HIQ LABS INC v. LINKEDIN CORPORATION (2022)

ORDER DENYING PLAINTIFF'S MOTION FOR SUMMARY JUDGMENT; GRANTING IN PART AND DENYING IN PART DEFENDANT'S MOTION FOR SUMMARY JUDGMENT; DENYING DEFENDANT'S MOTIONS TO EXCLUDE; AND GRANTING IN PART DEFENDANT'S MOTION FOR SANCTIONS

Docket Nos. 336-339, 355

I. INTRODUCTION

Before this Court are cross-motions for summary judgment. LinkedIn Corporation (“LinkedIn”) moves for partial summary judgment on its breach of contract claim and on hiQ Labs, Inc.’s (“hiQ”) tort claims. hiQ moves for partial summary judgment on its statute of limitations defense to LinkedIn's claim under the Computer Fraud and Abuse Act (“CFAA”). Additionally, LinkedIn moves to exclude the expert testimony of Stephen McElfresh, to exclude the damages opinions of Benjamin Sacks, and to impose spoliation sanctions on hiQ for the destruction of electronically stored information. For the reasons below, the Court GRANTS in PART and DENIES in PART LinkedIn's motion for partial summary judgment. It DENIES hiQ's motion for summary judgment. The Court DENIES LinkedIn's motions to exclude as moot and GRANTS in PART LinkedIn's motion for sanctions.

II. FACTUAL AND PROCEDURAL BACKGROUND 1

A. LinkedIn

LinkedIn is a professional network, with over 850 million members worldwide. (Docket No. 334-2 (8/5/22 Rockwell Decl.) at ¶ 1.) Members create and manage their profiles on the platform to network with other professionals. (Docket No. 216-14 (9/10/21 Rockwell Decl.) at ¶ 3; Docket No. 320 (LinkedIn's Amended Counterclaims (“Countercl.”)) at ¶ 2.) They populate their profiles with information such as job histories, skills, and educational background. (Countercl. at ¶¶ 2, 22.) LinkedIn allows members to control the information that they choose to publish about themselves to others. (9/10/21 Rockwell Decl. at ¶ 3.) Specifically, LinkedIn members can choose to: (1) keep their profile information private; (2) share only with their direct connections; (3) share with connections within certain degrees of separation; (4) allow access only to “logged-in” users, i.e., other signed-in LinkedIn members, or (5) allow access to everyone, including “logged-out” users, i.e., members of the general public who can access the information without signing in to LinkedIn accounts. (Docket No. 131 (First Amended Complaint (“FAC”)) at ¶ 8; see also Countercl. at ¶¶ 31-34.) This last category of information consists of wholly public profiles. Members also can delete their own profiles permanently, or, using the “Do Not Broadcast” setting, choose not to broadcast to their connections that they have made changes to their profiles. (9/10/21 Rockwell Decl. at ¶ 17; Docket No. 49-1 (7/20/17 Rockwell Decl.) at ¶¶ 2-10.)

LinkedIn contends that scraping—a process of extracting information from a website using automated means (HCE ² Ex. 61 (Schmidt Rpt.) at ¶ 57)—is bad for it and its members: scraping burdens LinkedIn's servers, inhibiting the site's performance (LCE 0008-09 (Bray Dep. Tr.); LCE 0461-65 (Malackowski Rpt.)); scrapers may retain and sell members’ deleted information, interfering with members’ control over or expectations regarding their information (LCE 0947-50 (Rockwell Dep. Tr.)). LinkedIn's User Agreement during the relevant time period thus prohibits scraping. (Docket No. 336-1 (“Lawit Decl.”) Ex. C (“User Agreement”) § 8.2.) The User Agreement seeks to govern both registered users and visitors. (Id. at § 1.2 (“[B]y ․ accessing or using our services ․ you are entering into a legally binding agreement.”).)

