time the page has dozens and dozens of links to news stories, so each one of these links is going to do orders of magnitude worse than 56 percent. As of this writing, I count 96 news stories linked directly from the Google News page; even if a typical user clicks on three of these (a generously high estimate), that is a per-story click-through rate of 0.56 * 3/96, or 1.75 percent—and again, that is among those who are at that moment proactively seeking news on the site, making each web link many times more successful than if it were sent via e-mail blast. Thus, the difference between this estimated per-link rate for the Google News website and Meltwater's e-mails should be utterly unsurprising and certainly no grounds for dismissing Meltwater's claim to being an information retrieval tool.
It is regrettable—especially for the sake of Meltwater and its counsel—that the company did not produce its own estimates of comparable click-through rates. In particular, well-known public figures in e-mail and web ad marketing provide important context.75 Constant Contact, an industry leader in outsourced e-mail marketing for small businesses and nonprofits, provides hard data on their click-through rates by sector.76 By publishing these figures, the company implies that they stand up nicely against their competitors’ figures. It breaks these rates out across 38 sectors such as real estate, publishing, travel and tourism, and several categories of nonprofits. Sector averages for how often recipients even open an e-mail range from 9.9 percent to 40.2 percent. Click-through rates are measured per opened e-mail—not per e-mail sent—and the sector averages here range from 4.4 percent to 20.6 percent of the time. Multiplying these figures together—share of e-mails opened; and of those opened, share that garner clicks—yields click-per-e-mail rates, by sector, from 0.7 percent to 4.7 percent.
These are sector averages; several customers in a given sector are paying handsomely to send e-mails that get clicks even less often. Even this figure is still not a fair comparison with the link-specific rates used by the court, though, as the Constant Contact figures count a click on any link in an e-mail. If an average e-mail has 15 links and, among users who click on any links, they click an average of 1.3 links (both reasonable ballpark estimates), the click rate per link, relative to how many times that link is e-mailed, would be below 0.45 percent in all sectors and as low as 0.06 percent in some sectors—with many Constant Contact customers experiencing even lower rates.
One might object that marketing e-mails will naturally have very low response rates, and we should expect better for work-relevant e-mails that one subscribes to. Keep in mind, though, that Constant Contact strictly requires that all e-mail recipients consent to receive e-mails from the sender.77 They have zero tolerance for sending e-mails to people who don't affirmatively consent to contact from senders.78 If senders have high bounce (bad address) or mark-as-spam rates, Constant Contact gets involved, and fast. After all, for Constant Contact, being a trusted source of messages, as moderated by the providers of e-mail services such as Gmail and Hotmail, is at the core of their business model. Thus, Constant Contact rates are particularly high and particularly reflective of how often users tend to click on links in solicited but unimportant e-mails. That Meltwater's rates are in this range should be unsurprising, not shocking.
Another comparison, to web ad click-through rates, is also illustrative. Retailer clients of the ad firm Nanigans were thrilled in the third quarter of 2013, when Facebook ad click-through rates soared to 0.20 percent; the previous year, these had been just 0.05 percent.79 Across the entire industry, the average Internet advertisement has a click-through rate of 0.06 percent.80 While click-through rates are not dispositive, these figures from the world of Internet marketing are stunningly close to the Meltwater rate—0.08 percent. In other words, corporations, nonprofits, and governments willingly pay sizable sums for these levels of online engagement. These response rates are utterly consistent with a service having a meaningful role in delivering traffic to websites, a substantial rebuttal to the court's dismissal of Meltwater's claim on this count. Again, in this particular case the blame surely falls largely at the feet of Meltwater's attorneys, but the conclusion is erroneous and worth correcting.
It also says something that Meltwater delivered the kind of traffic to AP periodicals, for free, that other websites gladly pay for—whether via e-mail or web ads—and the AP sued them for the effort. This is illustrative rather than conclusive; delivering the kind of traffic that some sites would pay for does not make Meltwater's service transformative, and the failure to deliver more traffic does not make it non-transformative. Wildly popular information retrieval services can be infringing, and non-engaging ones can be perfectly above-board. Still, these other areas of the Internet economy show that even these seemingly low rates of engagement are actually consequential rather than scandalously low.
More fundamentally, it bears repeating that the participants in this case all spent too much energy on the debate over whether the service is really a search service. Even if the e-mails are generally opened and rarely lead to clicks, they offer a short summary of which topics are being covered across a range of periodicals, delivered to the kind of communication professionals who help make these stories happen. The articles are serving profoundly different functions in the two different contexts—one for consumers of news, and one for newsmakers engaged in research and monitoring. This substantial difference is why another S.D.N.Y. judge in the Fox News v. TVEyes case provided an implicit but important corrective on whether the Meltwater service is transformative. Judge Alvin K. Hellerstein writes, “In the cases cited by Fox News, save for Meltwater, defendants were copying the plaintiffs work and then selling it for the very same purpose as plaintiff.”81 By implication, Judge Cote's S.D.N.Y. colleague is pointing out that Meltwater's service served a purpose that is at least somewhat different from that of the original news articles. Little wonder, then, that Hellerstein finds a very similarly positioned defendant to have been engaged in fair use.
Also, while transformative use is critically important in this case, it is also not dispositive. As Dylan J. Quinn writes, “This is not to say that Meltwater, in this specific case, actually met its burden of presenting evidence to demonstrate the use was fair in regards to the other factors, but if the court had determined that the use was transformative, the outcome may have been different when analyzing all of the factors together under § 107.”82 If the court had made the more defensible conclusion that selling research tools is at least a transformative use, pending the other three factors, it likely would have changed the light in which those other factors were examined, and this may have changed the outcome.
Second Factor: Nature of the Copied Work
The second fair use factor, the nature of the copyrighted work that is being used, also breaks down into two components. One, whether the original has yet been published,83 is essentially a push in this case. The ruling elaborates on the other component, which is whether a copyrighted work “‘is expressive or creative, such as a work of fiction, or more factual.’ Works of fiction are ‘closer to the core of intended copyright protection’ than are works that are predominantly factual. Accordingly, the scope of fair use is broader with respect to factual works than it is with respect to works of fiction.”84 Despite acknowledging this framework, the ruling quickly neuters its impact: “AP's articles are news stories and therefore more vulnerable to application of the fair use defense than works of fiction. Moreover, Meltwater copied works that were already published. As a consequence, this factor ‘is at most neutral on the question of fair use,’ and should be weighed in favor of finding fair use.”85 Thus ends the court's discussion of the second factor, and the decision does not even mention it in bringing all of the factors together to reach a conclusion.86
Failing to engage with the highly factual nature of the copyrighted works is quite an omission. If the factual/creative component of the second factor means anything, surely a case dealing with highly factual copyrighted works merits more than a trivial aside. Rather, it should be an important part of the final analysis, creating much greater leeway for the amount used and potential market harm. The Supreme Court makes it exceptionally clear that it is easier to make fair use of newspaper articles than of artistic photographs, the latter being the works that were found to have been subject to fair use by search engines in both Kelly and Perfect 10. Judge Cote, however, has a very different view, and she has voiced that view before. The last quote—that a work being highly factual “is at most neutral on the question of fair use”—is from a 1999 Second Circuit ruling, Nihon v. Comline,87 which is a partial affirmation of Cote's own District Court ruling.88 In her own ruling in Nihon, Cote seems to skip the second factor entirely,89 ignoring both its equal placement in the statute and higher courts consistently dealing more extensively with this issue when it is so clearly applicable. Instead of a corrective, the higher court amplifies this error with their ruling that, when the copyrighted works in question are highly factual, this factor “is at most neutral.”90
Several major Supreme Court cases that came before the Nihon case state that a fair use defense is substantially more likely to succeed if the copied work is more a factual work than a creative one.91 In Stewart v. Abend, the court writes, “fair use is more likely to be found in factual works than in fictional works.”92 The Sony v. Universal decision includes the statement, “copying a news broadcast may have a stronger claim to fair use than copying a motion picture.”93 Campbell v. Acuff-Rose includes the statement, “some works are closer to the core of intended copyright protection than others, with the consequence that fair use is more difficult to establish when the former works are copied,”94 and thus, by logical implication, that fair use is easier to establish when works with thinner protection are copied.
