hiQ Labs, Inc. v. Linkedin Corporation

• Court: District Court, N.D. California
• Decided: September 9, 2020
• Docket: 3:17-cv-03301
• Status: Unknown

Opinion

1 2 3 4 UNITED STATES DISTRICT COURT 5 NORTHERN DISTRICT OF CALIFORNIA 6 7 HIQ LABS, INC., Case No. 17-cv-03301-EMC

8 Plaintiff, ORDER GRANTING IN PART AND 9 v. DENYING IN PART DEFENDANT’S MOTION TO DISMISS 10 LINKEDIN CORPORATION, Docket No. 137 11 Defendant.

12 13 Plaintiff hiQ Labs, Inc. has filed suit against LinkedIn Corp. seeking declaratory relief as 14 well as injunctive relief and damages. According to hiQ, it has not violated any laws by 15 “scraping” public information about LinkedIn users from LinkedIn’s website. Furthermore, hiQ 16 asserts, LinkedIn has violated various antitrust and fair practices laws by trying to prevent hiQ 17 from accessing the public information on the website. Currently pending before the Court is 18 LinkedIn’s motion to dismiss the first amended complaint (“FAC”). The motion primarily 19 challenges the antitrust claims asserted for the first time in the FAC. 20 Having considered the parties’ briefs as well as the oral argument of counsel, the Court 21 hereby GRANTS in part and DENIES in part the motion to dismiss. 22 I. FACTUAL & PROCEDURAL BACKGROUND 23 The main allegations in the FAC which are related to the antitrust claims are as follows. 24 “LinkedIn is the world’s largest professional social network, with over 660 million 25 members.” FAC ¶ 2. Individuals who use the LinkedIn network can make certain information 26 about themselves – e.g., resumes and work history – publicly available on the LinkedIn website. 27 See FAC ¶ 2. 1 information publicly available on LinkedIn. FAC ¶ 3. “People analytics” is

2 a new type of predictive data analysis aimed at providing employers in-depth, predictive insights into their workforce. People analytics 3 generally work by performing computerized analyses of employees’ public professional information and history that then show which 4 employees are at higher risk of looking for a new job, and which may have skills that are not being utilized in their current job. hiQ 5 created two specific analytics services: (a) “Keeper,” which tells employers which of their employees are at the greatest risk of being 6 recruited away, and (b) “Skill Mapper,” a summary of the breadth and depth of aggregate or individual skills of current or prospective 7 employees, which may not be obvious from internal company documents (such as internal performance reviews or the resume the 8 employee supplied as part of the hiring process) or conversations with those employees. 9 10 FAC ¶ 33. Before hiQ, people analytics either did not exist or was not very accurate – e.g., 11 employers relied on internal data only and took an ad hoc approach; hiQ, with its people analytics 12 services, was “able to reduce hard costs and transaction costs” for employers. FAC ¶ 35.

