Response

Debunking the Myth of the Copyright Troll Apocalypsee

I.      Introduction

Professor Matthew Sag’s Copyright Trolling, An Empirical Study tells a riveting tale of a litigation system run amok.1 A plaintiff files suit in federal court. Each instance of alleged unlawful conduct targeted in the suit may well entail little in the way of actual damages, and for that reason the complaint demands statutory rather than actual damages. The conduct in question is as common as it is allegedly unlawful and, in some people’s views, this conduct isn’t particularly objectionable. And the economics of the litigation in question simply don’t make sense without the aggregation of claims related to many individuals. Defendants in suits like this one have more—very much more—to lose than to gain by fighting the suit, so they feel intense pressure to settle early, even if the suit is totally meritless.

If you think you’ve seen this movie before, that’s because you probably have. Critics have complained that the features above characterize aggregate litigation in antitrust, securities litigation, drug and medical device liability, mass accidents, and who knows how many other theaters. What’s new about Sag’s interesting article is that he argues that, in the context of copyright law, these controversial features of aggregate litigation amount to “trolling,” a phenomenon that has invited the ire of courts and scholars in the last few years.2 All the old story’s villains are there, with both plot and technological twists—and trolls lurking under the Internet’s bridge devices!—all providing the contemporary hooks that might lure in new viewers.

We commend Sag for taking a deeper look into the copyright trolling phenomenon, and for basing his analysis on the statistics of copyright litigation. Copyright thinking and policy-making surely would benefit immensely from more data-driven, empirical approaches. Yet, Sag’s analysis fails to identify the copyright trolling phenomenon with sufficiently workable analytical precision, and his empirical analysis compounds this problem.

Part II of this Essay addresses the problems inherent in Sag’s conduct-based definition of copyright trolling. His purported move from a status-based conception of trolling to a conduct-based one makes it exceptionally hard to distinguish instances of zealous copyright enforcement from copyright trolling. As a result, he sweeps in wide swaths of litigation that we do not think can fairly be criticized (at least, without reference to the pornographic content of the copyrighted works involved, which we recognize some people regard negatively).

Part III then investigates and assesses the empirical observations that Sag draws from his data. Unfortunately, Sag’s data do not support his claims about the role of joinder, or even about trends in anything that can be considered representative of the overall phenomenon of copyright trolling. The data turn out to be driven substantially by one plaintiff, Malibu Media, LLC (“Malibu”), which sues few defendants at a time, and which hardly seems to meet any plausible definition of a copyright troll. An important shortcoming in Sag’s empirical work is that he mixes up units of analysis. On its face, his story is directed at explaining the growth in filings of a particular type of copyright lawsuits, yet important parts of his empirical discussion instead focus on the number of defendants sued in such lawsuits. Because there are a few lawsuits in which many thousands of defendants are sued, and because the overwhelming number of lawsuits involve few defendants, this mismatch of object and discussion is highly consequential.

At the risk of stating the obvious, we agree that the law should identify and curb abusive copyright litigation. Copyright trolling, to the extent that it eviscerates the copyright system’s core goals and values, remains a deeply problematic phenomenon that courts are yet to develop a coherent strategy to curb. But by identifying the phenomenon too broadly, by failing to appreciate the importance of the runaway leader in what Sag calls “John Doe–Porn” lawsuits,3 and by conflating units of analysis, Sag’s article seems to greatly overstate the extent of copyright trolling within U.S. copyright litigation.4

II.     A “Conduct” Based Approach to Copyright Trolling

A.     Status Versus Conduct

We begin by trying to identify just who a copyright troll actually is according to Sag’s description of the phenomenon. Sag is certainly right to identify the first wave of copyright trolling with the purported assignment of claims to third parties who have no interest in the work other than in its enforcement.5 This in turn produced a separation between the incentives to create and litigate, with the former never moving from the author/‌owner to the third party. One of us previously offered a definition of the copyright troll as “an entity whose business revolves around the systematic legal enforcement of copyrights in which it has acquired a limited ownership interest.”6 The idea behind this definition was less to cabin the phenomenon based on formal criteria—such as the presence of a putative assignment (e.g., Righthaven7)—than to identify the fulcrum of trolling. We think this fulcrum lies in the disaggregation of copyright’s incentive to create (vested in the owner) and the incentive to litigate (vested in a third party).

As Sag rightly notes, the existence of law firms that work on a contingency basis achieves the same result as a formal assignment of rights, since such a law firm might indeed make decisions about when/‌whom to sue for copyright infringement.8 But such a law firm would satisfy the disaggregation-based characterization of trolling rather well, insofar as the law firm (rather than the creator) maintains control over the lawsuit. Sag characterizes the disaggregation-based approach to understanding trolling as a “status” based conception, painting it as unduly formalistic.9 Yet it is hardly the case that a party’s status as a troll is capable of being ascertained independent of its conduct. The supposedly “status-based” conception depends importantly on an actor’s conduct, both in the creative process (or lack thereof) and in control over the litigation at hand (whether by a de jure assignment or a de facto one). In the end, it appears that Sag’s real issue is with the disaggregation approach’s focus on disruptions to the equilibrium of “tolerated use[s].”10 We say this because it is clear from Sag’s article that he considers lawsuits brought by creative copyright owners themselves to potentially involve trolling.11 To get to this result, he offers his own “conduct” based approach, which we discuss below.

B.     Opportunism and a Cluster of Attributes

Who, then, is a copyright troll in Sag’s understanding? Here, Sag’s approach offers little conceptual guidance. He rejects prior definitions of trolling, since they would “exclude the majority of copyright troll litigation.”12 But that seems to beg the question: How do we know that litigation is trolling, without an a priori characterization of trolling? The first characteristic of a troll that Sag identifies is that of “systematic opportunis[m].”13 Opportunism can, however, mean a variety of different things. The scholarship that Sag relies on for this understanding builds on the work of Henry Smith, and characterizes opportunism as behavior that “us[es] the letter of the law to achieve objectives that are inconsistent with the law’s purpose, and in doing so creat[es] net social costs.”14 Translated into the context of copyright trolls, the ideal of opportunism could be seen to trace what Sag dismisses as the “status” based conception, under which the troll uses the letter of copyright law (i.e., its allowance for an infringement action for every instance of wrongful copying) in a way inconsistent with the law’s purpose (i.e., that the infringement action ought to be brought in furtherance of copyright’s incentive to create), generating net social costs (i.e., a reduction in tolerated uses and downstream creativity).

To Sag however, opportunism entails “basing a business on litigation.”15 Nowhere in the paper however, does Sag tell us why this is so. As a general matter, the premise appears to partake of the notion that litigation is a necessary evil in the copyright context, one that ought to be avoided unless absolutely necessary. If this is the core assumption motivating Sag’s idea behind opportunism, then it deserves further explication. It might well mean that Sag isn’t opposed just to trolling, but perhaps to all forms of market-based third party involvement in lawsuits (e.g., including contingency lawyering for copyright plaintiffs and third party litigation funding in the area). Placing reliance on the ephemeral ideal of opportunism isn’t particularly satisfying on its own as an analytical hook to identify trolling.

