Thomson Reuters Enterprise Centre GmbH v. ROSS Intelligence Inc.

Opinion

IN THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF DELAWARE THOMSON REUTERS ENTERPRISE CENTRE GMBH and WEST PUBLISH- ING CORP., Plaintiffs, No. 1:20-cv-613-SB v. ROSS INTELLIGENCE INC., Defendant. Jack B. Blumenfeld, Michael J. Flynn, MORRIS, NICHOLS, ARSHT & TUNNELL LLP, Wilmington, Delaware; Dale M. Cendali, Eric A. Loverro, Joshua L. Simmons, KIRKLAND & ELLIS LLP, New York, New York. Counsel for Plaintiffs. David E. Moore, Andrew L. Brown, Bindu A. Palapura, POTTER ANDERSON & COR- ROON LLP, Wilmington, Delaware; Warrington Parker, Christopher J. Banks, Joa- chim B. Steinberg, Jacob Canter, Anna Z. Saber, CROWELL & MORING LLP, San Francisco, California; Emily T. Kuwahara, Jordan Ludwig, CROWELL & MORING LLP, Los Angeles, California; Mark A. Klapow, Keith J. Harrison, Lisa Kimmel, Crinesha B. Berry, CROWELL & MORING LLP, Washington, D.C.; Ryan H. Seewald, CROWELL & MORING LLP, Denver, Colorado. Counsel for Defendant. MEMORANDUM OPINION September 27, 2024 BIBAS, Circuit Judge, sitting by designation. To make it to trial, litigants must substantiate their claims. Ross contends that Thomson Reuters violated and still violates antitrust laws, but Ross has not backed

up its allegations with enough evidence. So I grant summary judgment to Thomson Reuters on those counterclaims. I. BACKGROUND The antitrust issues here are just one part of a larger case. Thomson Reuters’s Westlaw platform has an extensive collection of legal sources (including judicial opin- ions) and accompanying search tools to navigate them. D.I. 170 at 1–2. Ross tried to build a better legal search tool using artificial intelligence. Id. Thomson Reuters sued

Ross for copyright infringement, alleging that Ross had used Thomson Reuters’s intel- lectual property to build its search tool. Compl., D.I. 1 at 2–3. The copyright lawsuit is pending before me. D.I. 667. Ross then brought these antitrust counterclaims. D.I. 24 at 42–44. This Court dis- missed some of these claims, D.I. 170, but three counts survived to discovery on a theory that Westlaw ties its caselaw database to its search tools. Those counts now

return on summary judgment: Count VI, 15 U.S.C. § 2 (Sherman Act), Count VII, 15 U.S.C. § 1 (Sherman Act), and Count VIII, California’s Business and Professions Code § 17200 et seq. D.I. 24 at 42–44; D.I. 170 at 9, 12, 15; D.I. 519, 521, 523, 525, 527. The parties do not distinguish among the three counts in their summary-judgment brief- ing because all rise and fall on the same issues. In this opinion, I resolve five summary-judgment motions and two motions to pre- clude or exclude expert opinions and testimony relevant only to the antitrust coun- terclaims. Thomson Reuters moved for summary judgment on five issues: separate

products, market definition and power, statute of limitations, injunctive relief, and damages relating to testimony from Dr. Alan Cox. D.I. 519, 521, 523, 525, 527. Thom- son Reuters also moved to preclude certain opinions of Ross’s experts Dr. James Rat- liff and Dr. Gillian Hadfield. D.I. 529. And Ross moved to exclude certain opinions of Thomson Reuters’s expert Dr. Chad Syverson. D.I. 516. I will grant summary judgment if there is “no genuine dispute as to any material fact.” Fed. R. Civ. P. 56(a). I view facts in the light most favorable to Ross, the non-

movant. Lamont v. New Jersey, 637 F.3d 177, 179 n.1 (3d Cir. 2011). I must enter summary judgment “against a party who fails to make a showing sufficient to estab- lish the existence of an element essential to that party’s case.” Celotex Corp. v. Catrett, 477 U.S. 317, 322 (1986). II. ROSS HAS FAILED TO SHOW THAT THOMSON REUTERS IS TYING SEPARATE PRODUCTS AND HAS FAILED TO DEFINE THE MARKETS FOR THEM Ross’s remaining antitrust counterclaims are all based on tying theories. Some tying arrangements violate antitrust law. Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 461–62, 464 (1992). Tying occurs when a party will “sell one prod- uct but only on the condition that the buyer also purchases a different (or tied) prod-