LinkedIn dedicates a team to prevent scraping on LinkedIn's platform and deploys mechanisms to block data scrapers except for those with LinkedIn's express permission to do so, such as the Google search engine. (Docket No. 29 (6/24/17 Rockwell Decl.) at ¶¶ 4, 12.) One such mechanism is LinkedIn's “Anti-Abuse Infrastructure” and “Anti-Scraping Technical Strategy” that is effective at blocking scrapers without regard to or knowledge of their identities. (LCE 0010-12 (Bray Dep.); LCE 1298-1302 (Wu Dep.); LCE 1313-15 (Rockwell Decl.); LCE 0810-11 (Murphy Rpt.).) LinkedIn also uses a “robots.txt” file—a technical protocol providing instructions to automated technologies visiting LinkedIn's site. (6/24/17 Rockwell Decl. at ¶ 12.) Those instructions prohibit unauthorized access and warn that use of bots to access LinkedIn without express permission is strictly prohibited. (Id.; Countercl. at ¶ 33.) Entities seeking to crawl LinkedIn apply for permission and, if allowed, must abide by the terms of LinkedIn's robots.txt and LinkedIn Crawling Terms and Conditions. (LCE 0812 (Murphy Rpt.).)

B. hiQ

Founded in 2012 and dormant by 2019, hiQ was a “people analytics” company that provided information to businesses about their workforces. (FAC at ¶¶ 33, 34; Docket No. 23-4 (6/22/17 Weidick Decl.) at ¶ 3; Docket No. 219-2 (9/24/21 Weidick Decl.) at ¶ 6.) It offered two products: “Keeper” and “Skill Mapper.” Keeper analyzed and predicted the retention risk for the employees of a given employer, and indicated which employees were at greatest risk of being recruited away. Employers could then develop action plans to retain its talent. (6/22/17 Weidick Decl. at ¶ 5.) “Skill Mapper” aggregated and summarized the breadth and depth of the skills possessed by an employer's workforce by analyzing all of the skills its employees listed in their LinkedIn profile, including skills acquired from previous positions. Using this information, employers could build succession plans, drive employee engagement, promote internal mobility, and reduce costs associated with external talent acquisition. (Id. at ¶ 6.)

hiQ relied on LinkedIn for its data primarily by scraping wholly public LinkedIn profiles using automated software. (FAC at ¶ 34; Docket No. 335-4 (Schmidt Decl.) Ex. A at 107–10, Table 3.) hiQ had continuously attempted to circumvent LinkedIn's general technical defenses since May 2014. (Schmidt Decl. Ex. A at 42; LCE 0729 (randomly varied time delay); LCE 0703-04 (hiQ CTO Miller: “[w]hen we started to get blocked ․ we started doing business with companies that provide libraries of IP addresses”).) It experimented and attempted to reverse engineer LinkedIn's systems and to avoid detection by simulating human site-access behaviors. (LCE 0706, 0714-16 (Miller Dep.); LCE 1601-03 (notes on reverse engineering LinkedIn's systems); LCE 0730 (make requests in small sets; change user agent and cookies); LCE 0737-48 (Scraper Wars Presentation).) hiQ also hired independent contractors known as “turkers” to conduct quality assurance while “logged-in” to LinkedIn by viewing and confirming hiQ customers’ employees’ identities manually. (HCE Ex. 102 (turker training document).) When LinkedIn's general defenses restricted the turkers’ real accounts, hiQ instructed them to create fake ones. (LCE 0148-49 (internal hiQ email).)

hiQ has known that LinkedIn prohibits scraping since at least 2015. (LCE 0927 (email to hiQ then-CEO excerpting relevant portion of LinkedIn's Use Agreement); LCE 0106-08 (Graves Dep.).) It accepted LinkedIn's User Agreement in running advertising and signing up for LinkedIn subscriptions. (Lawit Decl. ¶¶ 7-8, Exh. J-M; LCE 0118-19 (Graves Dep.); LCE 0615 (Medeiros Dep.); Docket No. 327 at 15, 39 (hiQ Answer to Countercl.).) hiQ also was a LinkedIn Member with a Company Page. (Docket No. 24 (hiQ Br.) at 12; Docket No. 327 at 15, 39 (hiQ Answer to Countercl.) at 14; Lawit Decl. ¶ 5, Ex. I (record of hiQ company page).)