13 LinkedIn eventually realized that it might be able to profit by providing the same type of innovative and revolutionary analytics 14 hiQ pioneered, and it developed its own competing version of that analytics service. Then, in May 2017, LinkedIn abruptly, 15 unlawfully and without cause denied hiQ access to the portion of the LinkedIn website containing wholly public member profiles. 16 17 FAC ¶ 6. 18 hiQ is not alone in being denied access to the public portion of LinkedIn’s website; 19 LinkedIn has denied access to other people analytics providers as well. See FAC ¶ 46. However, 20 LinkedIn has not denied access to companies that do not provide people analytics services (e.g., 21 Google and Bing). See FAC ¶ 47. 22 “The most immediate anticompetitive effect of LinkedIn’s conduct . . . was to eliminate – 23 effectively overnight – nearly all of its people analytics competitors.” FAC ¶ 53. People analytics 24 providers who remain either offer their services at much higher prices than LinkedIn does, or at 25 the same price but with inferior quality. See FAC ¶ 54. Through its actions, LinkedIn has thus 26 reduced consumer choice and price competition. See FAC ¶ 58(d). 27 LinkedIn’s conduct implicates two product markets. First, there is the market for 1 differentiated from traditional social networking platforms because the former, unlike the latter, 2 focus only on business relationships. See FAC ¶¶ 20-21 (alleging, inter alia, that “consumers 3 would reasonably switch to other professional social network platforms for business purposes (if 4 they could), but would not similarly use more traditional social networks for the same purposes”). 5 “LinkedIn was, for many years, the only real professional social networking platform . . . .” FAC 6 ¶ 25. Today, LinkedIn has “well over 75% of all professional social network users in the United 7 States.” FAC ¶ 25. 8 Second, there is the market for people analytics services. This “type of service . . . did not 9 exist until hiQ first came into being, and the only alternative to such services is the previous set of 10 ad hoc measures that companies employed when trying to guess employee attrition risk and 11 employees’ full and current skillsets.” FAC ¶ 36. 12 Based on, inter alia, the above allegations, hiQ has asserted three antitrust claims. 13 • Monopolization, in violation of § 2 of the Sherman Act. According to hiQ, 14 LinkedIn has acquired and maintained monopoly power in the markets for 15 professional social networking platforms and people analytics services through 16 unlawful means, including “leveraging, lock-in, raising rivals’ costs, tying, 17 unilateral refusal to deal, denial of essential facilities, and vertically-arranged 18 boycotts.” FAC ¶ 12. 19 • Attempted monopolization, also in violation of § 2 of the Sherman Act. According 20 to hiQ, there is a dangerous probability that LinkedIn will monopolize the market 21 for people analytics services because it has engaged in anticompetitive conduct 22 such as “leveraging, lock-in, raising rivals’ costs, tying, unilateral refusal to deal, 23 denial of essential facilities, and vertically-arranged boycotts.” FAC 150. 24 • Unreasonable restraint of trade, in violation of § 1 of the Sherman Act. According 25 to hiQ, LinkedIn and its members have entered into contracts or combinations that 26 have the effect of unreasonably restraining trade. See FAC ¶ 159. The contracts or 27 combinations include “tying arrangements and vertically-arranged boycotts.” FAC 1 In addition to the antitrust claims, hiQ has asserted claims for, inter alia, declaratory relief, 2 intentional interference with contract and prospective economic advantage, and violation of 3 California Business & Professions Code § 17200. 4 LinkedIn moves to dismiss certain claims asserted in hiQ’s FAC. In particular, LinkedIn 5 argues that all claims for damages (i.e., the antitrust and intentional interference claims) should be 6 dismissed based on the Noerr-Pennington doctrine and the California litigation privilege. 7 LinkedIn further argues that the antitrust claims should be dismissed on various merits grounds 8 (e.g., failure to allege antitrust injury, a product market, and anticompetitive conduct). 9 II. DISCUSSION 10 A. Legal Standard 11 Federal Rule of Civil Procedure 8(a)(2) requires a complaint to include “a short and plain 12 statement of the claim showing that the pleader is entitled to relief.” Fed. R. Civ. P. 8(a)(2). A 13 complaint that fails to meet this standard may be dismissed pursuant to Federal Rule of Civil 14 Procedure 12(b)(6). See Fed. R. Civ. P. 12(b)(6). To overcome a Rule 12(b)(6) motion to dismiss 15 after the Supreme Court’s decisions in Ashcroft v. Iqbal, 556 U.S. 662 (2009), and Bell Atlantic 16 Corp. v. Twombly, 550 U.S. 544 (2007), a plaintiff’s “factual allegations [in the complaint] ‘must . 17 . . suggest that the claim has at least a plausible chance of success.’“ Levitt v. Yelp! Inc., 765 F.3d 18 1123, 1135 (9th Cir. 2014). The court “accept[s] factual allegations in the complaint as true and 19 construe[s] the pleadings in the light most favorable to the nonmoving party.” Manzarek v. St. 20 Paul Fire & Marine Ins. Co., 519 F.3d 1025, 1031 (9th Cir. 2008). But “allegations in a 21 complaint . . . may not simply recite the elements of a cause of action [and] must contain sufficient 22 allegations of underlying facts to give fair notice and to enable the opposing party to defend itself 23 effectively.” Levitt, 765 F.3d at 1135 (internal quotation marks omitted). “A claim has facial 24 plausibility when the plaintiff pleads factual content that allows the court to draw the reasonable 25 inference that the defendant is liable for the misconduct alleged.” Iqbal, 556 U.S. at 678. “The 26 plausibility standard is not akin to a probability requirement, but it asks for more than a sheer 27 possibility that a defendant has acted unlawfully.” Id. (internal quotation marks omitted). 1 B. Claims for Damages 2 In the FAC, hiQ seeks declaratory relief, injunctive relief, and damages. The request for 3 damages is based on the following causes of action: (1) the antitrust claims and (2) the intentional 4 interference claims. LinkedIn moves to dismiss both damages claims based on the Noerr- 5 Pennington doctrine. LinkedIn also moves to dismiss the intentional interference claims based on 6 the California litigation privilege. 7 1. Noerr-Pennington Doctrine

8 The essence of the Noerr-Pennington doctrine is that those who petition any department of the government for redress are immune 9 from statutory liability for their petitioning conduct. The doctrine derives from two Supreme Court cases holding that the First 10 Amendment Petition Clause immunizes acts of petitioning the legislature from antitrust liability. The doctrine has since been 11 applied to actions petitioning each of the three branches of government, and has been expanded beyond its original antitrust 12 context. 13 Theme Promotions, Inc. v. News Am. Mktg. FSI, 546 F.3d 991, 1006-07 (9th Cir. 2008). For 14 example, the doctrine can apply to a state claim for tortious interference with prospective 15 economic advantage. See id. at 1007; see also Hi-Top Steel Corp. v. Lehrer, 24 Cal. App. 4th 570, 16 577-78 (1994) (noting that, “[w]hile the Noerr-Pennington doctrine was formulated in the context 17 of antitrust cases, it has been applied or discussed in cases involving other types of civil liability, 18 including liability for interference with contractual relations or prospective economic advantage or 19 unfair competition”). 20 LinkedIn concedes that it has not directly petitioned the government for any relief. 21 However, it points out that “[c]onduct incidental to a lawsuit, including a pre-suit demand letter, 22 falls within the protection of the Noerr-Pennington doctrine.” Id. (emphasis added). LinkedIn 23 maintains that hiQ’s antitrust and intentional interference claims “derive from the cease-and-desist 24 letters LinkedIn sent that ostensibly ‘cut off’ hiQ’s access to LinkedIn’s website.” Mot. at 8. See, 25 e.g., FAC ¶¶ 7, 39 (referring to a cease-and-desist letter sent on May 23, 2017); FAC ¶ 43 26 (referring to letter sent on June 24, 2017, reiterating demand to cease and desist). 27 In response, hiQ argues that “petitioning activity may be actionable if it is part of a broader 1 cutting off of competitors’ access to the public information on LinkedIn’s website – which is not 2 petitioning activity. See Opp’n at 15. 3 hiQ’s argument relies in large part on Hynix Semiconductor Inc. v. Rambus, Inc., 527 F. 4 Supp. 2d 1084 (N.D. Cal. 2007) (Whyte, J.). In Hynix, Rambus was a member of a SSO 5 (standard-setting organization) and allegedly used its membership in the SSO to discover how a 6 particular standard was developing and then drafted patent claims to cover the standard. “Once 7 the industry became ‘locked in’ to the DRAM standard, Rambus sprang the ‘patent trap’ and 8 demanded royalties,” and further “backed up its royalty demands with infringement litigation.” Id. 9 at 1089. Hynix subsequently brought an antitrust claim based on Rambus’s “‘overall course of 10 conduct’ – including [its] patent litigation.” Id. 11 Before Judge Whyte, Rambus argued that, “because its use of the courts to enforce its 12 patents is protected petitioning activity pursuant to the Noerr-Pennington doctrine . . . , Hynix 13 could not claim its litigation expenses as damages.” Id. at 1086. Judge Whyte ultimately 14 disagreed. 15 Judge Whyte noted that there were “three possible approaches to whether patent litigation 16 can be an anticompetitive act in violation of the antitrust laws.” Id. at 1091. 17 • Under the first approach (the Second Circuit’s “absolute bar”), “only frivolous 18 patents can be anticompetitive and part of an unlawful scheme.” Id. at 1091-92. 19 • Under the second approach (labeled the “Kobe claim”) – and at “the opposite end 20 of the spectrum” – even nonfrivolous patent litigation can be anticompetitive if part 21 of a broader anticompetitive scheme (i.e., “brought in conjunction with other 22 antitrust misconduct”). Id. at 1092-94. 23 • The third approach was a middle ground between the first two approaches. See id. 24 at 1094. Under the third approach, “good faith litigation” could be “unlawful if 25 done as part of an anticompetitive scheme” but, in order to get litigation costs as 26 part of the damages for the antitrust violation, there must be “an explicit linkage 27 between [the] antitrust violation and the litigation” – i.e., a causal connection. Id. 1 Judge Whyte endorsed the third approach.