uct.” N. Pac. Ry. Co. v. United States, 356 U.S. 1, 5 (1958). Ross’s theory is that Westlaw’s caselaw database (“the public law database”) is a product that many con- sumers want (the tying product), so Thomson Reuters decided to sell it only packaged with the Westlaw search tools (the tied product) to ensure that those products also sell. D.I. 24 at 31–32. In other words, Ross claims that Thomson Reuters forces people to buy its Westlaw search tools if they want to use its caselaw database. Ross contends

that this arrangement is both per se illegal under the Sherman Act and illegal under a rule-of-reason theory of liability. D.I. 554 at 7, 23; see Town Sound & Custom Tops, Inc. v. Chrysler Motors Corp., 959 F.2d 468, 475, 481–82 (3d Cir. 1992). Under either theory, Ross must make two threshold showings before it may pro- ceed with its arguments that Thomson Reuters violated antitrust law. First, Ross must show that the two products are separate. Avaya Inc., RP v. Telecom Labs, Inc., 838 F.3d 354, 397 (3d Cir. 2016). Products that are not separate cannot be tied together.

Second, Ross must define the relevant market before it can hope to show an improper use of power in that market. The relevant markets are different for per se and rule- of-reason claims. For a per se tying claim, Ross must define the tying product market. Brokerage Concepts, Inc. v. U.S. Healthcare, Inc., 140 F.3d 494, 513 (3d Cir. 1998). And for Ross’s rule-of-reason arguments, it must define the tied product market. Id. at 519. Ross does neither.

A. Ross has not shown that the two products are separate (i) Legal principles To show that a seller is illegally tying two products, one must first show that the two products are separate. Avaya, 838 F.3d at 397. Whether products are separate is typically an issue of law to be resolved by a judge. Phillip E. Areeda & Herbert J. Hovenkamp, Antitrust Law (Lexis Online 2024) ¶ 1743c. The test for whether the two products are distinct is whether there is “sufficient consumer demand so that it is efficient for a firm to [sell them separately].” Eastman Kodak, 504 U.S. at 462. “[N]o tying arrangement can exist unless there is a sufficient demand for the purchase of”

the products separately. Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 21 (1984). The sufficient-demand requirement has teeth. “Idiosyncratic buyers” are not enough; we must look at what consumers generally want. Areeda & Hovenkamp ¶ 1744d. And, when looking at analogous markets, “if less than 20 percent of the tying item is sold unbundled in the competitive market analogue, then the items are a sin- gle product.” Id. The parties dispute what Ross must do to show sufficient consumer demand. Ross

argues that it need not prove that people actually buy the products separately, only that independent markets could exist. D.I. 555 at 5. Ross points out the danger of “perversely immuniz[ing] the worst-case scenario of a successful tie by which a monop- olist successfully leverages a monopoly in the tying product into a monopoly in the tied product.” Id. at 8 (quoting Areeda & Hovenkamp ¶ 1745d). To avoid this unfair outcome, Ross says, I must “assess[] … what consumer demand would be at the costs

and quality that would prevail if the items were separately provided.” Id. at 9 (citing Areeda & Hovenkamp ¶ 1745d). To support its view, Ross also cites the ABA Model Jury Instructions in Civil Antitrust Cases on separate and distinct products. Id. at 9. But the Model Jury Instructions are not binding on courts. And though they do in- struct the jury to “consider whether there would be separate demand for each [prod- uct] if they were offered separately,” they further explain that “[t]he relevant issue is whether there is sufficient demand from customers to induce sellers to provide them separately.” D.I. 558-1, Ex. B at 26. So they do not make Ross’s point. Unsurprisingly, Thomson Reuters seeks a more concrete standard than Ross. It