Sag does offer something of a definition for copyright trolls when he notes that “[c]opyright trolls are best defined by a cluster of attributes rather than any single definitive feature.”16 In his conduct-based conception, “[a] troll that deserves its name asserts rights it does not have, makes poorly substantiated claims, or seeks disproportionate remedies” and “do[es] at least one of these things systematically.”17 Let us consider each of these elements in turn.

First, the “asserts rights it does not have” criterion may be easy to make sense of in hindsight, but it is not obviously a workable criterion. What Sag seems to have in mind is an entity that knows well in advance that its claims are non-existent, but nonetheless chooses to sue for infringement. An entity that goes after obvious instances of fair use might be a good example here. All the same, none of Sag’s examples point to such behavior, where a plaintiff brings altogether non-existent or frivolous claims—what might reasonably be called strike suits. Even with a copyright plaintiff that is obviously devious based on ex post facts—e.g., Righthaven—it can be difficult ex ante to be sure its actions are brought with the knowledge that they were fraudulent. Indeed, it was only when the courts eventually found such a lawsuit that they managed to rein in Righthaven.18

The reason that Sag’s first attribute is problematic is because it seems to fly in the face of one of the copyright system’s fundamental structural attributes: its reliance on litigation to ascertain the existence and scope of the plaintiff’s rights.19 Unlike patent and trademark law, which involve a formal grant of rights upon scrutiny by an administrative agency, protection under copyright law commences from the moment of creation and fixation in a tangible medium of expression. Registration serves as a mere formality, meaning that it is only ever in court that a plaintiff gets to have its rights formally delineated for the first time. Given the nature of the federal judicial process, a plaintiff seeking to push the boundaries of copyright law for its rights—for instance, an entity seeking protection for novel subject matter—might well try its luck in a variety of different courts, in an effort to have an appellate court or the Supreme Court eventually weigh in favorably. Would such a plaintiff now be systematically asserting rights that it does not have, simply because eight out of ten district courts around the country have ruled for defendants? Take Ira Arnstein, the well-known copyright plaintiff who earned a reputation for commencing copyright infringement claims against successful composers, accusing them of having plagiarized his music.20 Most courts dismissed his claims; some even characterized them as “fantastic[a].”21 In Sag’s account, Arnstein might well be a copyright troll, rather than just a misguided, maybe even delusional, copyright plaintiff.

This leads us to the second element in Sag’s definition: the requirement that the troll make poorly substantiated claims. Poorly substantiated claims might arise for a variety of reasons, including bad lawyering, unclear law, cost-cutting litigation strategy, or simply weak claims. Sag clearly has the last of these in mind. Again, we are presented with a hindsight problem. A claim might well be weak ex post, but hardly ex ante.22 Again, what Sag seems to be getting at is the situation where the plaintiff knows the claim to be weak when bringing the action and therefore chooses to substantiate it poorly. Here, Sag equivocates on the question of whether the weakness so to speak is in the abstract (as a question of law), or in a relative sense, i.e., against the particular defendant (as a question of fact). If the former, we have the same problem as with Sag’s first proposed attribute, since copyright’s entitlement structure is vividly unclear. If the latter though, it isn’t clear why additional discovery might not solve the issue. What Sag has in mind is of course the phenomenon of multi-defendant John Doe (“MDJD”) porn trolling, where the action is commenced against an entire “swarm” of BitTorrent-based file sharers without further evidence of every individual defendant’s involvement, under permissive joinder rules.23 If that is a problem, it has little to do with a conscious decision to poorly substantiate the claim as Sag’s definition requires, but is instead an allowance in the law.24

Sag’s third requirement—seeking disproportionate remedies­­—fares no better. “Disproportionate” to what exactly? Again, what Sag seems to have in mind is plaintiffs who seek exorbitant statutory damages when their actual harm is negligible or non-existent. Yet, that is known to be true of a good many plaintiffs. As Sag himself seems to acknowledge later on,25 statutory damages can—and often do—serve a punitive/‌deterrent function that transcends their compensatory or restitutionary basis. When this happens, the argument for disproportionality starts becoming even more tenuous. A plaintiff may begin a series of lawsuits, systematically, under a “deterrence campaign,” in order to curb behavior of a certain kind. In Sag’s conception, such a plaintiff could well be a troll (we discuss the Recording Industry Association of America’s (“RIAA”) deterrence campaign in the next section of this Part).

Sag qualifies his “cluster of attributes” approach by the requirement that at least one attribute be done “systematically.”26 Unfortunately, it isn’t clear what exactly Sag means by this requirement. We take it to suggest that the plaintiff’s behavior isn’t restricted to an isolated lawsuit, meaning that the plaintiff’s behavior extends to multiple lawsuits that exhibit at least one of his attributes. In that understanding, however, the requirement of systematic behavior does little to remedy the pervasive ambiguity in his criteria.

In the end, Sag’s definition appears tailor-made for the primary form of nefarious behavior that he identifies and critiques in his paper: John Doe copyright lawsuits filed by porn producers. Once understood as limited to that particular context, and the particular actors involved in that domain, his attributes start making rhetorical sense, even if not analytical (or, as we shall see, empirical).

C.     The Real Trolling Attribute: Litigation as a Content-Independent Revenue Stream

On the face of things, the RIAA’s famous deterrence campaign against music downloaders would result in the RIAA being classified as a copyright troll.27 It often brought poorly substantiated claims seeking disproportionate damages on a systematic basis. Sag seems to acknowledge this, since he implies that the reason why the RIAA was not a troll isn’t because they fail his cluster of attributes, but because its “end-user litigation strategy was to send a message, not create an independent revenue stream.”28 Creating an independent revenue stream through litigation thus appears as an additional qualifying element to being a copyright troll. Sag imposes this condition importantly to distinguish between plaintiffs who seek to bring defendants back into the market, and those that seek to create a revenue stream through an alternative distribution channel.

Litigation as a revenue stream independent of any content creation thus seems to lie at the heart of Sag’s problems with copyright litigation that he regards as trolling, an idea not adequately captured by his other criteria. This idea goes back to the assumption, then, that litigation is an avoidable evil in the copyright system.29 Yet, this idea is precisely what makes the case against copyright trolls hard to make and maintain. In a wide variety of areas, the law continues to remain agnostic to the question of whether litigation—in general—should be used as an independent revenue stream. The emergence of third-party litigation funding, and the debates surrounding investments in profitable lawsuits, has in recent times brought this question to the surface.30 Lawyers and scholars remain divided on the issue. One school of thought argues that it will clog up the courts with frivolous lawsuits, while another asserts that it will help plaintiffs who cannot afford lawsuits to assert their rights.