Perhaps no Supreme Court case is clearer on this point than Harper & Row. The court's explanation in this case, in which the original work is a political biography, is especially detailed and illustrative. The case centers on the defendant's use of the work's most expressive elements, including “subjective descriptions and portraits of public figures whose power lies in the author's individualized expression.”95 Yet even in this case, the court especially stresses that fair use includes a fairly broad right to use the factual parts of a work. Even copying the expression of those facts can be and often is fair use:
The law generally recognizes a greater need to disseminate factual works than works of fiction or fantasy. “[Even] within the field of fact works, there are gradations as to the relative proportion of fact and fancy. One may move from sparsely embellished maps and directories to elegantly written biogra-phy. The extent to which one must permit expressive language to be copied, in order to assure dissemination of the underlying facts, will thus vary from case to case.”96
The treatment of this question in Meltwater is thus at odds with the widely shared understanding of this factor, skating past the clear, relevant dicta in several Supreme Court cases to reach exactly the opposite conclusion from that recommended by the high court.97 It would be unremarkable if the Meltwater court had merely given this factor relatively less weight than the other three factors; many others dealing with similar fact patterns have also perhaps given inadequate gravity to the factual nature of copied works. Yet the ruling's dismissal of any need to give the factor any weight in this case whatsoever—in particular, through the “at most neutral” formulation—is, to my knowledge, only present in Nihon and Meltwater.98 In this way, these cases stand out for their problematic, especially suspect treatment of the second factor.
Sweaty Brows and Copyrightable Creativity
Throughout, the decision repeatedly reveals a desire to ensure that AP will be rewarded for their effort and expense. For instance, “Investigating and writing about newsworthy events occurring around the globe is an expensive undertaking.”99 The ruling decries Meltwater's taking of “the news reporting and research that AP labored to create” and Meltwater's attempt “to take the fruit of AP's labor for its own profit.”100 This emphasis on rewarding hard work led the lower courts astray in Feist v. Rural. With this focus on the effort required to collect facts—the “sweat of the brow”101—the district and circuit courts ruled that an utterly uncreative factual work could be protected by copyright.102 The Supreme Court overruled this interpretation, reminding us that facts are not copyrightable.103 “Factual compilations … may possess the requisite originality [to be copyrightable, but this] protection is subject to an important limitation. The mere fact that a work is copyrighted does not mean that every element of the work may be protected. Originality remains the sine qua non of copyright; accordingly, … protection may extend only to those components of a work that are original to the author.”104
I hold the highest regard for the work and craft that goes into writing a good newspaper article, but few copyrightable works are more clearly a compilation of facts and thus subject to relatively thin copyright protection. A good news writer fits as many important facts as possible into as little space as possible, including especially the headline and lead paragraph, and conveys these facts with as little embellishment as possible.105 Further, there are only so many ways to convey these facts; straight news articles are perhaps the single best example of the Harper & Row court's implicit point that, in some cases, “one must permit expressive language to be copied, in order to assure dissemination of the underlying facts.”106
The Meltwater decision's Appendix contains three illustrative articles and excerpts, and the very first example is a splendid illustration of the mostly factual nature of a typical article. With the headline, “‘Help wanted at new casino for Toledo, Ohio,’” the excerpted portion of the lead paragraph is, “‘Job seekers can roll the dice to land work at another of the four casinos coming soon to Ohio. Hollywood Casino Toledo has posted more than 600 job listings on its website this week.’”107 The subsequent excerpt continues, “‘… restaurant workers, slots and table games supervisors, groundskeepers and security officers. The casino is scheduled to open in the spring.’”108
The only excerpted portion that is not purely factual is a nominally creative metaphor, and the rest is almost pure conveyance of facts. A non-AP article has the headline, “Toledo casino posts 600 job openings,” and its opening paragraph is remarkably similar.109 This similarity is surely not because the publisher copied the AP story. If there is any copying afoot, it is surely that both articles copied heavily from the casino's press release. Several other regional news sources apparently did the same.110 One opens with, “Roll the dice and pick a card—Hollywood Casino is looking to hire more than 600 employees this spring.”111 Another starts, “If you were betting on landing a job at Toledo's Hollywood Casino, it's time to roll the dice.”112 This is surely an example of a story that was widely copied, from a press release, by a wide range of news outlets. This practice, derisively called “churnalism,”113 is hardly a rational basis for a newspaper being afforded anything like the same thickness of protection afforded a novel or even a memoir.
The court offers no accounting for the problem that many AP articles themselves are based in large part on the works created by other authors. The present study does not conduct an article-by-article analysis of all 33 articles that were the cause of action, but I find it remarkable that, even in the handful that are included for illustration in the court's case, several are transparent churnalism. The Ohio casino story is representative, not exceptional, on this count.
These are the articles that AP chose as causes of action, and in any other court, the defense counsel likely would have jumped all over this; in front of Cote, however, they may have chosen not to pursue an issue that she had already swatted away in an earlier ruling. In Nihon, the defense raises the issue of copying from press releases, but Judge Cote dismisses it as not disqualifying the relevant articles from copyright protection.114 I concur, but an article may be copyrightable and yet be subject to very thin copyright protection.
Thus, there are actually two quite substantial points that are relevant to the question of these works’ expressivity: their largely factual nature, and the degree to which they are transparent copies of somebody else's work. A mostly factual, mostly copied work can still be protected by copyright, but the protection afforded to such a work becomes so thin that fair use is exceptionally easy to demonstrate.
Third Factor: Amount and Substantiality of the Portion Used
The third fair use factor considers “the amount and substantiality of the portion used in relation to the copyrighted work as a whole.”115 The former is, more or less, a question of what percentage of the original is copied; the higher a proportion is copied, the less likely a finding of fair use. The latter asks, of the portion that is taken, how important is that portion to the original work? If it is the “heart” of the work—if “‘the most interesting and moving parts’”116 of the original work are copied—then this also leans substantially against a finding of fair use, and it does so even if the amount taken is a small portion of the original.
The Court's Ruling on Amount and Substantiality
The amount used in each case is highly consistent in terms of the number of characters taken: up “to 300 characters … from the opening text of the article or lede … [plus] approximately 140 characters … ‘surrounding a single, algorithmically chosen appearance of one of the customer's matched search keywords.’”117 Thus, the proportion varies depending on the length of the original article; “it probably ranged from as low as 4.5% to slightly over 60%.”118 In the latter case, the article is the one discussed earlier, “Modern Pentathlon Tightens Anti-Doping Policy.”119 In terms of the importance of what is taken, the court writes, Meltwater “automatically took the lede from every AP story. As described by AP's Standards Editor, the lede is ‘meant to convey the heart of the story.’ A lede is a sentence that takes significant journalistic skill to craft. There is no other single sentence from an AP story that is as consistently important from article to article.”120
The court decries Meltwater for not attempting “to show that it took no more than necessary to perform as a search engine, which is how it seeks to justify its infringement.”121 In contrast, “Google News Alerts do not systematically include an article's lede and are—on average—half the length of Meltwater's excerpts.”122 The court elaborates:
[S]earch engines regularly display briefer segments of news articles. As the CCIA describes the segments taken by search engines, they are no more than “a headline and a snippet of context” designed to direct users to the original source. Indeed, in its foreign operations, it is undisputed that Meltwater provides its customers with far less material than it provides in the United States. In Canada it delivers only headlines, and in the United Kingdom its excerpts are far shorter.123
Because the service delivers far more substantial excerpts to US audiences, then, the court seems to suggest the amount taken is presumptively more than is necessary, tilting this factor against a finding of fair use.