argues that Ross must show that consumers either “bought the items separately” or “requested or wanted the items separated.” D.I. 583 at 5 (quoting Areeda & Hovenkamp ¶ 1743a). Thomson Reuters is mostly right. I conclude that Ross must show at least one of three things to meet its burden on separate products: (1) that consumers “bought the items separately,” (2) that consum- ers “requested or wanted the items separated,” or (3) “that some customers would have requested or wanted the items separated but for” the fact that they were “intim-

idated not to make such requests or were deprived of the information that would have made them want the items separated.” Areeda & Hovenkamp ¶ 1743a. So Ross must show (1), (2), or (3) to establish that “independent markets could exist.” Areeda & Hovenkamp ¶1745d. But it has not. (ii) Ross fails to show separate products The facts, taken in the light most favorable to Ross, do not make any of these three

showings. Ross submits the following facts and arguments to attempt to show sepa- rate products. 1. Demand for public law separately Books. Ross notes that Thomson Reuters used to sell its public-law database in books (rather than online) without the legal search tools that are now packaged with Westlaw. Id. at 12–13. True, historical market analogues can support a separate- products finding. See Areeda & Hovenkamp ¶ 1744c2; see also Kaufman v. Time Warner, 836 F.3d 137, 142 (2d Cir. 2016) (“Relevant evidence of separate and distinct consumer demand for the tying product and the tied product is, inter alia, the history

of the products being, or not being, sold separately … or the sale of products sepa- rately in similar markets.”). And this is one reason why the previous judge in this case, Judge Stark, did not dismiss the tying claim. As he noted, “Plaintiffs’ public law database was effectively sold as a stand-alone product in the form of books for decades before modern technology allowed plaintiffs to develop an online legal search tool.” D.I. 170 at 8. But this is just one possible data point, and the bar is higher at summary judgment.

The idea behind historical market analogues is that they are evidence of consumer preferences from a time when the markets were competitive. See Areeda & Hovenkamp ¶ 1744c2. But the analogy to books suffers from two flaws. First, Ross is wrong that books were sold without search tools. True, books were sold without Westlaw’s current technological capacity. But if we can analogize online legal data- bases to printed legal databases, we can also analogize online search tools to printed

search tools: tables of contents, indices, and page numbers. So its database was not sold unbundled from search tools. Second, “consumer preferences, production costs, and quality concerns … vary over time.” Id. The evolution from book search tools (say, a table of contents) to Westlaw’s digital search tools (say, Boolean search terms) is like how the horse-drawn carriage market evolved into the car market. Just as we no longer use horse-drawn carriages for transportation (except for fun), few consumers want caselaw separated from the sophisticated search tools that make it digestible. A market for public law in book form used to exist, but that does not mean that a market for separate caselaw

still exists in a world with more sophisticated search tools. Ross has not done enough to show that this market is a helpful analogue. Consumer requests for public-law databases. Ross says that consumers asked to buy Westlaw’s public-law database separately. D.I. 555 at 15. But that smidge of consumer demand is not enough. There must be “sufficient consumer demand so that it is efficient for a firm to” sell the two products separately. Eastman Kodak, 504 U.S. at 462. Ross cites emails from two law firms, Baker Hostetler and

Skadden. D.I. 558-1, Ex. Q at 280–82; D.I. 558-1, Ex. R at 283–84. But the Baker Hostetler email chain does not ask about buying public law separately. D.I. 558-1, Ex. Q at 280–82. And the Skadden email alone is not enough to infer “sufficient con- sumer demand.” Eastman Kodak, 504 U.S. at 462; D.I. 558-1, Ex. R at 283–84. Ross could have done more to try to develop the record during discovery, perhaps subpoe- naing Skadden representatives. But it did not.