C. Pre-Litigation History And The Filing Of This Action

hiQ organized annual “Elevate” conferences for participants to share insights and best practices in the people analytics field, as well as to market itself. (HCE Exs. 2-5, 14, 15.) Some LinkedIn employees started discussing attending the Elevate conference as early as October 2014. (HCE Ex. 91.) Between 2015 and 2017, LinkedIn employees attended, spoke at, and received awards at these conferences. (HCE Ex. 81 (Reid Tr. 46:5-54:11); HCE Ex. 55 (“LinkedIn Attendees at hiQ's Elevate Conferences”); LCE 0062-63 (LinkedIn employee notes from conference); HCE Ex. 75 (Jennings Dep. Tr.) at 65:14-66.3; HCE Ex. 56 (awards to LinkedIn employees).)

hiQ began developing Skill Mapper in 2016 and planned to demonstrate it at the April 20, 2017 Elevate conference. (HCE Ex. 74 (Graves Dep. Tr.) at 80:21-81:9.) It, however, had trouble scraping LinkedIn public profiles due to LinkedIn's general technical defenses. (LCE 1174-75 (4/5/17 CEO Weidick email reporting technical and legal issues with scraping); LCE 1258 (Weidick email noting that the “first order of business is to get scraping up and running to some viable place.”).) hiQ was seeing ban rates of 99% or more. (LCE 0051 (ban rates between 99%-100%); LCE 0770 (reporting ban rate greater than 99%).) During this time, hiQ's data scientist anonymously requested access to LinkedIn's APIs, apparently on its CEO's instruction, to “test[ ] the water.” (LCE 1608 (Weidick email); LCE 0039-44 (Dev Dep.); LCE 1165-66 (Weidick Dep.).) LinkedIn rejected the request. (LCE 0049-50.)

hiQ also experienced financial strains at that time. It had less than six months of cash left before May 2017. (LCE 1150–51, 1161.) It had trouble renewing its existing customers—a challenge that LinkedIn experienced also with its own product that competed with hiQ's. (LCE 0673–76, 1288.) On May 15, 2017, hiQ's Board of Directors met to consider a complete reexamination of its business model. (LCE 1220–53.) hiQ, however, maintains that it was a thriving startup on the cusp of its third and largest round of investor funding, “with more than a dozen Fortune 500 clients, 23 employees, and a first-mover advantage in people analytics.” (Docket No. 365-3 (“hiQ Opp.”) at 1.)

Around 2016 and 2017, LinkedIn was developing an offering known as “Talent Insights” similar to hiQ's Skill Mapper. In October 2016, LinkedIn's Director of Analytics and Business Development, Lorenzo Canlas, circulated screenshots of and notes on hiQ's products—including the then un-announced Skill Mapper—to the LinkedIn Talent Insights development team. (HCE Ex. 59.) Canlas forwarded that team his invitation to hiQ's 2017 Elevate Conference, reminded them to attend, and noted that “hiQ will be unveiling their new skills mapped [sic] product if you want to get some intel on similar products to [Talent Insights].” (HCE Ex. 27.) Before the conference, LinkedIn internally identified hiQ as an “initial direct competitor” to Talent Insights. (HCE Ex. 32.) After attending the conference, two Talents Insight team members shared their notes on hiQ and noted that Skill Mapper “[wa]s powered mostly by scraped LinkedIn data.” (HCE Ex. 29.) In response, LinkedIn's Senior Director of Product Marketing, Kate Garvey asked about “LinkedIn's policy on blocking (soon-to-be) competitors from scraping our data.” (Id.) Although LinkedIn generally did not “pursue cases of public profile scraping,” Eric Owski who headed the Talent Insight's team stated that he would “see if we push a bit harder.” (Id.) Ms. Garvey responded that she was “a big fan of making my job easier by knocking out the competition.” (Id.)