Lots of Facts Are Still Facts
On this point, the court again fails to engage, in any meaningful way, with the highly factual nature of the works in question. The ruling's summary of the case law is adequate to convey the gist of the third factor, but the court would have done better to include highly relevant guidance from an earlier Second Circuit ruling, Salinger v. Random House. Specifically, Salinger held, “what is relevant is the amount and substantiality of the copyrighted expression that has been used, not the factual content of the material in the copyrighted works.”124 The amount that can be fairly taken from a highly factual work, such as a news report, will often thus be substantial. The “heart of the story” conveyed by the lede is a tight package of facts. Yes, it takes substantial skill and craft to write a good lead sentence; it also takes substantial skill and craft to wire a house safely, but electrical wiring is functional rather than expressive, so a wiring diagram merits very thin copyright protection. The craft of journalism, as embodied by this kind of opening sentence, is to fit as little expression and as much fact as possible into a small space. That creates room for a lot of copying within the space of fair use, pending the other factors.
The extent to which articles are copied from press releases also matters on this count. A work that is itself mostly a copy can be copied more freely than a work that is genuinely original. To take another example, the article “Modern Pentathlon Tightens Anti-Doping Policy” is clearly based on the international sports body's own press release.125 Because of this, there remains little expression to protect in the lead sentence, “‘Modern pentathlon has joined other sports in adopting a ‘no needles’ policy as part of its anti-doping rules ahead of the 2012 London Olympics,’” or the other excerpt, “‘… says athletes can receive injections only from a ‘certified medical professional’ after an appropriate diagnosis and only if there is no alternative.’”126 The press release includes the headline, “UIPM Approves No Needle Policy,” and the opening sentence, “At its 63rd Congress, UIPM approved the proposed ‘No Needle Policy’ which comes into force immediately to assist keeping Modern Pentathlon a doping free sport.”
The article is nothing but a summary of facts, presented to—and then rearranged by—an AP writer. So little expression is even there to be copied that a finding of fair use for this article by Meltwater is quite likely warranted.
Some of AP's articles show substantial creative expression and originality. For instance, the service also circulates opinion pieces,127 though these seem not to have been among the works in the complaint. Ironically, these far more creative, expressive works help illustrate the problem in emphasizing the importance of the headline and lead paragraph from a copyright standpoint. An op-ed or an editorial is surely more expressive, and thus subject to thicker copyright protection, than a straight news piece—and in the case of opinion pieces, the headline and lead paragraph are especially poor substitutes for the entire article. The more creative and expressive a journalistic work, the less of the proportional value of the entire work is embodied in the headline and lede. Conversely, the more the headline and lede are the heart of the value of the work, the more likely the value of the whole article is in conveying facts rather than expression.
The only example article in the ruling that is substantially expressive is a fantastic illustration of how poorly the headline and lede capture the heart of such a work. The article, entitled “Sun Shines on the Mountain and Pearce Rides Again,”128 is a touching 800-word tale of a world-class snowboarder who suffered a nearly fatal head injury, spent nearly two years in recovery, and finally returned to the slopes, where he could at least begin enjoying casual rides again. In this case, the headline and lede are not at all a replacement for the full article; they don't even inform the reader how he was injured, the seriousness of his injuries, or the duration and difficulty of his rehabilitation. The only way to convey all of these facts in a headline and lede would be to include a tight summary of the facts, with as little expression as possible. Thus, the court unintentionally includes a great contrast with the “just the facts” style of most news articles, giving an example of what expressive newswriting actually looks like—and, unwittingly, how poorly an e-mail excerpt from Meltwater substitutes for such a work.
I concede that Meltwater might have chosen to use a smaller take of each individual work. Because of this point in particular, it would definitely be possible to construct an argument that Meltwater infringed at least some of the articles that were causes of action—and write such an opinion in a way that is much more consistent with other fair use case law and more useful to future courts than the court's reasoning in Meltwater. There is surely at least some daylight between the service provided by Google News and that provided by Meltwater at the time of the suit, and a court arguing that this is the lynchpin of the case might have reached a more sound conclusion that found partially or entirely in favor of AP.
I would argue, though, that even the amount taken should not lead to a finding of infringement of any of the articles. This is because of the way the first two factors should have been applied in this case. In addition to being able to take more when a use is transformative, one can also take a great deal more from highly factual works than from highly expressive works. In Salinger, the high court clarifies that third factor calculations are to include only the amount of expression copied, not the facts that are embodied in such expression.129 Fortunately for newspaper writers and readers alike, their works are a degree more expressive than purely factual works like a statistical table or an alphabetized phone directory. Yet most straight news reports are closer to the “pure facts” end of the expressivity scale than the highly expressive end where we find works like novels and poems. Accordingly, copying the elements most likely to contain the highest concentration of facts—the headline and lede—should generally be fair use as long as the purpose is reasonably transformative and market effect is reasonably modest.
Fourth Factor: Effect on Market Value
The fourth fair use factor is “the effect of the use upon the potential market for or value of the copyrighted work.”130 This also has two components. First, a court must consider whether sales of the work itself are harmed. Since there is essentially no market for yesterday's news as news, this component is not especially important. The second component, the effect on the market for derivative works, is far trickier. “By definition every fair use involves some loss of royalty revenue because the secondary user has not paid royalties. Therefore, if an insubstantial loss of revenue turned the fourth factor in favor of the copyright holder, this factor would never weigh in favor of the secondary user. And if we then gave serious deference to the proposition that it is ‘undoubtedly the single most important element of fair use,’ fair use would become defunct.”131
The Meltwater court, drawing on an earlier Second Circuit ruling,132 starts this analysis by asking whether and to what extent a use competes with authorized derivative works, as expressed in an actual marketplace. “Thus, in order to prevent the loss of licensing fees from becoming a syllogistic consideration, courts consider only the loss to potential licensing revenues from ‘traditional, reasonable, or likely to be developed markets.’”133 Further, “when the use is transformative or takes place in a market that the copyright holder is unlikely to develop, it is more likely that the defendant has engaged in a fair use of the material.”134
Transformativeness of use can even preempt a copyright holder's attempt to extract licensing revenues from otherwise fair users. Judge Cote explains:
[W]hile a copyright holder's current participation in a given market is relevant to the determination of whether the market is “traditional, reasonable, or likely to be developed,” it is not determinative. “[A] copyright holder cannot prevent others from entering fair use markets merely by developing or licensing a market for parody, news reporting, educational or other transformative uses of its own creative work.” In other words, “[c]opyright owners may not preempt exploitation of transformative markets.”135
Thus, while a market for licensing said works is informative, it is not dispositive if the use in question is substantially transformative. As a subsequent court found, “any economic harm caused by transformative uses does not factor into this analysis.”136
This writing by Judge Cote deserves sincere praise. It is an especially clear explication of the tricky point that a functioning market does not automatically mean that licensing fees are due. From within this framework, then, her finding that the market effect also favors AP is primarily a consequence of her findings on the other factors. Phrased more critically, the gravest errors in the Meltwater ruling are based in a misreading of the first three factors, discussed earlier. If Meltwater's service is highly transformative, primarily a use of mostly factual works and mostly of the facts therein, and using portions that are at most a bit outsized, then AP should have no more right to demand a license from Meltwater than an earlier scholar has to demand fees from a later researcher for the right to quote an article. It is only because the court has already found stridently in the opposite direction on these previous points that the lack of licensing fees is fairly characterized as an injustice.