APIs. Ross also argues that Thomson Reuters’s decision to make some Westlaw content available via application programming interfaces shows separate consumer demand for public law. D.I. 555 at 18–20. This also fails. These interfaces are “another method that customers use to access Thomson Reuters’ platforms.” D.I. 520 at 23. Some of them let customers search internal firm data and external Westlaw data through a single search. D.I. 555 at 19. But Ross admits that these are just interfaces; they do not allow “independent access to [Thomson Reuters’s] full U.S. public law database.” Id. So this is not an example of selling public law separately. Ross submits evidence that Thomson Reuters is considering offering data products that include the

public law. D.I. 558-2, Ex. EE at 83. But Thomson Reuters is not doing this yet, so it is unclear what, if any, demand there would be for this data. 2. Demand for search tools separately Consumer requests for Ross’s search tool. As evidence that consumers wanted to buy its search tool on its own, Ross points to an interrogatory response and depositions of its own employees. D.I. 558-1, Ex. S at 291; Ex. T at 309–11 (transcript 100:24–102:15); Ex. U at 323–37 (transcript 14:18–18:2); Ex. V. at 342–43 (transcript

155:5–156:22) (irrelevant because it discusses buying Ross as the company, not Ross’s products); and Ex. W at 360–6 (transcript 269:23–271:11). But careful review of these transcripts shows that, at best, most employees gestured toward unspecified “custom- ers” who wanted to buy Ross’s search product without the public law. See, e.g., D.I. 558-1, Ex. T. at 310–11 (transcript 101–02) (a former Ross employee acknowledging that he never actually sold the search tool separately and responding to a question

about who wanted to buy Ross’s search tool separately from the public law database with, “[t]here was a customer that came into the chat. There were customers that would come into the chat.”). These statements do not suggest many interested buyers nor allege a specific number of consumers, and the only one they name specifically is the U.S. Department of Justice. So even taking these statements as true, they do not show “sufficient con- sumer demand so that it is efficient for a firm to” sell the two products separately. Eastman Kodak, 504 U.S. at 462. And Ross’s argument that its “legal search tool re- ceived significant attention and praise” is irrelevant. D.I. 555 at 13. Westlaw’s product tiers. Ross also tries to find evidence of separate products in

Westlaw’s own sales. Ross argues that Westlaw’s choice to offer different levels of search plans shows demand for buying the public law database and accompanying search tools separately. Id. at 17. But it does not matter that Westlaw offers three plans, each bundling its public-law collection with a different search tool. This shows demand for different levels of search-tool sophistication, not demand for a public-law database with no search tool at all or for a search tool on its own. West KM. Ross’s last point is that Thomson Reuters sells West KM separately,

but that argument fails too. Id. at 21. As Ross itself explains, “West KM uses Westlaw’s search technology to review the [customer] organization’s own documents, promising an experience that mimics searching for public law on Westlaw.” Id. So West KM is not a search tool for public law that is being sold separately from public law itself. To sum up, Ross has failed to satisfy any of the three standards for showing dis-

tinct products. It has not shown any consumers “bought the items separately.” Areeda & Hovenkamp ¶ 1743a. Nor has Ross adduced enough evidence that consumers “requested or wanted the items separate.” Id. And Ross has made no effort to show that consumers would have asked for the items separated if they were not “intimi- dated not to make such requests or were deprived of the information that would have made them want the items separated.” Id. I thus grant Plaintiffs’ Motion for Summary Judgment on Ross’s Antitrust Coun- terclaims (No. 1) Separate Products [D.I. 519]. Without showing separate products, none of Ross’s remaining antitrust counterclaims can proceed.

B. Ross has not defined the markets Even if Ross’s allegation of tying had not failed on separate products, it would still fail because Ross has not properly defined the market in which Thomson Reuters allegedly exercises excessive power, either on a per se tying theory or on a rule-of- reason theory.

(i) Legal principles At a minimum, Ross must define either the tying market or the tied market. For per se tying, Ross must show two threshold things about the market. First, it must “properly define[]” the tying product market. Brokerage Concepts, 140 F.3d at 513. That means defining both the product and the geography of the market. Id. Second, Ross must show that Thomson Reuters “possesses market power in the tying product market.” Town Sound, 959 F.2d at 477. One cannot determine market power until

after one defines the market. So a successful per se claim defines the tying market. For a rule-of-reason tying theory, Ross need define only the market for the tied product. A rule-of-reason theory is more flexible than a per se theory, letting Ross “state some plausible theory of antitrust harm that does not depend implicitly on power leveraging.” Town Sound, 959 F.2d at 486. To support a rule-of-reason theory, Ross must be able to show that the alleged tie affected “competition in the tied mar-

ket.” Brokerage Concepts, 140 F.3d at 511. So under this theory of liability, Ross need not define the tying market. But it does need to define the tied market. “Before we can determine whether there was harm to competition in the tied market, that mar- ket must be defined.” Id. at 519.