About a month after that email, on May 23, 2017, LinkedIn sent hiQ a cease-and-desist letter that threatened action under the CFAA, the Digital Millennium Copyright Act (“DMCA”), California Penal Code § 502(c), and the California common law of trespass. (HCE Ex. 17.) The letter asserted hiQ's unauthorized scraping of LinkedIn's profiles violated the law. (Id.) It advised hiQ LinkedIn had “restricted” hiQ's company page and “ha[d] implemented technical measures to prevent hiQ from accessing, and assisting others to access, LinkedIn's site, through systems that detect, monitor, and block scraping activity.” (Id.) According to hiQ, these actions by LinkedIn were part of a larger plan to “collect competitive intel from hiQ then shut down its business,” once LinkedIn identified hiQ as a competitor. (hiQ Opp. at 5.)

In response to the cease-and-desist letter, hiQ demanded that LinkedIn recognize its right to access LinkedIn's public pages. (FAC at ¶¶ 42-43.) LinkedIn refused. (Id. at ¶ 44.) hiQ subsequently sued LinkedIn, seeking injunctive relief based on California law and a declaratory judgment that LinkedIn could not lawfully invoke the CFAA, the DMCA, California Penal Code § 502(c), or the common law of trespass against it. (Docket No. 1.) Additionally, it asserted against LinkedIn tortious interference claims, claims under the California Unfair Competition Law (“UCL”), promissory estoppel, and violation of right to free speech under the California Constitution. (Id.) hiQ concurrently filed a request for a temporary restraining order that the parties later agreed to convert into a motion for a preliminary injunction. (Docket No. 23.)

D. Preliminary Injunction

On August 14, 2017, this Court granted hiQ's motion for a preliminary injunction. (Docket No. 63.) It credited hiQ's assertion that the survival of its business was threatened absent a preliminary injunction. (Id. at 5.) It determined that the balance of hardships tipped sharply in hiQ's favor because LinkedIn's interest in preventing hiQ from scraping the public profiles of LinkedIn's members did not outweigh hiQ's interest in continuing its business. (Id. at 7–8.) The Court held that hiQ raised at least serious questions going to the merits of its California UCL claims and the applicability of the CFAA, but not the promissory estoppel and free speech claims. (Id. at 16, 20–21, 23.) Finally, the Court determined that the public interest favored hiQ because giving companies like LinkedIn free rein to decide who could collect and use data otherwise available to the public risked the possible creation of information monopolies. (Id. at 24.)

The Court ordered LinkedIn to withdraw its cease-and-desist letter, to remove any existing technical barriers to hiQ's access to LinkedIn members’ public profiles, and to refrain from erecting any legal or technical barriers that block hiQ's access to those profiles. (Id. at 25.)

E. hiQ's Business Post-Injunction

After this Court issued the preliminary injunction, hiQ's business diminished significantly because of, according to hiQ, “the cloud of uncertainty caused by LinkedIn's conduct [that] lingered over the business.” (Docket No. 219-2 (Weidick Decl.) at ¶ 4.) hiQ lost out on funding from investors, all of its employees left, and it could no longer solicit new clients or renew current client contracts. (Id.)

hiQ wound down its operations in 2018, though its servers continued running into 2019 to deliver on client contracts. (Id. at ¶ 6.) hiQ let lapse its accounts with several cloud service providers key to hiQ's business, including Salesforce that stored its client data, and Splunk and Amazon Web Services that stored its scraping data.

F. LinkedIn's Appeal Of The Injunction

LinkedIn appealed this Court's preliminary injunction order in September 2017 and the parties stipulated to stay all interim deadlines. (Docket No. 80.) The Ninth Circuit upheld this Court's order in September 2019. HiQ Labs, Inc. v. LinkedIn Corp., 938 F.3d 985, 1004 (9th Cir. 2019) (“hiQ I”).

LinkedIn then filed a petition for writ of certiorari to the United States Supreme Court. (Docket No. 129.) The Supreme Court granted the petition, vacated the judgment, and remanded this case for further consideration in light of Van Buren v. United States, ––– U.S. ––––, 141 S. Ct. 1648, 210 L.Ed.2d 26 (2021). See LinkedIn Corp. v. hiQ Labs, Inc., ––– U.S. ––––, 141 S. Ct. 2752, 210 L.Ed.2d 902 (2021).