Adding Up the Four Factors
Adding up the four factors, this case looks more like fair use than infringement. The court essentially grants that standard search engines, and even targeted search engines like Google News, are obviously transformative, and none of the reasons for distinguishing the facts in Meltwater from cases like Kelly are persuasive or grounded in law. An information company that charges subscription fees should be treated the same as a firm that gives away access and sells ads. Moreover, Meltwater was directing users to specific bits of AP-licensed content with roughly the same success rate as paid online marketing, making it look an awful lot like an information retrieval tool. As such a tool, it places all of this information in a new context: alongside other articles on the same topic of the user's choosing, from multiple publishers. It is thus quite transformative. Most of the works that are quoted are only nominally expressive and thus should be subject to greater leeway under fair use. The window for fair use is even broader for those articles that are copied from press releases, made even more ironic by the fact that in many cases Meltwater clients are actually more responsible for the creation of the works in question. In a case featuring such starkly factual works, repurposed from other writers’ creations, this factor must be taken seriously, reshaping how we weigh the other three factors. The amount taken may be more than is strictly necessary, weighing slightly against fair use, but this weight is slight indeed compared to the first two factors. Finally, as per market effect, the AP's right to demand licensing fees is seriously diminished by the other three factors, and even the functioning of such a market is a sign of their market power rather than a fair valuation of such uses of any given individual work.
A correct reading of these cases and the facts would likely lead to a finding that Meltwater was generally or entirely engaged in fair use of these articles. A finding of no acts of infringement would have been fair and reasonable, as would a finding of infringement in some portion of the alleged instances.137 The Meltwater court thus likely reached an erroneous conclusion. The outcome, however, is not even the worst part about this holding; the means by which it is reached are far more insidious. There is enough room for varying interpretations of precedent that one could certainly construct an aggressively pro-copyright holder reading of these precedents, by more legally and logically valid means, that would side with AP without this decision's dizzying combination of unprecedented new standards (unacknowledged as such) and reintroduction of long-since-discarded theories. Instead, the court's errors are likely to reverberate in ways that are a net negative for the future of the information and media industries.
Reverberations in Search and News
The Meltwater adds unfortunate uncertainty to what had been a safer space for digital innovation. Further, it does so for little benefit to the news industry.
Impacts on Search and Innovation
Search engines need an exception to copyright to function; negotiating with all of the relevant content producers would be economically unfeasible.138 Despite the Meltwater decision, it is widely acknowledged that Internet search services are very likely to be engaged in fair use, and that digital search tools are the kind of highly transformative uses that are likely to be fair.139 Yet Meltwater injects just enough uncertainty into this understanding that the builders of future innovative search services—and, even more so, would-be investors—may hesitate before investing substantial time and money into something that may only lead to lawsuits down the road. In particular, the decision “may undermine the ability of other aggregators to distribute and disseminate information to the public by taking too narrow a view of what constitutes ‘transformative’ use.”140 Further, it provides at least a bit of straight-faced credibility to a would-be plaintiff—one who might be seeking to collect on the next innovator in search. This makes a substantial award of attorney's fees a great deal less likely, substantially reducing the expected cost of litigation.141
The area of parody provides an illustrative best-case scenario for search technologies. The 1994 Campbell decision142 was viewed even at the time as creating a much greater berth for fair use,143 and the decision has indeed created that legacy.144 A decade later, Mattel was ordered to pay nearly two million dollars in attorney's fees and costs to a defendant-artist who was obviously engaged in parodic fair use.145 This decade, another court ruled that a parody was fair use based solely on the pleadings,146 inviting courts to make fair use defenses even simpler and cheaper for defendants.147 While people engaged in obvious fair uses could still stand greater certainty,148 parodists have probably never been freer to mock—at least, according to the black letter law.
Search probably was and potentially still is on a similarly pro-fair-use trajectory. A Mattel-like award of substantial attorney's fees in particular would discourage baseless litigation against obviously transformative users. Thanks to Meltwater, however, this day may be delayed. It provides would-be plaintiffs with a far more colorable claim of infringement, even against services that are making exceptionally transformative and overall fair uses of a given set of works. This is especially problematic if a plaintiff's suit would only stand under something like the Meltwater court's highly flawed reasoning. With this decision still on the books, an award of attorney's fees is much less likely, and that makes such litigation more likely. Major players in the Internet ecosystem are extremely sensitive to even the threat of copyright litigation.149 Thus, even the odd out-of-line decision like Meltwater has a real potential to discourage development and innovation.
Little Help to Journalism
A shift in copyright that provides slightly more protection to news from online search is unlikely to have any meaningful impact on the production and distribution of quality news reporting. First and foremost, licensing to such services is a very small part of even the AP's budget. Ken Doctor of Nieman Journalism Lab observes that, in 2012 at least, a measly eight percent “of newspaper revenues came from something other than advertising or circulation.”150 Doctor's post breaks out the different elements of that sliver that have at least some potential for real growth, but search and search licensing are apparently not even worth mentioning.
Further, if Meltwater has an impact on the industry, it will further incentivize over-reliance on public relations sources, a problem that has already negatively affected journalism for years151—nay, decades.152 As a rational actor, the AP already has a strong cost-cutting incentive to hire fewer reporters who can produce more content per person by rewriting press releases. Some of the results discussed earlier illustrate the extent of this problem and the extent to which the AP sees nothing wrong with this type of journalism.
The AP and its attorneys presumably could have filed suit over any of thousands of articles that appeared in Meltwater searches, but they still chose several that are obvious examples of churnalism. Protecting these less creative articles as though they are in the same category as long-form essays and investigative journalism—let alone the same category as Toni Morrison—only further entrenches this production model. The Meltwater court fails to separate original, expressive writings from journalistic recycling of bare facts. Perhaps we should be saying thank heavens that this is almost certainly not the model for newspapers to grow their revenue. Even the lowest-budget news website—to say nothing of hobbyist bloggers and social media users—can repeat the news that the local casino is hiring. If that is what the AP is hanging its financial hat on, our civic discourse is in deep trouble regardless of the level of copyright protection for news articles.
It is clear that the old business model for print journalism is in trouble and could benefit from major policy changes. Greater minds than this author have proposed several.153 Enhanced copyright protection, however, would provide little if any real gain for the quality and quantity of original news reporting.
Future courts should ignore the Meltwater ruling. The misunderstanding of transformativeness; the misapplication of reasoning to the nature and norms of the technology at hand; the refusal to treat a work's being primarily factual as a substantial factor; and the legion other errors are not a great model of fair use jurisprudence. Fortunately for the AP, Meltwater quickly reached a licensing agreement.154 This outcome may prove unfortunate, though, for the next creator of an innovative search technology. The further a new search service deviates from the existing search engine market—in other words, the more innovative that service is—the more skeptically the next federal magistrate may view it. Before it even gets to court, engineers and investors may also look elsewhere rather than risk expensive litigation.
Without claiming any special insight into Judge Cote's inner states, it is obvious that she is deeply concerned about the future profitability of the newspaper industry. I share these concerns, as I assume is true for the vast majority of educated, thoughtful citizens. Yet the newspaper industry should not repeat the mistake that the record industry made of over-reliance on copyright as a means to limit digital technologies.155 Using copyright to endanger the future of news search/research services is about as sensible a solution to the woes of the newspaper industry as banning coffee would be for the problem of insomnia. We should find another way to “save” newspapers and put the reasoning in the Meltwater decision to bed.
- 1.Cisco, “Cisco Global Cloud Index.”
- 2.Cisco, “Visual Networking Index IP Traffic Chart.”
- 3.Cisco, “Cisco Global Cloud Index.”