Sag’s assertion that a litigation-based revenue stream, especially for authors without an established market, deserves further elaboration. It involves normative considerations of fairness and distributive justice, which a focus on the positive economics of litigation might not fully capture. To the extent that other mechanisms such as grant funding and run-of-the-mill market incentives provide creators with a level of support for their creative efforts that is socially sub-optimal, litigation might well serve copyright’s promise of optimally incentivizing creative expression. Whether this is true or not is both a normative question—how much expression and what kind do we want?—and an empirical one—are there enough non-litigation incentives to induce the desired results? While Sag may have strong intuitions about the viability of the argument, it would be well if he spelled out his arguments more clearly.

III.   Trends in Large-N John Doe Lawsuits: A Reassessment

Sag’s main empirical point is well summarized in the following passages:

This empirical study of copyright trolling presents new data that reveal[] the astonishing growth of a particular kind of copyright trolling—the multi-defendant John Doe (“MDJD”) lawsuit that alleges copyright violation through the file sharing software known as BitTorrent. Generally, these suits take the form of “Copyright Owner v. John Does 1–N” where N is a large number. MDJD suits are not just any form of copyright trolling, they are the dominant form. In 2013, these MDJD suits were the majority of copyright cases filed in 19 of the 92 federal districts. . . .

The essence of trolling is that the plaintiff is more focused on the business of litigation than on selling a product or service or licensing their IP to third parties to sell a product or a service. The paradigmatic troll plays a numbers game in which it targets hundreds or thousands of defendants, seeking quick settlements priced just low enough that it is less expensive for the defendant to pay the troll rather than defend the claim. . . .

Of the 3817 copyright law suits filed in 2013, over 43% were against John Does and more than three-quarters of those related to pornography. . . . [T]he association with pornography is far from coincidental. The MDJD model works especially well for pornography because the potential embarrassment of being accused (accurately or not) of downloading such material is a great motivation to settle.31

As we see it, Sag makes six key empirical or conceptual claims in these passages:

1.     There has been astonishing growth in “copyright trolling.”32

2.    “Copyright trolling” is synonymous with MDJD lawsuits alleging copyright violation through BitTorrent file sharing.33

3.     The typical copyright trolling suit is filed against a large number of John Doe defendants.34

4.     These suits are not only the dominant form of copyright trolling, but also make up a substantial share of copyright litigation.35

5.     Joinder is important in the paradigmatic copyright trolling case, with the plaintiff relying importantly on being able to sue hundreds or thousands of defendants.36

6.     Copyright trolling is systematically related to pornography because the porn-related embarrassment factor enables quick and easy settlements.37

Based on both the tenor of Sag’s Article and his suggestion that existing statutory damage levels are unconstitutionally high in file sharing-based copyright cases, it is evident that Sag regards the developments he describes above as a bad thing.38 But Sag’s account does not match the reality of the litigation he summarizes. Certainly there are copyright cases in which each of the large-N, John Doe and porn characteristics that trouble Sag all play a role. Indeed, Sag cites a few of these cases. But Sag’s data do not actually show that such cases have undergone “astonishing growth.”39

In the rest of this Part, we make these points in two main subparts. Subpart A discusses the empirical facts concerning a single pornographic video producer and plaintiff, Malibu. These facts paint a picture that differs radically from the one that Sag describes. Subpart B illustrates how Sag’s focus on the number of defendants sued overstates the role of a few outlier cases, which obscures the fact that most cases he seems to consider copyright trolling necessarily involve not “hundreds or thousands of defendants,” but rather just a few.40 Whatever problems these cases might entail, they aren’t the ones that Sag identifies in his Article.

A.     Malibu

Consider the role of Malibu in Sag’s data. Malibu is a producer of pornographic videos, to which it sells access through the website http:/‌/‌www.x-art.com.41 As of July 11, 2015, this website indicates such access may be purchased for $29.95 for a single 30-day period, for $99.95 for one year, or for $19.95 per month with automatic repeat billing.42 According to court documents, Malibu had 50,000 subscribers as of June 2013.43 Thus, there is little question that Malibu is engaged in the production of copyrighted material for sale.

Since 2012, Malibu has also been engaged in John Doe litigation to enforce its copyrights. So heavily engaged, in fact, that Malibu dominates the “John Doe–Porn” category in the spreadsheet that Sag has posted on his website.44 In Sag’s spreadsheet of copyright suits filed during the 2000–2014 period, 3544 lawsuits are assigned to this category.45 Of these, 2144, or 66%, were filed by Malibu.46 Consequently, Malibu is far and away the most prolific filer of suits Sag categorizes as “John Doe–Porn” suits.47 Unfortunately, when he discusses or presents data related to Malibu, Sag does so in a way that either underplays Malibu’s importance (e.g., in his Table 148) or limits it to a table in an appendix that some readers might miss.49

To appreciate Malibu’s numerical importance, consider Figure 1, infra, which we constructed using only the data in Sag’s spreadsheet. The dashed line in the figure plots the yearly numbers of “John Doe–Porn” lawsuits filed by all plaintiffs other than Malibu. The solid line plots the yearly numbers of such suits filed by Malibu.

 

Figure 1. Number of Cases Sag Categorizes as “John Doe–Porn” by Filer

Table Copyright

 

We learn four important things from Figure 1. First, as Sag discusses, there were virtually no “John Doe–Porn” lawsuits filed in the years before 2010. Second, as Sag also discusses, the number of such lawsuits grew rapidly after 2010. But third, the trend in “John Doe–Porn” lawsuits filed by plaintiffs other than Malibu quickly reversed itself, sinking like a stone following 2012’s peak. Sag’s spreadsheet shows that there were 693 “John Doe–Porn” lawsuits filed by non-Malibu plaintiffs in 2012, 253 filed in 2013, and only 15 filed in the first six months of 2014.

Fourth, having filed no suits at all in 2011 or earlier, Malibu went bananas thereafter. Malibu filed 318 “John Doe–Porn” lawsuits in 2012, filed a staggering 1014 suits in 2013, and then actually increased its pace by filing 812 in just the first six months of 2014. To appreciate Malibu’s numerical importance, consider that Malibu filed more “John Doe–Porn” lawsuits in the first half of 2014 than all other filers filed in the peak year for non-Malibu filers. Figure 1 makes clear that what’s going on in “John Doe–Porn” lawsuits is what’s going on with Malibu. Sag’s picture of “astonishing growth” of troubling John Doe lawsuits involving both porn and a “large-N” number of defendants either fits litigation involving Malibu or it doesn’t fit at all.

Unfortunately, it’s the latter. Sag’s spreadsheet indicates that Malibu has never sued more than 81 defendants in a single case. Data in his Table 1 imply that in the cases represented in that table, Malibu filed 1709 lawsuits against 6280 defendants.50 That’s a lot of people to sue, but it’s still fewer than four defendants per lawsuit. And Sag acknowledges that “Malibu Media, LLC only filed suits against single IP addresses in the first six months of 2014.”51 A paragraph later, he tells readers that “Malibu Media, now the most prolific litigator in this area, has adapted its strategy and no longer relies on joinder at all.”52

Indeed, in between the two quotes above, Sag tells us that “although the average number of defendants per suit has decreased precipitously since 2010, there has been a steady increase in filings.”53 Sag is speaking here of all John Doe suits, rather than just “John Doe–Porn” lawsuits, but we suspect that the facts underlying his observations are driven by the huge number of suits filed by Malibu, the fact that Malibu has sued few defendants per suit on average, and the virtual disappearance of “John Doe–Porn” lawsuits filed by everyone other than Malibu. While Sag does refer to these facts in an offhand manner, they are fatal to his story of a burgeoning problem of lawsuits involving (1) a large number of; (2) unnamed defendants; and (3) porn.