The Ninth Circuit again affirmed the preliminary injunction. See hiQ Labs, Inc. v. LinkedIn Corp., 31 F.4th 1180 (9th Cir. 2022) (“hiQ II”). It concluded that Van Buren reinforced its determination that hiQ has raised serious questions about whether LinkedIn may invoke the CFAA to preempt hiQ's possibly meritorious tortious interference claim under California Unfair Competition Law. Id. at 1180.

G. Post-Preliminary Injunction Procedural History In This Court

On February 14, 2020, three months after the issuance of hiQ I and before LinkedIn petitioned for writ of certiorari, hiQ filed its Amended Complaint. It alleged, inter alia, that “LinkedIn effectively eviscerated hiQ's business,” that hiQ lost most of its employees, was unable to find further investors, and lost major clients, and that it “has largely been forced out of business entirely.” (FAC at ¶¶ 45, 81.) Compared to the original complaint, the FAC removed the promissory estoppel and free speech claims and added antitrust claims under Sections 1 and 2 of the Sherman Act. (Id. at ¶¶ 109-168.)

On April 14, 2021, LinkedIn moved to dismiss certain claims in the FAC. (Docket No. 137.) The Court granted that motion in part. (Docket No. 158.) It denied LinkedIn's motion to dismiss the tortious interference claims as they were not barred by the Noerr-Pennington doctrine or the California Litigation privilege. (Id.) It dismissed the antitrust claims because hiQ failed to allege anticompetitive conduct. (Id.) The Court granted hiQ leave to amend its antitrust claims to the extent they are based on the theories of unilateral refusal to deal and the essential facilities doctrine. (Id.) hiQ subsequently declined to amend the pleadings. (Docket No. 163.)

After LinkedIn learned about hiQ's dormancy, it filed a motion to dissolve the preliminary injunction. (Docket No. 216.) On August 1, 2022, this Court dissolved the preliminary injunction, holding that LinkedIn had established a significant change in facts by showing that hiQ no longer had an ongoing business. (Docket No. 329.)

III. SUMMARY JUDGMENT LEGAL STANDARD

Federal Rule of Civil Procedure 56 provides that a “court shall grant summary judgment [to a moving party] if the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(a). An issue of fact is genuine only if there is sufficient evidence for a reasonable jury to find for the nonmoving party. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248–49, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). “The mere existence of a scintilla of evidence ․ will be insufficient; there must be evidence on which the jury could reasonably find for the [nonmoving party].” Id. at 252, 106 S.Ct. 2505. At the summary judgment stage, evidence must be viewed in the light most favorable to the nonmoving party and all justifiable inferences are to be drawn in the nonmovant's favor. See id. at 255, 106 S.Ct. 2505. Evidence may be presented in a form that is not admissible at trial so long as it could ultimately be capable of being put in admissible form. See Fed. R. Civ. P. 56(c)(2) (“A party may object that the material cited to support or dispute a fact cannot be presented in a form that would be admissible in evidence”); Fonseca v. Sysco Food Servs. of Ariz., Inc., 374 F.3d 840, 846 (9th Cir. 2004) (“Even the declarations that do contain hearsay are admissible for summary judgment purposes because they ‘could be presented in an admissible form at trial.’ ”) (quotation omitted).

Where a defendant moves for summary judgment based on a claim for which the plaintiff bears the burden of proof, the defendant needs only point to the plaintiff's failure “to make a showing sufficient to establish the existence of an element essential to [the plaintiff's] case.” Celotex Corp. v. Catrett, 477 U.S. 317, 322, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986).

IV. LINKEDIN'S MOTION FOR SUMMARY JUDGMENT (DOCKET NO. 336)

LinkedIn seeks partial summary judgment finding hiQ liable for breach of contract. It also seeks summary judgment on hiQ's tort claims, namely, intentional interference with contract, intentional interference with prospective economic advantage, and unfair competition under Business and Professions Code § 17200. LinkedIn argues California's litigation privilege, Cal. Civ. Code § 47(b), bars these claims. Alternatively, LinkedIn argues that hiQ cannot establish the required elements of hiQ's tort claims.