- 4.“Global Top 100 Companies by Market Capitalisation.”
- 5.“Top Sites.” Nine other Google search pages—local versions for other countries, such as India, Japan, and Germany—are also among the top 50 websites.
- 6.Patry, 68.
- 7.For example, Kelly v. Arriba Soft. Corp., 336 F.3d 811 (9th Cir. 2003); Agence France-Presse v. Google Inc., No. 1:05 Civ. 00546 (D.D.C. Oct. 26, 2005) 2002WL 3174409; Perfect 10, Inc. v. Amazon.com, Inc., 508 F.3d 1146 (9th Cir. 2007).
- 8.For example, AP v. All Headline News Corp., 608 F. Supp. 2d 454 (S.D.N.Y. 2009); Barclays Capital Inc. v. Theflyonthewall.com, Inc., 650 F.3d 876 (2d Cir. 2011).
- 9.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537 (S.D.N.Y. 2013).
- 10.Quinn. Quinn's comment is also a relatively detailed analysis of the decision. That paper and this one are, however, still quite different. For instance, Quinn's comment is a less thorough critique of the court's reasoning, focusing on transformativeness rather than re-examining the full fair use calculus. Quinn's piece also contains other important differences, including an emphasis on the theory of implied consent, which this piece does not consider. Quinn is, nonetheless, quoted and cited where appropriate.
- 11.“AP, Meltwater Settle.”
- 12.Meltwater, 931 F. Supp. 2d at 545.
- 13.The decision points out “that Google News Alerts do not systematically include an article's lede and are—on average—half the length of Meltwater's excerpts.” Ibid., 555. Google News Alerts do sometimes contain the lede, however, as this is sometimes the first result for a given keyword.
- 14.Ibid., 546.
- 15.“About Us.”
- 16.17 U.S.C. § 107.
- 17.“Copyright Law Revision,” 65, H. R. Rep. No. 94-1476 (1976). (“Although the courts have considered and ruled upon the fair use doctrine over and over again, no real definition of the concept has ever emerged. Indeed, since the doctrine is an equitable rule of reason, no generally applicable definition is possible, and each case raising the question must be decided on its own facts. On the other hand, the courts have evolved a set of criteria which, though in no case definitive or determinative, provide some gauge for balancing the equities.”)
- 18.Campbell v. Acuff-Rose Music, 510 U.S. 569, 578 (1994).
- 19.Pub. L. 94-553, 90 Stat. 2541 (1976).
- 20.American Geophysical Union v. Texaco, Inc., 802 F. Supp. 1, 12 (S.D.N.Y. 1992), aff'd, 60 F.3d 913 (2d Cir. 1994).
- 21.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 551 (S.D.N.Y. 2013) (citing Infinity Broadcast Corp. v. Kirkwood, 150 F.3d 104, 108 & n.2 [2d Cir. 1998]).
- 22.Ibid. (citing Bill Graham Archives v. Dorling Kindersley, 448 F.3d 605, 611 [2d Cir. 2006]).
- 23.Ibid., 552 (citing Harper & Row Publishers, Inc. v. Nation Enterp., 471 U.S. 539, 562 ; ellipses in original).
- 24.Ibid., 552.
- 25.Leval, “Toward a Fair Use Standard,” 1111.
- 26.Campbell v. Acuff-Rose Music, 510 U.S. 569 (1994).
- 27.Bunker, “Transforming the News,” 311–12.
- 28.Tushnet. (“Transformativeness has indeed become almost synonymous with fairness.” Ibid., 870.)
- 29.Meltwater, 553. (“Meltwater's argument that its use of the Registered Articles is transformative is premised on a single contention. It characterizes Meltwater News as a search engine that directs users to a source of information online and whose search results provide insight into ‘where, when, how often, and in what context’ certain words or phrases appear on the Internet.”)
- 30.If anything, the Meltwater client is engaged in a purpose that is more consistently different from the primary intended purpose than that of a professor evaluating a book. It is hardly unheard-of for a professor to acquire evaluation copies with at least a partial interest in learning new materials themselves. In contrast, if a Meltwater subscriber actually wanted to use the service to learn about an issue in depth, she would have to click on the links and read the original articles.
- 31.For example, ibid., 556 (“the purpose of search engines is to allow users to sift through the deluge of data available through the Internet and to direct them to the original source. That would appear to be a transformative purpose”).
- 32.Kelly v. Arriba Soft Corp., 336 F.3d 811 (9th Cir. 2003).
- 33.Perfect 10, Inc. v. Amazon.com, Inc., 508 F.3d 1146 (9th Cir. 2007).
- 34.Kelly, 819.
- 35.The Kelly and Perfect 10 courts found zero or nearly zero substitution for the original, but the lesson from these cases is not and should not be that fair use only applies when there is zero or nearly zero substitution. The courts held that there was almost no substitution in those cases, and while this is highly helpful to a defense and was part of the court's reasoning, the courts did not hold that it is required. This is not least because a standard of zero substitutions would be highly unrealistic. For instance, if a critique of a copyrighted work is so damning that those who consume the critique will generally not consume the original work, this outcome does not preclude fair use.
- To pick an example from the world of image search, consider all of the instances in which a user can't quite picture something or someone—from an exhaust manifold to Ron Perlman. In these instances, the thumbnail is a direct substitute for the user who then need not click through to the original. That does not mean that Google Images is making an infringing use of millions of pictures of character actors, and if a photographer of one such image were to sue Google, the courts would very likely invoke Kelly and Perfect 10 and find for Google.
- 36.Perfect 10, 1165.
- 37.Campbell v. Acuff-Rose Music, 510 U.S. 569 (1994).
- 38.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 554 (S.D.N.Y. 2013) (emphasis added).
- 39.Oxford Dictionaries.
- 40.Brad Reed.
- 41.“Torrentz Search Engine.”
- 42.The court also notes that the debate about whether or not it is a search engine is not the same thing as whether or not it is copyright infringement. Meltwater, 556. (“Moreover, Meltwater's discussion of search engines is to some extent beside the point. While it is important to understand how Meltwater News functions, even if it were a search engine it would still be necessary to examine whether Meltwater had acted to violate the Copyright Act.”) This makes it all the more baffling that the decision draws all these novel and misguided lines in the sand about what does and does not count as a search engine, implying along the way that this is an important legal distinction. A much better starting point would have been, “Call it a search engine if you want; is it transformative?”
- 43.Meltwater, 556.
- 44.Meltwater, 555.
- 46.Grimmelmann, 939–40.
- 47.Meltwater, 554.
- 48.AV ex rel. Vanderhye v. iParadigms, LLC, 562 F.3d 630 (4th Cir. 2009).
- 49.Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379, 396–97 (S.D.N.Y. 2014). (“TVEyes subscribers use this service to comment on and criticize broadcast news channels.… Clearly, TVEyes provides substantial benefit to the public.” The court expresses no concern that only the members of the public that can pay for the service can so benefit.)
- 50.Quinn, 1214.
- 51.Ibid., 639.
- 52.Tushnet, 876.
- 53.Meltwater, 552. The court is implicit in the claim that, because there is no added commentary or insight, it is not transformative. Still, I am not the only one to read the implicit “because” into the claim. See Quinn, 1205 (“The court found the use was not transformative because Meltwater did not add commentary or insight, and just automatically captured and republished portions of copyrighted works.”), and Fox News v. TVEyes, 43 F. Supp. 3d at 391-92 (“The [Meltwater] court ruled that this use was not transformative because it ‘uses its computer programs to automatically capture and republish designated segments of text from news articles, without adding any commentary or insight in its New Reports’”).
- 54.Perfect 10 and Kelly were in the 9th Circuit.
- 55.Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379, 392 (S.D.N.Y. 2014).