But that’s not all there is to say about Malibu. By any standard—whether status-based or conduct-based—Malibu simply doesn’t look like a copyright troll. Judge Michael Baylson of the Eastern District of Pennsylvania wrote exactly this in a memorandum summarizing the evidence presented at a bellwether trial he held in 2013.54 We will quote from this memorandum at length, to illustrate the extent to which illegal file sharing has affected Malibu:

For the first couple of years of its existence, Malibu had a production budget of between $150,000 and $200,000 a year. It now spends over $2,000,000 a year to produce its movies. Its subscriber base has grown from about 500 in year one to approximately 50,000 now.

      Malibu spends substantial amounts of money to operate its business. Malibu’s expenses include compensation to actors, payments for servers and website maintenance, and purchases of bandwidth, among other things. Malibu’s subscription base has not increased over the last several years because people are downloading its movies from [sic] free via the BitTorrent Protocol.

      In May 2013, Malibu estimated over 80,000 people illegally downloaded Malibu’s movies in the United States through BitTorrent and over 300,000 people illegally downloaded Malibu’s movies in the fifteen countries that IPP, Ltd., Malibu’s investigator, tracks. This represented a typical month. Malibu subscribers routinely ask Malibu why they should pay a subscription fee when they can get its movies for free through BitTorrent. . . .

      Twice in 2013, unknown third parties hacked into Malibu’s servers and put its movies onto BitTorrent prior to the time that these movies were released onto Malibu Media’s website. These incidents cost Malibu thousands of dollars in lost subscription revenue.55

Judge Baylson goes on to state that Malibu’s additional security measures, involving file encryption, cost Malibu $15,000 per month; Judge Baylson notes that infringement is still possible once the files have been decrypted.56 Further, Judge Baylson reports that “Malibu subscribers have also complained that they can download its movies faster from BitTorrent than they can from its website.”57 To respond to these complaints, “Malibu started spending approximately $20,000 more a month” to increase download speed above that available through BitTorrent.58 Just these two examples of BitTorrent-related costs amount to $420,000 on an annual basis. For these and additional reasons, Judge Baylson concluded that “[t]he evidence that Malibu presented at trial was persuasive as to the fact that it had suffered real damages as a result of illegal downloading of its movies through BitTorrent.”59

Ultimately, the bellwether trial over which Judge Baylson presided involved two John Doe defendants who conceded liability but contested damages and a third defendant, Bryan White, who contested both liability and damages.60 The Doe defendants settled with Malibu before Judge Baylson was to determine damages. As for Mr. White, he first “perjured himself at the Rule 104 [evidentiary] hearing” concerning Malibu’s allegation that Mr. White had tampered with his hard drive to hide the evidence of his illegal file sharing and then admitted both to the perjury and the illegal file downloads.61 It is worth quoting Judge Baylson’s description of his rationale for setting the terms of his judgment against Mr. White:

      As many judges in copyright cases automatically begin with the minimum amount of statutory damages, $750 per infringement, and then treble that amount to note that the defendant committed a serious tort, I followed this practice.

      However, in this case, because of Bryan White’s having perjured himself at the Rule 104 hearing and having taken steps to destroy and conceal evidence, a further substantial award of damages was necessary. . . . Bryan White’s wiping clean of his computer in attempting to cover up the fact that he had downloaded the BitTorrent software, as well as five of Malibu’s movies, required a substantial penalty, and also to make a statement that would effectively deter others from acting as Bryan White had acted in this case.

      I have previously noted the fact that the infringement cases brought by Malibu and other similarly situated companies have attracted a great deal of attention, not only of federal district judges but also of bloggers who assert that “copyright trolls” are using unscrupulous tactics and false accusations to collect millions of dollars from innocent and injured computer users. I do not have any opinion as to the truth of the situation as regards to other plaintiffs in other courts. But here, Malibu has satisfied its burden of proof with substantial evidence and deserves a large award. Bryan White, because of his conduct, deserves a heavy dose of damages which should also act as a deterrent to others.62

Judge Baylson subsequently entered an order of judgment against Mr. White that required him to pay damages of $112,500 and attorney’s fees and costs of $128,350.50—for a total of over $240,000.63 It is worth noting that Malibu incurred substantial expert costs, and also that Judge Baylson appointed a special master under Rule 706, whose costs were presumably covered by the judgment against defendant White.64

The bellwether trial and Judge Baylson’s memorandum teach that illegal file sharing has posed a substantial challenge to Malibu’s conduct of its business as a producer and seller of copyrighted works through normal market channels. While Malibu has filed an enormous number of copyright infringement lawsuits, it also appears to have been the victim of an enormous amount of copyright infringement. Further, it is clear from Judge Baylson’s memorandum that Malibu’s litigation tactics, while possibly aggressive, were hardly out of bounds.

In sum, there seems no reason to view Malibu as either a troll or an abuser of the civil justice system based on the empirical facts that Sag provides. We do not make this point out of any great respect for Malibu or the content of its copyrighted works. Rather, we make the point because it undermines altogether Sag’s suggestion that the increase in John Doe copyright suits, especially those involving porn, constitutes copyright trolling that must be stopped via constitutional or statutory change.65 Malibu’s litigation activity is the 800-pound gorilla in the copyright room, and Malibu does not appear to be a copyright troll under any useful definition of the term. Malibu’s litigation behavior in other contexts may have raised eyebrows,66 but such behavior bears no connection to its classification as a troll in Sag’s definition.

B.     How Outliers Distort the Picture: The Importance of the Unit of Analysis

In light of the persuasive case that Sag seems to make in his Article, how can it be that the trend is so strongly away from large-N John Doe lawsuits? How can Sag’s data contradict his suggestion that growth in copyright lawsuits involving many unnamed defendants is substantial and problematic, when there are so many infamous lawsuits involving gobs of unnamed defendants?