2 [T]he court believes that the Federal Circuit and the Supreme Court would recognize some “scheme” antitrust allegations that include 3 constitutionally protected litigation within the “overall course of conduct,” but only those in which the patent litigation is “causally 4 connected” to anticompetitive harms. . . . The court concludes that before otherwise protected litigation can be a part of an 5 “anticompetitive scheme” claim, the court must first find that the other aspects of the scheme independently produce anticompetitive 6 harms. Once this step has been established, the court should ask whether the accused patent litigation was causally connected to 7 these anticompetitive harms. If yes, an antitrust plaintiff may then include good faith patent litigation as part of the anticompetitive 8 scheme. 9 Id. at 1096-97.1 He added: “[W]here the patent litigation is used to further the harm caused under 10 a ‘more traditional antitrust theory,’ a plaintiff should be allowed a full recovery.” Id. at 1097 11 (emphasis added); see also Arista Networks, Inc. v. Cisco Sys. Inc., No. C-16-0923 BLF (N.D. 12 Cal.) (Docket No. 281) (Order at 16) (taking note of plaintiff’s argument that defendant’s 13 intellectual property “lawsuit was the ‘enforcement mechanism’ of the purported ‘open early, 14 closed late’ anticompetitive scheme” – i.e., defendant used the litigation and conduct incidental to 15 that litigation “as tools to further the ‘closing’ step of the ‘open early, closed late’ scheme”) 16 (emphasis added). 17 According to LinkedIn, the Court should reject the application of Hynix to the instant case 18 because the facts in Hynix and the facts in the instant case are not sufficiently similar. LinkedIn 19 maintains that Hynix-type cases all involve “a bait-and-switch, where a defendant affirmatively 20 1 In its papers, LinkedIn argues that the anticompetitive nonpetitioning activity must be 21 “ineffective” without the petitioning activity in order to avoid the Noerr-Pennington doctrine. See Reply at 6-7. LinkedIn comes up with this requirement because Judge Whyte stated in Hynix as 22 follows:

23 Because Rambus’ alleged conduct at JEDEC can independently qualify as an anticompetitive harm under section 2, the court finds 24 that Rambus’ current patent litigation is “causally connected” to that behavior and therefore properly included in an “anticompetitive 25 scheme” allegation. To be clear, the causal connection is that a patent “ambush” or “hold-up” is ineffective without the threat of 26 litigation.