- Not any re-use of news is part of or targeted at transformative research. Perhaps the closest example of non-transformative re-use is presented by the financial news and reporting cases Nihon Keizai Shimbun, Inc. v. Comline Business Data, Inc., 166 F.3d 65, 73 (2d Cir. 1999) and Barclays Capital, Inc. v. Theflyonthewall.com, Inc., 650 F.3d 876 (2nd Cir. 2011).
- In both of these cases, the aggregator was engaged in selling access to portions of works protected by copyright, expressly competing in the same market for such works. Comline was selling roughly translated Nihon news reports as news reports. Theflyonthewall.com was selling excerpts from financial firms’ reports as financial analysis reports. In this way, the defendants were sued for engaging in a highly similar purpose, competing in the same overall market as the market served by the originals. Importantly, both of those defendants sought to serve their clients with the protectable expression in the original, providing the same benefit as if the clients had consumed the original. When a highly trusted source of financial news provides a report, word choices move markets; thus, including the specific expression (even as translated, as in Nihon) seeks to convey the same insight that results in part from the protectable parts of the articles, for the same purpose of conveying that expression to investors.
- The TVEyes case, in contrast, shows a service that is transformative enough to lend toward a finding of fair use, pending the other factors. In TVEyes, the protectable expression is also of potential import to the service's clients, but the clients come with a research purpose—with a desire to monitor, criticize, compare, and so on. The difference between Nihon and Theflyonthewall on one hand, and TVEyes on the other, is striking. In this light, the error of the Meltwater court is perhaps more visible. Of course, TVEyes may have benefitted from some combination of better counsel, a more favorable judge, and more favorable precedent (TVEyes was decided after the defense-friendly HathiTrust decision, discussed in the next section).
- 56.Quinn, 1211.
- 57.Meltwater, 544.
- 58.Ibid., 556.
- 59.Ibid., 561.
- 60.Authors Guild, Inc., v. HathiTrust, 755 F.3d 87 (2nd Cir. 2014).
- 61.The lower court ruling, Authors Guild, Inc. v. Hathitrust, 902 F. Supp. 2d 445 (S.D.N.Y. 2012), was filed before the Meltwater ruling. The lower court was corrected, in part, on its theory of transformativeness as it relates to the first factor. This correction, had it come before the Meltwater ruling, could have provided valuable guidance. Even the lower court's reasoning in Hathitrust was far more sympathetic to search services than that in Meltwater, but the difference is more stark between Meltwater and the 2nd Circuit's Hathitrust reasoning. To say nothing of the fact that, had the appeals court decided a year sooner, the decision would have provided a binding precedent for the Meltwater court.
- 62.Authors Guild, Inc., v. HathiTrust, 755 F.3d at 97 (2nd Cir. 2014).
- 64.Ibid., 96. This reading of transformativeness adds a great deal of clarification and predictability, in sharp contrast to earlier mistakes—some by none other than the Second Circuit Court of Appeals. For instance, in Castle Rock Ent. v. Carol Publ'g Group, Inc., 150 F.3d 132 (2d Cir. 1998), the court held that a trivia book about the TV show Seinfeld is not sufficiently transformative, apparently because the judges found the book to be lacking in cultural gravitas. As one scholar concludes, “The Second Circuit's confused application of the transformative use doctrine would result in a search for authorial motives, whether explicit or implicit, rather than an examination of more objective features of the borrowing work.” Bunker, “Adventures in the Copyright Zone.”
- 65.Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379 (S.D.N.Y. 2014).
- 66.Ibid., 393.
- 67.Fox News Network, LLC v. TVEyes, Inc., 2015 U.S. Dist. LEXIS 112836 (S.D.N.Y. 2015). In particular, the court found the service's e-mail, downloading, and (surprisingly) date-time search features not to be fair use. Ibid., *20–28.
- 68.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 554 (S.D.N.Y. 2013).
- 69.Ibid., 547. For more on the methodology that lead to the 0.08 percent figure, see ibid., n. 5. The court also highlighted a UK tribunal finding of 0.5 percent, ibid., 554, but with less detail explaining how this latter rate was calculated.
- 70.One would also, of course, hope for a much, much larger sample size of articles.
- 71.Google News was launched in 2002, and users could sign up for customized news alert e-mails no later than August 2004. See Galvin. Meltwater started offering their comparable service in 2005. Meltwater, 543. Google's service is so close an analog that one wonders why anybody would pay Meltwater for the same, if this is indeed the heart of their value proposition. It seems highly improbable that the extra text per article, alone, would justify the substantial expense for many—let alone all—Meltwater clients.
- 72.Ibid., 554, n.13, citing an amicus brief by New York Times et al.
- 73.Franssila, Okkonen, and Savolainen.
- 74.Kalman and Ravid.
- 75.In fairness to Judge Cote, the primary burden to do this research falls on the defendants and their attorneys, who—based on my reading of the key filings and transcripts—seem not to have introduced it into the record at all. This is consistent with an overall performance by the defendants and/or their attorneys that is disappointing in general and especially so in regards to their handling of the discovery process.
- 76.“What Are the Average Open.”
- 77.“Constant Contact's Email Permission Policy.”
- 78.“Constant Contact Anti-Spam Policy.”
- 81.Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379, 392 (S.D.N.Y. 2014).
- 82.Quinn, 1212.
- 83.See Harper & Row, Publishers, Inc. v. Nation Enters., 471 U.S. 539 (1985).
- 84.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 557 (S.D.N.Y. 2013) (citations omitted).
- 85.Ibid. (citing Nihon Keizai Shimbun, Inc. v. Comline Business Data, Inc., 166 F.3d 65, 73 [2d Cir. 1999]).
- 86.Ibid., 561.
- 87.Nihon, 166 F.3d at 73.
- 88.Nihon Keizai Shimbun, Inc. v. Comline Bus. Data, 1998 U.S. Dist. LEXIS 6806 (S.D.N.Y. Apr. 14, 1998).
- 89.In Cote's Nihon decision, she does ask whether the works are copyrightable at all. Ibid., 31–35. However, mere copyrightability does not determine the thickness of protection merited under the second fair use factor.
- 90.Nihon, 166 F.3d at 73.
- 91.In each of these rulings, the court's reasoning on this factor comes in dicta rather than in the holding; to my knowledge, the highly factual nature of a work has not been central to any of the conclusions it has reached. Nonetheless, even though these holdings are not binding on this question, they are persuasive authorities. For a theory of persuasive authority, see Flanders.
- Further, even in a lecture cautioning the audience against conflating holdings with dicta, Leval writes, “dicta often serve extremely valuable purposes. They can help clarify a complicated subject. They can assist future courts to reach sensible, well-reasoned results. They can help lawyers and society to predict the future course of the court's rulings.” Leval, “Judging Under the Constitution,” 1253.
- When the US Supreme Court has issued consistent dicta on a given issue in several cases, this should be treated as quite persuasive authority indeed. Thus, lower courts should take quite seriously the high court's repeated statements that it is far easier to make fair use of a factual work than a work of fiction.
- 92.Stewart v. Abend, 495 U.S. 207, 237 (1990).
- 93.Sony Corporation of Amer. v. Universal City Studios, Inc., 464 U.S. 417, 455 n.40 (1984).
- 94.Campbell v. Acuff-Rose Music, 510 U.S. 569, 586 (1994).
- 95.Harper & Row Publishers, Inc. v. Nation Enterp., 471 U.S. 539, 563 (1985).
- 97.See also, for example, Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379, 394 (S.D.N.Y. 2014) (“[T]here is ‘greater leeway’ for a determination of fair use when the work is factual or largely informational. In these cases, the scope for fair use is greater.”).