This requirement holds especially true for Ross’s specific arguments. The “tying injury” alleged “depend[s] on a careful definition of the tied good market.” United States v. Microsoft Corp., 253 F.3d 34, 95 (D.C. Cir. 2001). Ross argues that it “has offered evidence of cognizable harm in the tied product market in the form of reduced innovation.” D.I. 554 at 24. But I cannot assess that injury if it “takes place in a mar- ket that has not been defined.” Microsoft, 253 F.3d at 95 (quoting Jefferson Parish, 466 U.S. at 29). Ross also argues that there was “foreclosure in the legal search tools

market.” D.I. 554 at 24. When “the theory is one of market foreclosure, [I] must exam- ine the structure of the market supposedly foreclosed.” Town Sound, 959 F.2d at 493. Defining a market requires, at a minimum, “describing those groups of producers which, because of the similarity of their products, have the ability … to take signifi- cant amounts of business away from each other.” SmithKline Corp. v. Eli Lilly & Co., 575 F.2d 1056, 1063 (3d Cir. 1978). “The outer boundaries of a product market are

determined by evaluating which products would be reasonably interchangeable by consumers for the same purpose.” Brokerage Concepts, 140 F.3d at 513. To recap, Ross had to define the market for both its per se and rule-of-reason theory. The per se tying theory requires defining the tying market—the market for public law. And here, the rule-of-reason tying theory requires defining the tied mar- ket—the market for legal search tools. (ii) Ross failed to define either market Ross failed to define either the tying or tied product market. It tried to do that by relying on the expert opinion of Dr. Ratliff. But he spent just three short paragraphs

defining the market for electronically searchable collections of U.S. law and six short paragraphs defining the market for legal search tools. D.I. 532-1, Ex. 33 at 73, 80–82. In these short paragraphs, he fails to analyze in any depth companies or products that are interchangeable with the products here. Brokerage Concepts, 140 F.3d at 513. This makes it difficult to understand the scope of the market he is attempting to interact with. His analysis is not enough. In fact, his analysis is so lacking that I grant in part Plaintiffs’ Motion to Preclude

Dr. James Ratliff and Dr. Gillian Hadfield. D.I. 529. I exclude the section of Dr. Rat- liff’s opinion devoted to market definition under Federal Rule of Evidence 702 and Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 592–93 (1993). To be admissible, qualified experts’ opinions must be based on reliable methodology and must fit the facts of the case. Elcock v. Kmart Corp., 233 F.3d 734, 741 (3d Cir. 2000). Here, Dr. Ratliff essentially has no methodology for defining the relevant markets. He includes

no math or economic modeling. He never analyzes potential competitors in any depth. All he does is make brief, conclusory assertions. That is not enough. Ross fails to define the relevant product markets. Without defining them ade- quately, it cannot reach the issue of market power. So I grant Plaintiffs’ Motion for Summary Judgment on Ross’s Antitrust Counterclaims (No. 2) Market Definition and Market Power. D.I. 521. III. OTHER OUTSTANDING MOTIONS RELEVANT TO THE ANTITRUST COUNTERCLAIMS ARE MOOT I need not reach Thomson Reuters’s other summary-judgment motions, because granting summary judgment on either separate products or market definition alone resolves the outstanding antitrust counterclaims. Turning to the expert motions, the portions of Plaintiffs’ Motion to Preclude Dr. James Ratliff and Dr. Gillian Hadfield that I did not already exclude for failing to

meet Daubert are moot. D.I. 529. These expert reports were submitted to further the antitrust counterclaims and no longer matter now that those counterclaims are resolved. For the same reason, Defendant’s Motion to Exclude Certain Opinions of Plaintiffs and Counter Defendants’ Expert Chad Syverson is moot too. D.I. 516. If either party tries to use these expert opinions in the copyright portion of this case, the parties may resubmit motions to exclude them.

* * * * * For all these reasons, I grant summary judgment to Thomson Reuters on its motions on separate products and market definition and market power. I dismiss as moot the other motions on the antitrust counterclaims.