The answer involves the important role that outlier cases play. Below is a reproduction of Sag’s Table 1.67 As the title suggests, Sag reports what he terms “top” John Doe plaintiffs in descending order of the number of defendants sued. Thus, an individual named Ott, who sued 15,551 defendants, is listed first, while Malibu appears seventh in this list despite the fact that the table reports that Malibu filed 1709 times as many suits as Ott.68

 

Table 1.     Top 20 Copyright John Doe Plaintiffs 2001–2014 by Total Number of Defendants

(Sag’s Table 1)

Table3

 

It is not clear that it makes sense to look at the data this way. Yes, Ott’s lawsuit against more than 15,000 defendants necessarily involved massive joinder. But Sag’s article never tells us who Ott is, or whether the subject of the lone lawsuit Ott filed was pornography. Having looked up the complaint using the link that Sag provided in his spreadsheet, we determined that Ott v. Does 115,551 does appear to involve pornography (we base this assessment on the titles of the video works alleged to have been copyrighted and downloaded via BitTorrent).69 Ott voluntarily dismissed the suit in June 2011, before the court decided whether to grant his motion to discover the names of Internet subscribers whose IP addresses he alleged were involved in unlawfully downloading his videos.70 While the dismissal was without prejudice,71 we can find no evidence of any other copyright suit filed by Mr. Ott in any U.S. district court. And no matter how important Ott’s suit is in terms of the total number of defendants sued, Mr. Ott’s litigation is just a drop in the sea of copyright suits, accounting for just one of over a hundred thousand suits referred to in Sag’s Table 1.

If what we are trying to understand is what Sag describes as “the astonishing rise of MDJD lawsuits [in United States district courts] over the past decade,”72 then the appropriate unit of analysis is not the number of defendants each plaintiff sued, but rather the number of lawsuits each plaintiff filed. And when we reconsider the data in Sag’s Table 1 in service of the latter approach, the results are problematic for Sag’s story.

Our own Table 2, infra, reports the data from Sag’s Table 1 with two important differences. The first difference is that we have sorted the plaintiffs reported in Sag’s table not according to the total number of defendants sued, but rather in terms of the total number of lawsuits filed. As we have seen, Malibu was by far the most prolific filer of “John Doe–Porn” lawsuits, and it is also the leading filer in Sag’s top 20 when we rank by lawsuits filed; Malibu’s filings account for a whopping 72% of the lawsuits represented in Sag’s Table 1.

 

Table 2.     Top 20 Copyright John Doe Plaintiffs 2001–2014 by Number of Suits Filed

(Data taken from and calculated based on data from Sag’s Table 1) 

                Table2

 

How is it possible that the plaintiff that sued only the seventh largest number of defendants filed nearly three-fourths of all the lawsuits filed by the “top 20” plaintiffs represented in Sag’s table? It is possible because—as we have seen—when Malibu sues, it typically sues few defendants.

That brings us to the second difference between Table 2 and Sag’s Table 1: We report information not just on the number of suits filed or the total number of defendants sued, but also on the average number of defendants each plaintiff sued per suit filed. As the fourth column in Table 2 shows, Malibu’s suits proceeded against an average of just four defendants each. Looking down the fourth column of Table 2, we see that the average number of defendants per suit exceeded 1000 for only four of the “top 20” plaintiffs listed in Sag’s table. And those four defendants filed only seven suits all together.

These figures suggest that Sag’s repeated reference to real-world and hypothetical examples in which thousands of defendants are joined in a single lawsuit has little to do with trends in the filing of MDJD lawsuits. To recap our two points: The plaintiffs that sued many defendants filed few lawsuits, and the single plaintiff whose filings accounted for nearly three-fourths of all suits represented in Sag’s Table 1 sued an average of only four defendants per suit.

Still, one might argue, the fourth column in Table 2 shows that a majority of the top 20 plaintiffs listed in Sag’s Table 1 did sue an average of far more than 100 defendants per suit. Sag’s focus on lawsuits filed against thousands of defendants might be off by an order of magnitude. But hundreds of defendants in a case is still a lot of defendants. But that argument wouldn’t fare well, because it appears that for many plaintiffs the largest single suit filed was far larger than their typical suits.

In the fifth column in Table 2, we report the average number of defendants sued in suits other than a plaintiff’s largest single suit. Only four defendants had an average of more than 100 defendants per suit in their non-largest suits. Even counting their largest single suits, these four other defendants filed just 50 lawsuits in total.73

In sum, a quick re-sorting of Sag’s Table 1 data, together with some simple calculations based only on those data, yield a picture of copyright John Doe suits among Sag’s top 20 plaintiffs that differs radically from Sag’s story. Yes, there are a smattering of plaintiffs who have sued bazillions of defendants. But Sag’s story is not intended to explain the increase in the number of defendants who are sued in MDJD actions. Rather:

This empirical study of copyright trolling presents new data that reveals the astonishing growth of a particular kind of copyright trolling—the [MDJD] lawsuit that alleges copyright violation through the file sharing software known as BitTorrent. Generally, these suits take the form of “Copyright Owner v. John Does 1–N” where N is a large number.74

Again, though, Sag’s data shows that even among the 20 plaintiffs that sued the largest total number of defendants between 2001 and 2014, nearly three-fourths of lawsuits were filed by plaintiff Malibu, which by 2014 had—as a matter of policy—eschewed using multi-defendant actions. Further, even among the other “top plaintiffs,” Table 2 shows that there were likely precious few lawsuits in which the level of joinder could fairly be described as anything approaching “massive.”

Consider that there were a total of 106,379 defendants sued by the “top plaintiffs” represented in Sag’s Table 1.75 This means that in the 14-year period represented by this table, these “top plaintiffs” could not possibly have filed more than 20 lawsuits with Sag’s exemplifying number of 5000 defendants.76Indeed, no more than 104 suits with 1000 or more defendants could have been filed by the “top plaintiffs” in these 14 years; nor could more than 208 suits with 500 or more defendants have been filed.

In conclusion, there may have been massive numbers of defendants sued in large-N MDJD suits. And there may have been an increase in the number of John Doe copyright suits over the period Sag considers. But the data in Sag’s Table 1 provide no support at all for the idea that an increase in John Doe suits filed against huge numbers of defendants caused the increase in John Doe suits.

C.     A Final Assessment of the Empirical Evidence for Sag’s Empirical Claims

We conclude this Part by assessing Sag’s six key claims in light of the empirical evidence we have discussed here:

1. Claim: There has been astonishing growth in “copyright trolling.”

Empirical Reality: It is true that there has been astonishing growth in the filing of “John Doe–Porn” lawsuits.

2. Claim: “Copyright trolling” is synonymous with MDJD lawsuits alleging copyright violation through BitTorrent file sharing.

Empirical Reality: Malibu, which filed the large majority of “John Doe–Porn” lawsuits,77 and which has dominated that field since 2013, is clearly not a troll even if its litigation behavior may have been lightly sanctioned on occasion.7

3. Claim: The typical copyright trolling suit is filed against a large number of John Doe defendants.

Empirical Reality: At least for 2013 and the first half of 2014, the typical “John Doe–Porn” lawsuit was filed by Malibu, whose lawsuits no longer involve more than one defendant, and, over the company’s copyright litigation history, have averaged fewer than four defendants.

4. Claim: These suits are not only the dominant form of copyright trolling, but also make up a substantial share of copyright litigation.

Empirical Reality: Malibu now dominates “John Doe–Porn” lawsuits, which themselves are a substantial share of copyright litigation.79 Further, based on the data, it is implausible that John Doe lawsuits filed against huge numbers of defendants make up a substantial share of copyright litigation measured in terms of the number of lawsuits.