27 Hynix, 527 F. Supp. 2d at 1098. But as should be clear from the above, although Judge Whyte did 1 misled competitors into using its technology and intellectual property on the understanding that no 2 legal action would result from doing so.” Reply at 5. LinkedIn’s argument is not persuasive. It 3 provides no rationale as to why Hynix should be narrowly construed to these facts. Hynix clearly 4 made a broader point: it asks whether the petitioning activity furthered an anticompetitive scheme 5 involving nonpetitioning activity; if the gravamen of the action centers on nonpetitioning activity, 6 the fact that petitioning activity is employed to further that conduct is not sufficient to implicate 7 the Noerr-Pennington doctrine. 8 LinkedIn argues that even if Hynix has broader application, its cease-and-desist letters 9 (petitioning activity) did not further the anticompetitive scheme that involved the blocking of 10 access to its website (nonpetitioning activity). LinkedIn points out that its first cease-and-desist 11 letter was issued before it blocked hiQ’s access to the website. See Reply at 7 (“The [first] letter 12 preceded any technical cut-off by a full month.”). But LinkedIn’s position is problematic for two 13 reasons. 14 • First, LinkedIn’s second cease-and-desist letter was issued on the same day that 15 LinkedIn blocked access. See Mot. at 5 (“LinkedIn implemented IP address blocks 16 on June 24, the same day it sent the second letter.”). Therefore, arguably, the 17 second letter was used to further the anticompetitive scheme involving the 18 blocking. 19 • Second, it is not clear that the causal connection requirement always demands that 20 the anticompetitive petitioning activity come after the anticompetitive 21 nonpetitioning activity; a cease-and-desist letter that comes before the actual 22 blocking still reinforces the overall anticompetitive scheme. 23 At bottom, LinkedIn’s argument defies the fundamental logic of Hynix. 24 Finally, LinkedIn contends that, should the Court favor hiQ on the above arguments, then 25 the Court should – at the very least – restrict hiQ’s damages claims. More specifically, LinkedIn 26 maintains that, if the anticompetitive nonpetitioning activity is the blocking of hiQ’s access, then 27 damages should be limited to the timeframe when hiQ was actually blocked – a “six-day period in 1 period is necessarily based on LinkedIn’s letters and should be dismissed”) (emphasis in original); 2 see also Reply at 6. But LinkedIn’s position is not persuasive because, even if petitioning activity 3 arguably cannot be a basis for liability under the Noerr-Pennington doctrine, that does not mean 4 that a plaintiff cannot get damages based on the petitioning activity,2 and the petitioning activity 5 can take place at a different time than the nonpetitioning activity. Furthermore, LinkedIn takes too 6 narrow a view of the impact that the nonpetitioning activity (blocking of access) could have – that 7 impact could extend beyond the time of the actual blocking. For example, even if the actual 8 blocking lasted for only six days, it seems plausible that this conduct might dissuade hiQ 9 customers from subsequently signing up for hiQ’s people analytics services. Cf. Arista (Docket 10 No. 281) (Order at 21-22) (noting that “a reasonable jury could find that [the defendant] raised the 11 specter of the . . . lawsuit to persuade customers to abandon [the defendant’s] competitors”). 12 Accordingly, the Noerr-Pennington doctrine does not bar hiQ’s antitrust and interference 13 claims. 14 2. California Litigation Privilege 15 As noted above, LinkedIn argues that the state claims for intentional interference are also 16 protected by the California litigation privilege. The parties’ briefs indicate that the above analysis 17 for Noerr-Pennington applies here as well. For the reasons stated above, the California litigation 18 privilege is not a bar to hiQ’s interference claims. 19 C. Antitrust Claims: Product Market 20 LinkedIn contends the antitrust claims are still deficient for independent reasons, including 21 a failure on the part of hiQ to adequately allege a product market – i.e., a people analytics market.3 22 2 See Hovenkamp, et al., IP & Antitrust § 11.04[F] (suggesting that liability should not be based 23 on protected petitioning but damages are a different matter: “‘a plaintiff that can prove an antitrust violation without the use of protected petitioning can recover damages caused by that petitioning 24 as well as damages by the conduct that proved the violation’”).

25 3 Technically, hiQ makes reference to two different product markets in the FAC – (1) the market for people analytics services and (2) the market for professional social networking platforms. See, 26 e.g., FAC ¶ 110 (alleging that “LinkedIn has willfully acquired and maintained monopoly power in the relevant markets for professional social networking platforms and people analytics”); FAC ¶ 27 159 (alleging that “LinkedIn and its members have entered into contracts or combinations that 1 A product market “encompass[es] the product at issue as well as all economic substitutes 2 for the product.” Newcal Indus. v. Ikon Office Sol., 513 F.3d 1038, 1045 (9th Cir. 2008). “‘The 3 outer boundaries of a product market are determined by the reasonable interchangeability of use or 4 the cross-elasticity of demand between the product itself and substitutes for it.’”4 Id. “[W]hat 5 constitutes a relevant market is a factual determination for the jury,” Image Tech. Servs. v. 6 Eastman Kodak Co., 125 F.3d 1195, 1203 (9th Cir. 1997); see also Todd v. Exxon Corp., 275 F.3d 7 191, 199-200 (2d Cir. 2001) (stating that, “[b]ecause market definition is a deeply fact-intensive 8 inquiry, courts hesitate to grant motions to dismiss for failure to plead a relevant product market”); 9 however, the relevant market must still be plausibly alleged to make it past a 12(b)(6) challenge. 10 In the instant case, the FAC indicates that people analytics is “a new type of predictive data 11 analytics aimed at providing employers in-depth predictive insights into their workforce.” FAC ¶ 12 33. The FAC also alleges that “[p]eople analytics generally work by performing computerized 13 analyses of employees’ public professional information and history.” FAC ¶ 33. Nevertheless, 14 the parameters of the people analytics market – as pled – are vague. Most notably, as the Court 15 discussed with the parties at the hearing, it is not clear what substitutes there are for people 16 analytics products such as those offered by hiQ. Should a product be considered a substitute if the 17 people analytics are based on an employer’s internally maintained data? Should a product be 18 considered a substitute if the people analytics are based on publicly available information other 19 than that available on LinkedIn’s website? Even if more established analytic techniques have 20 become dated, that does not mean there is no substitutability or cross-elasticity of demand between 21 different modes of people analytics. 22 At the hearing, hiQ asserted that a product that uses an employer’s internally maintained 23 data should not be deemed a substitute but the FAC fails to explain why. For example, the FAC 24

25 attempt to define the people analytics market.