- 98.The only citation LexisNexis Academic returns for the word Nihon and the phrase “at most neutral” is in another decision's attempt to find a token difference of opinion on this question, contrasted with several other decisions that state clearly that copied works’ highly factual nature does weigh substantially in favor of the defendant. The decision, LA Times v. Free Republic, 2000 U.S. Dist. LEXIS 5669 (C.D.C.A., 2000), reads:
- A number of cases that have analyzed alleged copying of news articles or videotapes of news events have concluded that the second fair use factor weighs in the defendant's favor. See Reuters Television, supra, 149 F.3d at 994 (the court held that the second factor weighed in favor of defendants that copied news footage); Los Angeles News Service v. KCAL-TV Channel 9, 108 F.3d 1119, 1122 (9th Cir. 1997) (the second factor weighed in favor of a finding of fair use where defendants copied news footage); Los Angeles News Service v. Tullo, 973 F.2d 791, 792, 798 (9th Cir. 1992) (the second factor favored a video news clipping service that used portions of copyrighted videotapes of newsworthy events). See also American Geophysical, supra, 60 F.3d at 925 (given the “manifestly factual character of the … articles” from scientific and medical journals copied by defendant, the court held that the second factor weighed in favor of fair use); Television Digest, supra, 841 F. Supp. at 10 (the court found that the second factor weighed in favor of a defendant that copied a newsletter containing original news stories). Compare Nihon Keizai Shimbun, supra, 166 F.3d at 72–73 (in a suit by a newspaper publisher against a defendant that gathered news articles from various sources and sold “abstracts” of them to its customers, the court recognized that newspaper articles are predominantly factual in nature and that expressive elements do not dominate, but nonetheless concluded that the second “factor is at most neutral on the question of fair use”).
- Ibid., *54–56.
- 99.Meltwater, 931 F. Supp. 2d at 553 (emphasis added).
- 100.Ibid. (emphasis added).
- 101.Feist Publications, Inc. v. Rural Telephone Service Co., Inc., 499 U.S. 340, 352–60 (1991).
- 102.Rural Telephone Service Co. v. Feist Publications, Inc., 663 F. Supp. 214 (Kan. 1987) (summary judgment for plaintiff); affirmed, 916 F.2d 718 (10th Cir. 1990).
- 103.Feist, 499 U.S. at 352–60.
- 104.Ibid., 348 (ellipses added, citations omitted).
- 105.See, for example, Spark and Harris, 50 (“Start with facts rather than comment.”).
- 106.Harper & Row Publishers, Inc. v. Nation Enterp., 471 U.S. 539, 563 (1985).
- 107.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 572 (S.D.N.Y. 2013).
- 109.“Toledo Casino Posts 600 Job Openings.”
- 110.Kelly; “Toledo Casino Posts Job Openings.”
- 111.Elizabeth Reed.
- 114.Nihon Keizai Shimbun, Inc. v. Comline Bus. Data, 1998 U.S. Dist. LEXIS 6806, *35 (S.D.N.Y. Apr. 14, 1998).
- 115.17 U.S.C. § 107(3).
- 116.Harper & Row Publishers, Inc. v. Nation Enterp., 471 U.S. 539, 565 (1985).
- 117.Meltwater, 545.
- 118.Ibid., 547.
- 120.Ibid., 558.
- 121.Ibid., 559. Emphasis is added to highlight how anxious Judge Cote is to find for AP, even while still nominally analyzing the case.
- 122.Ibid., 555 (“The sample Google News Alert excerpts provided with this motion are on average roughly 207 characters long [not including white spaces].”). Ibid., n. 16.
- 123.Ibid., 558–59.
- 124.Salinger v. Random House, Inc., 811 F.2d 90, 97 (2d Cir. 1987) (emphasis in original).
- 125.“UIPM Approves No Needle Policy.”
- 126.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 547 (S.D.N.Y. 2013).
- 127.For example, Heffernan.
- 128.Meltwater, 98.
- 129.Salinger v. Random House, Inc., 811 F.2d 90, 97 (2d Cir. 1987).
- 130.17 U.S.C. § 107(4).
- 131.Leval, “Toward a Fair Use Standard,” 1124–25.
- 132.American Geophysical Union v. Texaco, Inc., 60 F.3d 913, 929-31 (2d Cir. 1994).
- 133.AP v. Meltwater U.S. Holdings, Inc., 931 F. Supp. 2d 537, 557 (S.D.N.Y. 2013) (citing Am. Geophys. Union, 60 F.3d at 930).
- 134.Ibid., 560.
- 135.Ibid. (citing Bill Graham Archives, 448 F.3d at 614–15).
- 136.Fox News Network LLC v. TVEyes Inc., 43 F. Supp. 3d 379, 395 (S.D.N.Y. 2014).
- 137.The nearest example is Cambridge University Press v. Becker, 863 F. Supp. 2d 1190 (N.D.Ga. 2012), which found that Georgia State University's usage of academic writings in the school's e-reserves system was mostly fair use—finding infringement in just 5 of the 75 alleged instances of infringement. In those five instances, excerpts provided to the university's students took excessive amounts from undoubtedly expressive works. I have trouble seeing any similarly clear instances of infringement in the Meltwater decision for the reasons discussed in this article, though one could certainly argue that, for the shortest articles, Meltwater's takings are outsized and therefore infringing.
- 139.Samuelson, 832–33. She writes, “The Second and Ninth Circuits have decided such uses are not only fair, but also transformative, as illustrated by the next three cases.” She lists Kelly, Perfect 10, and HathiTrust.
- 140.Quinn, 1209–10.
- 141.Balganesh. Balganesh observes, “courts in most jurisdictions remain unwilling to award defendant's attorney's fees unless convinced of the plaintiff's blameworthiness or culpability in bringing the suit. As long as a plaintiff has a reasonable copyright claim, such culpability is extremely hard to establish—and is usually associated with ‘bad faith motivation,’ ‘hard-ball tactics,’ or ‘objective unreasonableness.’” Ibid., 2323.
- 142.Campbell v. Acuff-Rose Music, 510 U.S. 569 (1994).
- 144.Samuelson. For a contrary view on this, see Bunker, “Eroding Fair Use.” Bunker's critique is in no small part tethered to the inconsistencies with which the lower courts draw the line between transformative and non-transformative uses, and on this count this article provides more fuel for Bunker's glass-half-empty reading than Samuelson's glass-half-full version.
- 145.Mattel Inc. v. Walking Mountain Productions, 353 F.3d 792 (9th Cir. 2003), on remand 2004 U.S. Dist. LEXIS 12469 (C.D. Cal.). The artist, Thomas Forsythe, created several parodies of the copyrighted work at issue, Mattel's Barbie doll. When the artist won at trial and Mattel appealed anyway, the 9th Circuit reversed and remanded the non-award of attorney's fees to Forsyth. The district court then ordered Mattel to pay $1,584,089 in attorney's fees and $241,797.09 in costs.
- 146.Brownmark Films, LLC v. Comedy Partners, No. 10-CV-1013, 2011 U.S. Dist. LEXIS 72684.
- 148.For example, in an interview, Tom Forsythe, the defendant in Mattel v. Walking Mountain, remarks, “People [artists] don't feel particularly safe … [you] should always be a little wary.” Somerstein, 574 (paraphrases in original).
- 149.Carrier. Carrier's study is specific to music, but its results lead us to worry about the negative impact of litigation on digital innovation more generally.
- 150.Doctor, “Newsonomics of Recycling Content” (emphasis in original).
- 152.Gandy Jr.
- 153.For example, McChesney and Nichols. Not only do the authors not suggest increasing copyright-related licensing revenue, they strongly advise that newspapers engage in less—not more—enforcement of exclusivity, for instance by not enacting impenetrable paywalls that cut newspapers off from the rest of the web. Ibid., 73. The few years since the book's release have seen many newspapers adopt a sort of tiered paywall, in which regular readers will need to subscribe (or jump through many hoops), but irregular readers of a given paper can get a few articles, for example, via hyperlinks.