 5. Claim: Joinder is important in the paradigmatic copyright trolling case, with the plaintiff relying importantly on being able to sue hundreds or thousands of defendants.

Empirical Reality: There simply could not have been very many lawsuits filed against “hundreds or thousands of defendants”;80 if the paradigmatic copyright trolling case requires such large numbers, the paradigm is the rare unusual one.

6. Claim: Copyright trolling is systematically related to pornography, because the porn-related embarrassment factor enables quick and easy settlements

Empirical Reality: While the number of porn-related copyright lawsuits has grown enormously, Sag’s data indicates that two-thirds of them were filed by a firm, Malibu, which, as noted previously, is not a troll.81

IV.   What Explains the Explosion in “John Doe–Porn” Lawsuits, and Does It Matter?

Sag’s explanation for the rise in the number of multi-defendant John Doe suits—which we have already questioned—centers on a simple story of economies of scale. He posits that copyright lawsuits involve both fixed and variable costs.82 Fixed costs are those that would have to be spent even if there were only a single defendant; examples include the cost of researching the facts and the law and writing a legally sufficient complaint. Variable costs are those that rise with the number of defendants. An example of variable costs, one on which both Sag and a number of courts have focused attention, is the filing fee for suing in federal court, which is a few hundred dollars.83

According to Sag’s analysis (and that of a number of district courts), it would not make sense to file MDJD copyright suits if plaintiffs had to pay the filing fee for all defendants sued.84 Permissive joinder rules let them sidestep this problem by aggregating claims against many defendants while having to pay only a single filing fee. Just as the class action device makes feasible actions against a single defendant via aggregation of large numbers of plaintiffs who each have small claims, permissive joinder in Sag’s story allows one copyright plaintiff to wage a single action against a large number of plaintiffs, even if none is likely to pay very much.85

But once again the facts related to Malibu undermine Sag’s argument. Malibu has filed most of the porn-related suits,86 has apparently never sued more than 81 defendants at a time, and, on average, sues only a few defendants.

So what explains the rise in John Doe/‌BitTorrent copyright cases, whether or not they involve porn or large numbers of defendants? The best explanation we can think of centers on the economies-of-scale/‌fixed-costs aspects of Sag’s story, but without his emphasis on permissive joinder. The IT and legal work needed to make out a complaint against one set of John Does identified only through their IP addresses lends itself easily to cutting and pasting information—about IP law and allegedly infringing downloads—into a boilerplate complaint. Similar complaints can then be filed against any desired set of John Doe defendants in any federal district court, with the same law firm or firms representing the plaintiff in all suits.

Our unscientific sampling of Malibu complaints suggested that this is an accurate characterization of that company’s litigation behavior—which means it is an accurate characterization of the majority of John Doe/‌porn-related suits. The post-filing litigation issues that arise—relating, for example, to pre-Rule 26(f) conference discovery of ISP information; motions to quash ISP subpoenas; questions related to the timeliness of service (or lack thereof); and voluntary dismissal pursuant to Rule 41(a)87—seem to be substantially the same in the Malibu cases whose dockets we have inspected. Thus, once the basic legal work is done, the marginal costs in each suit involve the following: whatever it costs to monitor BitTorrent downloads; the cutting, pasting, and filing of complaints; subpoena-related litigation; and settlement or other post-subpoena activity.

This litigation model, like the one that Sag offers, entails the spreading of fixed costs over many defendants. But contrary to Sag’s story, it seems that Malibu and its counsel have been able to realize economies of scale spread across multiple civil actions—rather than having to spread them across massive numbers of defendants within the same action. Perhaps this is not surprising, since the only obvious gain from joining many defendants in the same action is the court filing fee of a few hundred dollars.88 Sag reports that typical settlements are in the $2000–$4000 range,89 which suggests that forcing the plaintiff to pay a separate filing fee per defendant may be unlikely to deter plaintiffs like Malibu. If all of this is right, then joinder rules have little if anything to do with trends in John Doe file sharing suits.

In addition to his joinder-based theory, Sag does offer up a “second structural explanation for the prevalence of pornographic subject matter in copyright trolling” (permissive joinder rules was the first), connected to the fact that “a significant amount of this litigation” has been initiated by “a small number of entrepreneurial lawyers and plaintiffs.”90 Sag might have used this observation to pivot in the direction of our alternative economies-of-scale narrative, with which it is consistent. Instead, though, Sag emphasizes the behavior of the now-dissolved Prenda Law firm, which he reports engaged in “ruthless . . . pursui[t of] profits” and was not only “accused of colluding in litigation, lying to the court, forging documents, and identity theft,” but also actually sanctioned in court.91 It may be narratively useful to have a villain to point to, but the one Sag casts seems to be relatively unimportant to the broader rise-of-MDJD-suits story. Not to beat a dead horse, but according to Sag’s Table 2, Prenda has apparently not been associated with Malibu.92

If the John Doe copyright litigation model isn’t strongly driven by traditional bargaining-in-the-shadow-of-the-law considerations, then what makes it go? We agree with Sag that the porn-related embarrassment factor seems likely to play: surely there is some value to plaintiffs in bargaining in the law’s shadows, quite apart from whatever dangers the substantive law poses. But we suspect that the simple threat of litigation expense matters the most. Lawyers aren’t cheap, and proving one’s innocence could be expensive if a plaintiff fought a case even part way to judgment.

But this raises another problem of economic analysis: How credible is the threat that a plaintiff really will litigate? This isn’t the place to develop a full-fledged economic model of litigation that takes credibility constraints seriously, but we have our doubts as to whether plaintiffs can expect their bluffs to keep going uncalled once word gets out that they habitually fail to serve identified defendants. As Sag himself discusses, there is some evidence that plaintiffs regularly do fail to serve defendants.93 A simple judicial policy of proactively dismissing cases pursuant to Rule 4(m),94 as at least some judges seem to do, would reduce the credibility of plaintiffs that aren’t willing to litigate against defendants who don’t settle immediately. Indeed, in the actions that were consolidated into Judge Baylson’s bellwether trial, Malibu voluntarily dismissed claims against a number of defendants whom it was unable to serve before the Rule 4(m) clock had struck midnight.

IV.   Conclusion

Everyone today dislikes “copyright trolls.” Yet, no one knows just who a copyright troll is! Copyright trolling remains a confused, moving target. Behavior that is variously litigious, unprofessional, unethical, morally questionable, overzealous, or even just zealous is today commonly characterized as trolling. Unfortunately, Sag’s attempt to marry the conceptual study of copyright trolling and empirical data winds up contributing to this confusion. Sag’s conduct-based definition of trolling is at once both overdetermined and underspecified. In addition, while the data Sag presents does show that the frequency of John Doe porn lawsuits has indeed risen dramatically, this fact itself does not support Sag’s claim that there are alarming trends involving copyright trolling through multi-defendant John Doe lawsuits.

We share Sag’s concern that copyright litigation suffers from a variety of different malaises, many of which may be exacerbated by litigants. But we cannot agree with his reading of his data concerning systematic trends in copyright trolling.