26 4 “Elasticity of demand is a concept used to signify the relationship between changes in price and responsive changes in demand.” United States v. LSL Biotechnologies, 379 F.3d 672, 697 (9th 27 Cir. 2004); see also Eastman Kodak Co. v. Image Tech. Servs., 504 U.S. 451, 469 (1992) 1 does not explain why internal data such as performance reviews and resumes should be considered 2 “limited” in nature. FAC ¶ 35. The FAC also fails to explain why, even if an employer was not 3 able to use internal data successfully itself, see FAC ¶ 35 (alleging that “employers tried to 4 manage attrition risk within their organizations through a variety of ad hoc internal methods, and 5 they tried to identify employees’ full skillsets through similarly ad hoc internal means”), the 6 employer could not hire a company to put that data to good use. Indeed, hiQ suggests that that 7 was the conventional approach before hiQ came on the scene with its analytics. It is not plausible 8 to suggest without specific facts that the entire field of analytics using internal data is now 9 obsolete. In fact, it would appear that internal data which might include direct polling could yield 10 information not otherwise accessible via LinkedIn. 11 As for a product that bases its people analytics on publicly available information other than 12 LinkedIn, hiQ made a different argument at the hearing. According to hiQ, in theory, such a 13 product would be a substitute (and thus in the same market as hiQ’s products) but, hiQ 14 maintained, as a practical matter, the only place to get publicly available data is LinkedIn. But 15 similar to above, the FAC contains no specific allegations establishing such. The Court does not 16 doubt that LinkedIn is a useful source for publicly available data given its focus on professional 17 social networking and its prominence in the professional social networking space; however, that 18 does not mean that useful publicly available information cannot be gleaned from other sources 19 such as Google and Facebook or other industry directories and sources. Indeed, in this day and 20 age, it would not be surprising if a person’s digital footprint – even apart from LinkedIn – were 21 enlightening to the analytics task. 22 The Court acknowledges hiQ’s suggestion that products using employer internal data or 23 publicly available data other than LinkedIn’s are different in quality from hiQ’s products – and 24 thus it is at least a question of fact whether there is some elasticity of demand between them and 25 whether those products are in the same market as hiQ’s products. See generally Brown Shoe Co. 26 v. United States, 370 U.S. 294, 326, 82 S. Ct. 1502, 1524-25 (1962) (“agree[ing] with the District 27 Court that in this case a further division of product lines based on ‘price/quality’ differences 1 Antitrust Litig., 247 F.R.D. 98, 129 (C.D. Cal. 2007) (indicating that consumers may differentiate 2 or distinguish among products based on performance, price, and so forth but that does not 3 necessarily mean that the products are in separate markets). The problem for hiQ is that it has not 4 yet shown that it is plausible that the relevant market should be defined as that which uses only 5 LinkedIn data. 6 Accordingly, the Court finds all of hiQ’s antitrust claims deficient for failure to adequately 7 allege a product market.5 The Court, however, shall give hiQ an opportunity to amend to correct 8 this deficiency. 9 D. Antitrust Claims: Anticompetitive Conduct 10 There are additional reasons why the antitrust claims are not viable as currently pled. In 11 particular, LinkedIn persuasively argues that hiQ has failed to adequately allege anticompetitive 12 conduct. 13 For its first antitrust claim – monopolization in violation of § 2 of the Sherman Act – hiQ 14 alleges as follows:

15 LinkedIn has willfully acquired and maintained monopoly power in the relevant markets [i.e., the market for professional social 16 networking platforms and the market for people analytics services], by means of predatory, exclusionary, and anticompetitive conduct, 17 including but not limited to by means of lock-in, raising rivals’ costs, tying, unilateral refusal to deal, denial of essential facilities, 18 and vertically-arranged boycotts . . . . 19 FAC ¶ 112. hiQ repeats these theories for its second antitrust claim, i.e., attempted 20 monopolization). See FAC ¶ 150. For the final antitrust claim – concerted action that 21 unreasonably restrains trade – hiQ asserts the tying and vertical boycott theories only. See FAC ¶ 22 162. Each of the above theories, as addressed below, is deficient. 23 1. Unilateral Refusal to Deal 24 As an initial matter, the Court rejects LinkedIn’s contention that hiQ cannot raise a theory 25 of unilateral refusal to deal because, during the Ninth Circuit proceedings, hiQ stated that it was 26

27 5 Because the Court finds a deficiency in the allegation of a product market, it does not address 1 not asserting such a theory. See Mot. at 21 (asserting that, “[h]aving discouraged the Ninth Circuit 2 from considering its case under a duty-to-deal framework, hiQ cannot now revive this theory 3 under the exact same facts”). Although LinkedIn makes an analogy to the doctrine of judicial 4 estoppel, that doctrine requires that a court rely on a representation being made by the party 5 against whom estoppel is asserted. See Casa Del. Caffe Vergnano S.P.A. v. Italflavors San Diego, 6 Ltd. Liab. Co., 816 F.3d 1208, 1213 (9th Cir. 2016) (noting that judicial estoppel “‘generally 7 prevents a party from prevailing in one phase of a case on an argument and then relying on a 8 contradictory argument to prevail in another phase’”). LinkedIn has made no showing that the 9 Ninth Circuit relied on hiQ’s representation above in making its decision on hiQ’s motion for a 10 preliminary injunction. 11 As for the merits of the unilateral refusal-to-deal theory, the Court begins with the 12 predicate that, “[a]s a general rule, businesses are free to choose the parties with whom they will 13 deal, as well as the prices, terms, and conditions of that dealing.” Pac. Bell Tel. Co. v. linkLine 14 Communs., Inc., 555 U.S. 438, 448 (2009). This includes dealing or not dealing with a rival. See 15 Aerotec Int’l, Inc. v. Honeywell Int’l, Inc., 836 F.3d 1171, 1184 (9th Cir. 2016) (stating that “there 16 is ‘no duty to deal under the terms and conditions preferred by [a competitor’s] rivals’”). 17 However, there are “limited circumstances in which a firm’s unilateral refusal to deal with its 18 rivals can give rise to antitrust liability,” as indicated in the Supreme Court’s seminal case Aspen 19 Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585 (1985). See Aerotec, 836 F.3d at 1184. 20 That being said, the Aspen Skiing exception is very narrow, as both the Supreme Court and Ninth 21 Circuit have made clear. See Verizon Communs., Inc. v. Law Offices of Curtis v. Trinko, LLP, 540 22 U.S. 398, 409 (2004) (characterizing Aspen Skiing as “at or near the boundary of § 2 liability”); 23 FTC v. Qualcomm Inc., 969 F.3d 974 (9th Cir. 2020), available at No. 19-16122, 2020 U.S. App. 24 LEXIS 25347, at *30 (9th Cir. Aug. 11, 2020) (stating that Aspen Skiing represents a “limited 25 exception to [the] general rule that there is no antitrust duty to deal”); Aerotec, 836 F.3d at 1184 26 (stating that Aspen Skiing is a “narrow exception”). 27 In Aspen Skiing, the defendant owned three of four mountain areas for skiing, and the 1 had cooperated for years in the issuance of a joint, multiple-day, all- area ski ticket. After repeatedly demanding an increased share of 2 the proceeds, the defendant canceled the joint ticket. The plaintiff, concerned that skiers would bypass its mountain without some joint 3 offering, tried a variety of increasingly desperate measures to re- create the joint ticket, even to the point of in effect offering to buy 4 the defendant’s tickets at retail price. The defendant refused even that. [The Supreme Court] upheld a jury verdict for the plaintiff, 5 reasoning that “[t]he jury may well have concluded that [the defendant] elected to forgo these short-run benefits because it was 6 more interested in reducing competition . . . over the long run by harming its smaller competitor.” 7 8 Trinko, 540 U.S. at 408-09 (2004). 9 The Ninth Circuit has explained that, under Aspen Skiing,