- Also, while most newspapers do continue to need additional reforms to stay solvent, it is encouraging that at least the New York Times has managed to bring enough digital ad revenue and digital subscriptions that, combined with cost cutting, it made a $16 million profit in the second quarter of 2015. Somaiya.
- 154.“AP, Meltwater Settle.”
- Balganesh, Shyamkrishna. “Copyright Infringement Markets.” Columbia Law Review, 113 (2013): 2322–24.
- Bohannan, Christina. “IP Misuse as Foreclosure.” Iowa Law Review, 96 (2011): 475–527.
- Bunker, Matthew D. “Adventures in the Copyright Zone: The Puzzling Absence of Independent First Amendment Defenses in Contemporary Copyright Disputes.” Communication Law and Policy 14, no. 3 (2009): 273–301. doi:10.1080/10811680903000963.
- Bunker, Matthew D. “Eroding Fair Use: The ‘Transformative’ Use Doctrine after Campbell.” Communication Law and Policy 7, no. 1 (2002): 1–24. doi:10.1207/S15326926CLP0701_01.
- Bunker, Matthew D. “Transforming the News: Copyright and Fair Use in News-Related Contexts.” Journal of the Copyright Society of the USA, 52 (2005): 309–27.
- Carrier, Michael. “Copyright and Innovation: The Untold Story.” Wisconsin Law Review, 2012 (2012): 891–962.
- Chaffney, Dave. “Display Advertising Clickthrough Rates.” Smart Insights. Last updated April 21, 2015. http://www.smartinsights.com/internet-advertising/internet-advertising-analytics/display-advertising-clickthrough-rates.
- Cisco. “Cisco Global Cloud Index: Forecast and Methodology 2013–2018 White Paper.” Cisco Systems Inc. Accessed July 16, 2015. http://www.cisco.com/c/en/us/solutions/collateral/service-provider/global-cloud-index-gci/Cloud_Index_White_Paper.pdf.
- Cisco. “Visual Networking Index IP Traffic Chart.” Cisco Systems Inc. Accessed July 16, 2015. http://www.cisco.com/assets/cdc_content_elements/networking_solutions/service_provider/visual_networking_ip_traffic_chart.html.
- Davis, Aeron. Public Relations Democracy: Public Relations, Politics and the Mass Media in Britain. New York: Manchester University Press, 2002.
- Doctor, Ken. “The Newsonomics of Where NewsRight Went Wrong.” Nieman Journalism Lab. May 15, 2013. Accessed July 16, 2015. http://www.niemanlab.org/2013/05/the-newsonomics-of-where-newsright-went-wrong.
- Flanders, Chad. “Toward a Theory of Persuasive Authority.” Oklahoma Law Review, 62 (2009): 55–88.
- Franssila, Heljä, Jussi Okkonen, and Reijo Savolainen. “Email Intensity, Productivity and Control in the Knowledge Worker's Performance on the Desktop.” In Proceedings of the 18th International Academic MindTrek Conference: Media Business, Management, Content & Services, 19–22. New York: ACM Press, 2014. http://dl.acm.org/citation.cfm?doid=2676467.2676513.
- Galvin, Nick. “Webs We Weave.” The Age (Melbourne, Australia), August 19, 2004.
- Gandy Jr., Oscar H. Beyond Agenda Setting: Information Subsidies and Public Policy. Communication and Information Science. Norwood, NJ: Ablex Pub. Co., 1982.
- Gaytan, Miguel. “The Hollywood Casino is Looking for Employees.” 13 ABC (Toledo, OH). Last updated December 28, 2011. Accessed October 31, 2013 (broken link). http://www.13abc.com/story/16323375/the-hollywood-casino-posts-is-looking-for-employees.
- “Global Top 100 Companies by Market Capitalisation: 31 March 2015 Update.” PriceWaterhouseCoopers. Accessed July 16, 2015. http://www.pwc.com/gx/en/audit-services/capital-market/publications/assets/document/pwc-global-top-100-march-update.pdf.
- Grimmelmann, James. “Speech Engines.” Minnesota Law Review, 98 (2013): 868–952.
- Jackson, Matt. “Commerce Versus Art: The Transformation of Fair Use.” Journal of Broadcasting and Electronic Media, 39 (1995): 190–99.
- Kalman, Yoram M., and Gilad Ravid. “Email Inbox Management by Information Overloaded Users.” In Proceedings of the Companion Publication of the 17th ACM Conference on Computer Supported Cooperative Work & Social Computing, 185–88. New York: ACM Press, 2014. http://dl.acm.org/citation.cfm?doid=2556420.2556478.
- Kanzelberger, Jennifer. “Brownmark Films, LLC. v. Comedy Partners: Why Fair Users Should Be Able to Rely On Fair Use.” North Carolina Journal of Law & Technology, 13 (2011): 35–60.
- Leval, Pierre N. “Judging Under the Constitution: Dicta About Dicta.” NYU Law Review, 81 (2006): 1249–82.
- Leval, Pierre N. “Toward a Fair Use Standard.” Harvard Law Review, 103 (1990): 1105–36.
- Lichtman, Doug. “Google Book Search in the Gridlock Economy.” Arizona Law Review, 53 (2011): 131–43.
- Logan, Wayne A. “Constitutional Cacophony: Federal Circuit Splits and the Fourth Amendment.” Vanderbilt Law Review, 65 (2012): 1137–95.
- McChesney, Robert W., and John Nichols. The Death and Life of American Journalism: The Media Revolution That Will Begin the World Again. 1st Trade Paper Ed. New York: Nation Books, 2011.
- Olson, David S. “First Amendment Based Copyright Misuse.” William and Mary Law Review 52 (2010), 537–606.
- Patry, William F. How to Fix Copyright. New York: Oxford, 2011.
- Quinn, Dylan J. “Associated Press v. Meltwater: Are Courts Being Fair to News Aggregators?” Minnesota Journal of Law, Science, & Technology, 15 (2014): 1189–220.
- Samuelson, Pamela. “Possible Futures of Fair Use.” Washington Law Review, 90 (2015): 815–68.
- Sinnreich, Aram. The Piracy Crusade: How the Music Industry's War on Sharing Destroys Markets and Erodes Civil Liberties. Amherst, MA: University of Massachusetts Press, 2013.
- Somerstein, Liz. “Who Is Really Protecting Barbie: Goliath or the Silver Knight? A Defense of Mattel's Aggressive International Attempts to Protect Its Barbie Copyright and Trademark.” The University of Miami Inter-American Law Review, 39 (2008): 559–85.
- Spark, David B., and Geoffrey Harris. Practical Newspaper Reporting, 4th ed. London: Sage, 2010.
- Stambor, Zak. “Facebook Ads’ Click-Through Rates Soar in Q3: Retailers Enjoy Strong Returns on Their Ads, a New Report Suggests.” Internet Retailer, October 17, 2013. http://www.internetretailer.com/2013/10/17/facebook-ads-click-through-rates-soar-q3.
- “Toledo Casino Posts 600 Job Openings.” The Blade (Toledo, OH), December 14, 2011.
- Tushnet, Rebecca. “Content, Purpose, or Both.” Washington Law Review, 90 (2015): 869–92.
- Van Vleet, Jacob E. Informal Logical Fallacies: A Brief Guide. Lanham, MD: United Press of America, 2010.
- “What Are the Average Open, Click-Through, and Bounce Rates of Other Constant Contact Customers?” Last modified February 2015. Accessed July 16, 2015. http://support2.constantcontact.com/articles/FAQ/2499.