  1. [1]. Matthew Sag, Copyright Trolling, An Empirical Study, 100 Iowa L. Rev. 1105 (2014).

  2. [2]. Id. at 1108.

  3. [3]. Matthew Sag, Data Sets—Copyright Trolling, Matthew Sag: Copyright L., Fair Use & Tech., http://matthewsag.com/wp-content/uploads/2012/08/Copyright-Filing-Data-to-June-30-2014.xls (downloaded on July 21, 2015).

  4. [4]. Sag, supra note 1, at 1126. There are several different types of cases that all are instances of Sag’s “John Doe pornography” category. See infra Part II.A.

  5. [5]. Sag, supra note 1, at 1110–14.

  6. [6]. Shyamkrishna Balganesh, The Uneasy Case Against Copyright Trolls, 86 S. Cal. L. Rev. 723, 732 (2013).

  7. [7]. Righthaven was an entity that entered into arrangements with content owners, agreeing to police the use of their works on the Internet and then commence actions for copyright infringement against them. See id. at 738–46 (providing a fuller discussion of the Righthaven episode).

  8. [8]. Sag, supra note 1, at 1112.

  9. [9]. Id. at 1113.

  10. [10]. Id.

  11. [11]. See id.

  12. [12]. Id.

  13. [13]. Id.

  14. [14]. Kenneth Ayotte et al., A Safety Valve Model of Equity as Anti-Opportunism 4 (Northwestern Law & Econ. Research Paper No. 13-15, 2013), http://papers.ssrn.com/abstract=2245098.

  15. [15]. Sag, supra note 1, at 1114.

  16. [16]. Id.

  17. [17]. Id.

  18. [18]. Balganesh, supra note 6, at 743–44.

  19. [19]. Shyamkrishna Balganesh, Copyright Infringement Markets, 113 Colum. L. Rev. 2277, 2286 (2013).

  20. [20]. See generally Gary A. Rosen, Unfair to Genius: The Strange and Litigious Career of Ira B. Arnstein (2012).

  21. [21]. Id. at 222.

  22. [22]. Or the opposite. Claims might be dismissed under Rule 12(b)(6) for failure to state a claim, due to the plaintiff’s inability to meet the plausibility pleading standard introduced by the Supreme Court’s Twombly and Iqbal cases, even when discovery would reveal facts that make the plaintiff’s merits case quite strong.

  23. [23]. Sag, supra note 1, at 1122–23.

  24. [24]. As Sag notes, district courts have been split on whether permissive joinder under Rule 20(a) is appropriate in such cases. See id. at 1123.

  25. [25]. See id. at 1136.

  26. [26]. Id. at 1114.

  27. [27]. For a general account, see Ben Depoorter et al., Copyright Backlash, 84 S. Cal. L. Rev. 1251, 1260–64 (2011).

  28. [28]. Sag, supra note 1, at 1114.

  29. [29]. See, e.g., Walter K. Olson, The Litigation Explosion: What Happened When America Unleashed the Lawsuit 2–4 (1991).

  30. [30]. See, e.g., Jonathan T. Molot, Litigation Finance: A Market Solution to a Procedural Problem, 99 Geo. L.J. 65 (2010); Maya Steinitz, Whose Claim Is This Anyway? Third-Party Litigation Funding, 95 Minn. L. Rev. 1268, 1275–85 (2011).

  31. [31]. Sag, supra note 1, at 1108, 1110 (footnotes omitted).

  32. [32]. Id. at 1108 (“This empirical study of copyright trolling presents new data that reveal[] the astonishing growth of a particular kind of copyright trolling....”).

  33. [33]. Id. (“[A] particular kind of copyright trolling—the multi-defendant John Doe (‘MDJD’) lawsuit... alleges copyright violation through the file sharing software known as BitTorrent.”).

  34. [34]. Id. (noting that cases which satisfy the “particular kind of copyright trolling” definition “[g]enerally... take the form of ‘Copyright Owner v. John Does 1–N’ where N is a large number”).

  35. [35]. Id. (“MDJD suits are not just any form of copyright trolling, they are the dominant form.”).

  36. [36]. Id. at 1108 (“The paradigmatic troll plays a numbers game in which it targets hundreds or thousands of defendants, seeking quick settlements priced just low enough that it is less expensive for the defendant to pay the troll rather than defend the claim.”); id. at 1109 (“The economic viability of MDJD litigation depends on suing as many defendants as possible in a single action to keep costs low and leveraging the threat of statutory damages in order to maximize the flow of settlement dollars.... [T]hese suits are a form of copyright trolling.”).

  37. [37]. Id. at 1108–9 (“Of the 3817 copyright law suits filed in 2013, over 43% were against John Does and more than three-quarters of those related to pornography.”); id. at 1110 (“[T]he association with pornography is far from coincidental. The MDJD model works especially well for pornography because the potential embarrassment of being accused (accurately or not) of downloading such material is a great motivation to settle.”).

  38. [38]. Id. at 1135–37.

  39. [39]. Id. at 1108.

  40. [40]. Id.

  41. [41]. Malibu Media, LLC v. John Does 1, 6, 13, 14, 950 F. Supp. 2d 779, 787 (E.D. Pa. 2013) (mem.).

  42. [42]. See Join, X-Art.com, http://www.x-art.com/join (last visited July 21, 2015).

  43. [43]. Malibu Media, LLC, 950 F. Supp. 2d at 787.

  44. [44]. This statement, and any other statement not directly attributed to a page in Sag’s article, is based on our calculations using the spreadsheet Sag posted on the Internet. See Sag, supra note 3.

  45. [45]. Id.

  46. [46]. There are an additional 184 suits filed by Malibu that are categorized as “Other Copyright.” Id. We suspect many if not all of these suits should have been categorized as “John Doe–Porn,” but we will ignore these cases for the rest of our discussion.

  47. [47]. Surprisingly, Sag hardly discusses Malibu’s role. Aside from citations to cases and entries in his Tables 1 and 2 (discussed infra Part III.B), he discusses Malibu on only one page of his article’s main text, and in a table in Appendix B that appears on his Article’s last page. Sag, supra note 1, at 1142, 1147 app.B.

  48. [48]. Id. at 1131 tbl.1.

  49. [49]. Id. at 1147 tbl. app.B.

  50. [50]. There are some inconsistencies between Sag’s spreadsheet and the data reported in his Article. Sag’s Table 1 table is titled “Top 20 Copyright John Doe Plaintiffs 2001–2014 by Total Number of Defendants.” Sag, supra note 1, at 1131 tbl.1. Yet Sag reports there that Malibu filed “only” 1709 lawsuits in this period, several hundred fewer than the 2144 categorized as “John Doe–Porn” in his spreadsheet. Id.; see also Sag, supra note 3. Further, Sag’s table in Appendix B lists a total of 2183 “Malibu v. Doe(s)” suits in years 2012–2014 (332 in 2012, 1034 in 2013, and 817 in 2014—each a bit higher than the numbers one observes in his spreadsheet). Sag, supra note 1, at 1147 tabl. app.B.