10 a company engages in prohibited, anticompetitive conduct when (1) it “unilateral[ly] terminat[es] . . . a voluntary and profitable course 11 of dealing”; (2) “the only conceivable rationale or purpose is ‘to sacrifice short-term benefits in order to obtain higher profits in the 12 long run from the exclusion of competition’“; and (3) the refusal to deal involves products that the defendant already sells in the existing 13 market to other similarly situated customers. 14 Qualcomm, 2020 U.S. App. LEXIS 25347, at *30-31. 15 In the instant case, hiQ has failed to adequately allege at least the first two Aspen Skiing 16 factors. For example, hiQ has failed to make a plausible case that it and LinkedIn had engaged in 17 a voluntary course of dealing. In ¶ 37, hiQ alleges as follows:

18 LinkedIn has known of hiQ since at least 2015, when it started participating in hiQ’s annual “Elevate” conference. The hiQ Elevate 19 conference was designed to build a community around the emerging field of people analytics and has provided a regular forum for 20 participants to share insights and disseminate best practices. LinkedIn has sent representatives to each iteration of that conference 21 since hiQ’s founding. hiQ has spoken freely about its public data collection from LinkedIn at Elevate; LinkedIn has thus always 22 understood what hiQ does. Over the years, LinkedIn has itself participated regularly in hiQ Elevate events. At a 2016 Elevate 23 conference, LinkedIn employee Lorenzo Canlas received special recognition and accepted the hiQ Elevate “Impact Award.” 24 25 FAC ¶ 37. But that certain LinkedIn employees knew about hiQ and its practices does not mean 26 that the two companies were thereby dealing with one another. It is not even clear whether these 27 LinkedIn employees were high-level employees who could act on behalf of the company. And 1 or an “implicit understanding” between the companies. LiveUniverse v. MySpace, Inc., 304 Fed. 2 Appx. 554, 557 (9th Cir. 2008) (noting “nothing in the complaint suggests an agreement, or even 3 an implicit understanding, between MySpace and LiveUniverse”) (emphasis omitted). Unlike 4 Aspen Skiing, there was no mutual commerce or course of dealing between LinkedIn and hiQ. 5 Moreover, even if there were a voluntary course of dealing between LinkedIn and hiQ, hiQ 6 still must show that LinkedIn sacrificed a profitable course of dealing in the short term in order to 7 benefit long term from the exclusion of competition. Absent what would otherwise be irrational 8 short-term behavior (like the defendant’s conduct in Aspen Skiing), there is no antitrust claim for 9 refusal to deal. To establish this element, hiQ maintains that “people analytics help prove 10 LinkedIn’s value proposition; specifically, they demonstrate why it is valuable for both employees 11 and employers to belong to LinkedIn’s professional social network, and why the public 12 employment database benefits those who participate.” FAC ¶ 38. Thus, hiQ implies that 13 LinkedIn lost business as a result of hiQ’s being excluded from access to information publicly 14 available on LinkedIn’s website. But hiQ has not specifically alleged that, e.g., employers stopped 15 using LinkedIn’s services (or discouraged employees from using LinkedIn’s services) after 16 LinkedIn blocked hiQ from accessing its website. Its theory of short-term sacrifice is far from 17 obvious or even intuitive. Absent specific factual allegations, it is speculative to assume that 18 LinkedIn gave up any short-term benefit by refusing to deal with hiQ. 19 Accordingly, to the extent hiQ’s antitrust claims are predicated on a unilateral refusal to 20 deal, the Court finds such a theory, as pled, implausible. 21 2. Denial of Essential Facilities 22 LinkedIn also asserts the essential facilities doctrine as another theory of anticompetitive 23 conduct.

24 [An essential facilities] theory is a variation on a [unilateral] refusal to deal claim. It imposes liability where competitors are denied 25 access to an input that is deemed essential, or critical, to competition. Although the Supreme Court has never recognized the 26 doctrine, [the Ninth Circuit has] continued to treat it as having a basis in § 2 of the Sherman Act. 27 control of an essential facility, (2) that [the plaintiff], as [the 1 defendant’s] competitor, is unable reasonably or practically to duplicate the facility, (3) that [the defendant] has refused to provide 2 [the plaintiff] access to the facility, and (4) that it is feasible for [the defendant] to provide such access. Because mandating access, as 3 the essential facilities doctrine implies, shares the same concerns as mandating dealing with a competitor, a facility is essential “only if 4 control of the facility carries with it the power to eliminate competition in the downstream market.” 5 6 Aerotec, 836 F.3d at 1184-85; see also Alaska Airlines, Inc. v. United Airlines, Inc., 948 F.2d 536, 7 544 (9th Cir. 1991) (stating that “[a] facility that is controlled by a single firm will be considered 8 ‘essential’ only if control of the facility carries with it the power to eliminate competition in the 9 downstream market”). 10 In the instant case, hiQ asserts that LinkedIn’s social networking platform amounts to an 11 essential facility. However, the Court cannot assess the viability of hiQ’s essential facilities 12 argument without there first being a properly defined downstream market (i.e., the people 13 analytics market). The Court therefore dismisses the essential facilities theory based on a failure 14 to adequately allege a people analytics market. 15 3. Lock-In 16 In the FAC, hiQ further alleges anticompetitive conduct on the basis that