  51. [51]. Sag, supra note 1, at 1142.

  52. [52]. Id.

  53. [53]. Id.

  54. [54]. Malibu Media, LLC v. John Does 1, 6, 13, 14, 950 F. Supp. 2d 779, 780 (E.D. Pa. 2013) (mem.) (“I emphasize that Malibu is not what has been referred to in the media and legal publications, and in the internet blogosphere, as a ‘copyright troll’—i.e., a non-producer who merely has acquired the right to bring lawsuits against alleged infringers.”).

  55. [55]. Id. at 787.

  56. [56]. Id.

  57. [57]. Id.

  58. [58]. Id.

  59. [59]. Id. at 788.

  60. [60]. Id. at 785–87.

  61. [61]. Id. at 788.

  62. [62]. Id.

  63. [63]. Id.

  64. [64]. Id. at 786.

  65. [65]. See Brad A. Greenberg, Copyright Trolls and the Common Law, 100 Iowa L. Rev. Bull. 77, 85–86 (2015).

  66. [66]. See, e.g., Joe Mullin, Porn Troll Malibu Media Sanctioned for Trying to Embarrass Defendants, Ars Technica (Sept. 11, 2013, 2:33 PM), http://arstechnica.com/tech-policy/2013/09/11/
    porn-troll-malibu-media-sanctioned-for-trying-to-embarrass-defendants (describing one judge’s Rule 11 sanctions in the amount of $2200 imposed on Malibu for its practice of attaching an “Exhibit C” with its complaints, which lists salaciously titled videos, not copyrighted by Malibu, allegedly downloaded by John Doe defendants, as well as the judge’s lifting of a stay on the 11 cases involved).

  67. [67]. Sag, supra note 1, at 1131 tbl.1.

  68. [68]. See supra note 50 and accompanying text (explaining the discrepancy in this 1709 figure and the numbers of Malibu-filed suits reported in Sag’s Appendix B and in his spreadsheet).

  69. [69]. See Complaint for Copyright Infringement at 4, Ott v. Does 1–15,551, No. 1:11-cv-00553, 2011 WL 2491677 (D.D.C. Mar. 11, 2011).

  70. [70]. Notice of Voluntary Dismissal of Doe Defendants at 1, Ott v. Does 1–15,551, No. 1:11-cv-00553-RLW (D.D.C. June 20, 2011), https://www.bloomberglaw.com/s/dockets/8c748a428853b2b453
    bc056240c8dae0/document/2048032166569730?documentName=10.pdf&search32=C9NMUR35C5N5USR5C5P66Q2VEHIN4R9T7HO62SJKF4UIGJRKEGKJS81640U68RR3DDIN8NREELMM4PBI7KH32EHH64MM6THD60O3AD9J48V3MEREDTFMIRBGBTO6GSJ1EDIN6F9H&fmt=pdf.

  71. [71]. Id.

  72. [72]. Sag, supra note 1, at 1110.

  73. [73]. Further, Top 20 plaintiffs Mr. Ott and On the Cheap, LLC both drop out of this analysis, since each filed only one suit.

  74. [74]. Sag, supra note 1, at 1108 (emphasis added).

  75. [75]. Id. at 1131 tbl.1.

  76. [76]. Were there 21 such suits, there would have been only 1379 defendants left to fill out the remaining 2349 lawsuits. Also, Sag cites the example of Voltage Pictures, LLC v. Does 1–5000, No. 1:10-cv-00873, 2010 WL 4955131 (D.D.C. May 24, 2010). Sag, supra note 1, at 1115 n.40. He also uses the hypothetical example of 5000 defendants to illustrate his arguments in multiple places. See id. at 1115, 1125, 1129.

  77. [77]. Malibu also had 10%, 27%, and 41% of all copyright cases from 2012–2014. Sag, supra note 1, at 1147 tabl. app.B.

  78. [78]. See supra note 66 (referring to Rule 11 sanctions).

  79. [79]. See supra tbl.2 (discussing Malibu’s substantial share of all copyright cases).

  80. [80]. Sag, supra note 1, at 1108.

  81. [81]. See supra note 54 and accompanying text.

  82. [82]. Sag, supra note 1, at 1115.

  83. [83]. See, e.g., Fee Schedule, U.S. District Ct. S. District of Fla., https://www.flsd.uscourts.
    gov/?page_id=2396 (last visited July 21, 2015) (listing a “[f]iling fee for opening civil action” of $400).

  84. [84]. Sag, supra note 1, at 1141–44.

  85. [85]. The availability of statutory damages in the copyright context means that each claim is nominally not small; we return to this issue below.

  86. [86]. We saw above that Malibu filed 72% of the suits filed by Sag’s Top 20 plaintiffs. Using Sag’s spreadsheet of characteristics of 42,900 copyright cases, we found that Sag categorized 8.3% of cases as involving pornography, while Malibu Media, LLC, is the plaintiff in 5.4% of cases. Thus, Malibu was the plaintiff in 66% of all porn cases (5.4% ÷ 8.3% = 66%). We note that the number of cases in which Malibu was the listed plaintiff was 2328, much greater than the 1709 that Sag reports in his Table 1. Sag’s spreadsheet contains three categories for the nature of suit: “John Doe–Porn,” “John Doe–Other,” and “Other Copyright.” Sag, supra note 3. It appears that this last category includes many cases that were initially filed as what Sag would call “John Doe–Porn” and were then re-captioned after the plaintiff discovered a defendant’s name. For example, Malibu Media, LLC v. Moore, No. 1:13-cv-01535 (D. Colo. Jun 13, 2013) began as MALIBU MEDIA, LLC, Plaintiff, v. JOHN DOE subscriber assigned IP address 50.152.34.5. This case is listed in Sag’s spreadsheet under “Other Copyright,” though the plaintiff’s identity and the content of the complaint and attached exhibits make clear that the case involves pornographic videos; in Sag’s intended taxonomy, the case pretty clearly should have been categorized as “John Doe–Porn”. See Sag, supra note 3.

  87. [87]. Fed. R. Civ. P. 41(a).

  88. [88]. 28 U.S.C. § 1914(a) (2012). One might think there are other economies of scale to joining defendants in the same action, but the only connection between defendants is likely to be their involvement in the same “swarm,” so there is likely little gain in litigation efficiency from proceeding against them together. Sag agrees: “Litigating a case all the way to trial against thousands of individuals in the same suit would be ridiculous and unwieldy”; he then drops a footnote encouraging readers to “[j]ust imagine how long a scheduling conference would take.” Sag, supra note 1, at 1109 & n.16.

  89. [89]. Sag, supra note 1, at 1009–10.

  90. [90]. Id. at 1129–30.

  91. [91]. Id. at 1130–31 (footnotes omitted).

  92. [92]. Id. at 1132 tbl.2.

  93. [93]. Id. at 1126.

  94. [94]. Fed. R. Civ. P. 4(m).

*

Professor of Law, University of Pennsylvania Law School.


**

Associate Professor of Law, University of Pennsylvania Law School.