17 LinkedIn lured users desiring professional social networking services into joining LinkedIn with the promise that their 18 employment information would be their own and could be publicly shared and accessed. . . . At a certain point, LinkedIn’s growth and 19 dominance became self-reinforcing, meaning that users needed to stay a part of its professional social network platform, or else be 20 excluded from the largest and most useful professional social network in the world. 21 22 FAC ¶ 116. Allegedly, a similar lure and lock into the network took place with respect to 23 employers: “LinkedIn promised prospective employer members that they could access users’ 24 public employment information without restriction.” FAC ¶ 117. According to hiQ, “[o]nce users 25 were locked into its professional social network platform, LinkedIn utilized the power that lock-in 26 conferred in order to advantage itself to the exclusion of its people analytics competitors . . . .” 27 FAC ¶ 118. 1 fails to map onto the “classic” lock-in situation “where a consumer buys one product from a 2 company, and then, without warning, is forced to buy aftermarket products or services from that 3 same company at monopolistic prices in order to use the product.” Mot. at 28 (citing Eastman 4 Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 476 (1992)). In response, hiQ contends that 5 there must be a flexible approach in evaluating whether conduct is anticompetitive. See Avaya 6 Inc., RP v. Telecom Labs, Inc., 838 F.3d 354, 404 (3d Cir. 2016) (noting that the court had 7 previously “declined to read Kodak as applying narrowly to only cases involving ‘[a]n aftermarket 8 policy change’”). 9 While hiQ makes a fair point regarding flexibility, it is still invoking the concept of a lock- 10 in as a basis for the antitrust violation – and a lock-in typically means that a locked-in consumer is 11 exploited, as even the Third Circuit case cited by hiQ indicates. See id. (stating that, under Kodak, 12 “exploitation of locked-in customers is one theory that courts will recognize to justify [antitrust] 13 liability” for conduct in an aftermarket related to a primary market) (emphasis added); see also 14 Mich. Div. – Monument Builders of N. Am. v. Mich. Cemetery Ass’n, 524 F.3d 726, 737 (6th Cir. 15 2008) (noting that “[m]arket power may exist in a lock-in case where once a customer buys one 16 product, he or she is locked in to buying another product because of the seller’s rules”); Xerox 17 Corp. v. Media Scis., Inc., 660 F. Supp. 2d 535, 546-47 (S.D.N.Y. 2009) (noting that post-Kodak 18 decisions “establish that for a nonmonopolist producer of a durable good to be held to have 19 monopoly power in the aftermarket for parts, service, or supplies, a plaintiff generally must show 20 that (i) customers who own the good are ‘locked in’ by the prohibitive costs of switching to an 21 alternate product, and (ii) the lock-in permitted those customers to be exploited”). In the instant 22 case, hiQ has not explained how LinkedIn has exploited its locked-in customers (i.e., users). 23 Indeed, hiQ does not appear to claim any exploitation of locked-in customers; instead, it suggests 24 that LinkedIn has used the lock-in to harm people analytics providers. See Opp’n at 22. Similar 25 to the problem with hiQ’s tying argument, see infra, this is not a case whether the same party 26 lured to do business with the defendant is then evoked in to purchase additional products and 27 serving from that defendant. 1 dismisses it. 2 4. Tying 3 “A tying arrangement is ‘an agreement by a party to sell one product but only on the 4 condition that the buyer also purchases a different (or tied) product, or at least agrees that he will 5 not purchase that product from any other supplier.’”6 Eastman Kodak, 504 U.S. at 461-62. “[I]f 6 the seller has ‘appreciable economic power’ in the tying product market and if the arrangement 7 affects a substantial volume of commerce in the tied market,” there is an antitrust violation. Id. at 8 462.

9 “The essential characteristic of an invalid tying arrangement lies in the seller’s exploitation of its control over the tying product to force 10 the buyer into the purchase of a tied product that the buyer either did not want at all, or might have preferred to purchase elsewhere on 11 different terms. When such ‘forcing’ is present, competition on the merits in the market for the tied item is restrained and the Sherman 12 Act is violated.” 13 Id. at 464 n.9. 14 In the FAC, hiQ alleges that the tying product is LinkedIn’s “professional social 15 networking platform” and the tied product is people analytics services. FAC ¶ 125. hiQ further 16 alleges that there is illegal tying because “LinkedIn has conditioned the provision of its dominant 17 professional social networking platform on the use of its people analytics services (or the non-use 18 of its competitors’ people analytics services).” FAC ¶ 126. 19 The tying claim, however, makes no sense. hiQ has not explained how LinkedIn has 20 essentially forced employers to purchase LinkedIn’s people analytics product (or forced employers 21 not to purchase other companies’ comparable products) in order for the employers to use 22 LinkedIn’s professional social networking platform. Again, there are two different consumer 23 groups: (1) those individuals who use LinkedIn’s professional social network, and (2) employers 24

25 6 A tying theory is usually brought under § 1 of the Sherman Act instead of § 2. See, e.g., Systemcare, Inc. v. Wang Labs. Corp., 117 F.3d 1137, 1142-43 (10th Cir. 1997) (“hold[ing] that a 26 contract between a buyer and seller satisfies the concerted action element of section 1 of the Sherman Act where the seller coerces a buyer’s acquiescence in a tying arrangement imposed by 27 the seller[;] [t]he essence of section 1’s contract, combination, or conspiracy requirement in the 1 want to purchase people analytics products. As noted above, there is no tying of products directed 2 at the same consumer group. 3 5. Vertically-Arranged Boycotts 4 hiQ’s theory of a vertical boycott suffers from a similar flaw as does its tying theory. 5 According to hiQ, “LinkedIn has induced, coerced, or otherwise forced LinkedIn members 6 (including without their knowledge) to boycott LinkedIn’s competitors for the people analytics 7 service.” FAC ¶ 142. But hiQ has failed to explain how LinkedIn was able to coerce its members; 8 there is no allegation that, e.g., LinkedIn does not allow anyone to use its professional social 9 networking platform if he/she/it purchases people analytics services from hiQ or any other 10 provider. In its opposition, hiQ argues that LinkedIn improperly prohibits its members from 11 sharing their information as they choose (contrary to what it represented to members to lure them 12 in). But even if LinkedIn engaged in some kind of misconduct there, that does not explain how 13 any members were thereby forced to boycott hiQ’s or other companies’ people analytics services. 14 6. Raising Rivals’ Costs 15 In the FAC, hiQ alleges that, “[b]y foreclosing people analytics competitors’ access to the 16 public employment information database on its site, LinkedIn raised those competitors’ costs to do 17 business.” FAC ¶ 121. As LinkedIn points out in its papers, this theory seems to be nothing more 18 than a unilateral refusal to deal, and this theory is problematic for the reasons discussed above. 19 In its opposition brief, hiQ tries to get around this problem by claiming that this theory is 20 based on both unilateral action by LinkedIn (a § 2 violation) and concerted action between 21 LinkedIn and its members (a § 1 violation). See Opp’n at 18 (arguing that the refusal-to-deal 22 standard discussed above “applies only to unilateral conduct” and not to concerted conduct). 23 According to hiQ, this theory involves concerted action because LinkedIn’s User Agreement 24 “prevent[s] people analytic providers [as users] from accessing [other] users’ otherwise public 25 employment” and, by relying on the User Agreement, “LinkedIn artificially raised its people 26 analytics’ competitors’ costs and pushed them out of the market.” Opp’n at 19. But the § 1 claim, 27 as pled in the FAC, is based only on tying and vertical boycotts. See FAC ¶ 162 (“These 1 vertically-arranged boycotts.”). No viable Section 1 claim is stated. Even if that were not the 2 case, § 1 claims typically involve concerted action between multiple defendants or between a 3 defendant and a third party that harms the plaintiff – not concerted action between the defendant 4 and the plaintiff. hiQ cites no precedent for this novel theory. 5 7. Leveraging 6 Finally, LinkedIn attacks the leveraging theory tendered by hiQ. With respect to 7 leveraging, hiQ alleges as follows in the FAC:

8 LinkedIn has used its monopoly power in [the market for professional social networking platforms] in a predatory, 9 exclusionary, and anticompetitive manner to monopolize the people analytics services market and exclude competitors from that market, 10 including but not limited to by means of lock-in, raising rivals’ costs, tying, unilateral refusal to deal, denial of essential facilities, 11 and vertically-arranged boycotts. 12 FAC ¶ 114. LinkedIn argues that, based on the above allegation, the leveraging theory is simply a 13 derivative theory. LinkedIn emphasizes that the fact of leveraging, in and of itself, does not give 14 rise to a viable claim; rather, there must also be some anticompetitive conduct. See Doe v. Abbott 15 Labs., 571 F.3d 930, 931 (9th Cir. 2009) (holding that “allegations of monopoly leveraging 16 through pricing conduct in two markets [does not] state a claim under § 2 of the Sherman Act, 17 absent an antitrust refusal to deal (or some other exclusionary practice) in the monopoly market or 18 below-cost pricing in the second market”). hiQ does not disagree that there be anticompetitive 19 conduct in order for there to be a viable leveraging theory. This makes sense because an antitrust 20 violation requires that there be anticompetitive conduct and leveraging by itself is not inherently 21 anticompetitive in nature. 22 Accordingly, as LinkedIn argues, the leveraging theory rises or falls with the other theories 23 identified above. 24 8. Summary 25 For the foregoing reasons, the Court finds each theory of anticompetitive conduct 26 implausible. The Court further finds the majority of these theories futile. The only theories where 27 hiQ may be able to allege a plausible claim for relief are (1) the unilateral refusal to deal and (2) 1 it notes it is skeptical of the refusal-to-deal theory in light of the narrowness of the Aspen Skiing 2 exception.7 3 III. CONCLUSION 4 LinkedIn’s motion to dismiss is granted in part and denied in part. 5 The motion to dismiss the interference claims is denied as they are not barred by the 6 Noerr-Pennington doctrine or the California litigation privilege. 7 The motion to dismiss the antitrust claims is granted. hiQ has failed to adequately allege a 8 product market (i.e., the people analytics market). In addition, hiQ has failed to adequately allege 9 anticompetitive conduct. The Court shall give hiQ leave to amend its antitrust claims, but only to 10 the extent they are based on the theories of unilateral refusal to deal and the essential facilities 11 doctrine. The other theories are futile. 12 hiQ shall have four weeks to file an amended complaint. LinkedIn shall have four weeks 13 thereafter to respond to the amended complaint. 14 This order disposes of Docket No. 137. 15 16 IT IS SO ORDERED. 17 18 Dated: September 9, 2020 19 20 ______________________________________ EDWARD M. CHEN 21 United States District Judge 22 23 24 25 26 27