A. Breach Of Contract

LinkedIn moves for partial summary judgment on its breach of contract claim for (1) hiQ's scraping of LinkedIn's site and using the collected data to sell its Keeper and Skill Mapper products, and (2) hiQ's use and for directing “turkers” to make fake accounts and to copy url data as part of hiQ's scraping operation. hiQ only contests the breach and damages elements. Specifically, as to the first accused conduct, hiQ argues that questions of fact remain as to whether the User Agreement is ambiguous and consequently whether hiQ breached that Agreement. For the turkers’ actions, hiQ argues that they did not constitute a breach that resulted in damages and that it was not responsible for them. hiQ also argues that its affirmative defenses bar the breach claim.

1. Liability

a. Scraping and Using Collected Data

LinkedIn's User Agreement expressly prohibits scraping of its site. Section 8 of the User Agreement (“LinkedIn ‘DOs’ and ‘DON'Ts’ ”) states:

8.2 Don'ts. You agree that you will not:

․

• Scrape or copy profiles and information of others through any means (including crawlers, browser plugins and add-ons, and any other technology or manual work);

․

• Use manual or automated software, devices, scripts[,] robots, other means or processes to access, “scrape,” “crawl” or “spider” the Services or any related data or information;

• Use bots or other automated methods to access the Services, add or download contracts, send or redirect messages;

․

As relevant to hiQ's argument, the User Agreement also delineates members’ rights and obligations as follows:

2. Obligations

․

When you share information, others can see, copy and use that information.

․

3.1 Your License to LinkedIn

․

c. We will get your consent if we want to give others the right to publish your posts beyond the Service. However, other Members and/or Visitors may access and share your content and information, consistent with your settings and degree of connection with them.

․

Despite the clear language cited above, hiQ argues that LinkedIn's User Agreement was nonetheless ambiguous because of (1) inconsistent provisions within the Agreement, and (2) extrinsic evidence, including LinkedIn's conduct which suggested that scraping was not categorically barred.

“Whether language in a contract is ambiguous is a question of law to be answered by the court.” Coremetrics, Inc. v. Atomic Park.com, LLC, No. C-04-0222 EMC, 2005 WL 3310093, at *3 (N.D. Cal. Dec. 7, 2005); accord Abifadel v. Cigna Ins. Co., 8 Cal. App. 4th 145, 159, 9 Cal.Rptr.2d 910 (1992). However, extrinsic evidence may be required to determine the intent of the parties to the contract. Abifadel, 8 Cal. App. 4th at 159, 9 Cal.Rptr.2d 910. The usual formula under California law when determining whether to admit parol evidence is as follows:

First, the court provisionally receives (without actually admitting) all credible evidence concerning the parties’ intentions to determine “ambiguity,” whether the language is “reasonably susceptible” to the interpretation urged by a party. If in light of the extrinsic evidence the court decides the language is “reasonably susceptible” to the interpretation urged, the extrinsic evidence is then admitted to aid in the second step—interpreting the contract.

Winet v. Price, 4 Cal. App. 4th 1159, 1165, 6 Cal.Rptr.2d 554 (1992) (citing Blumenfeld v. R. H. Macy & Co., 92 Cal. App. 3d 38, 45, 154 Cal.Rptr. 652 (1979)). The threshold issue of whether the parol evidence is necessary to determine the meaning of the language is a question of law. Id. Similarly, “when the parol evidence is not conflicting,” construction of the contract is a question of law. Id.

User Agreement Provisions. Contrary to hiQ's characterization, the User Agreement's provisions do not conflict with each other. According to hiQ, the User Agreement's statements that “Visitors may access and share your content and information consistent with your settings” and that “[w]hen you share information, others can see, copy and use that information” are inconsistent with the prohibition of scraping data—a means to access and copy LinkedIn members’ information. (User Agreement §§ 2, 3.1.) Thus, hiQ argues that the inconsistency creates a question of fact as to whether its conduct constitutes a breach of the User Agreement.

The Court disagrees. Informing members that their data may be “see[n], cop[ied], and use[d]” does not contradict the prohibition against “scrap[ing], crawl[ing], or spider[ing] the Server.” (User Agreement §§ 2, 8.2.) The two concepts are not mutually exclusive—a warning to members that a third party may collect their public-facing data is not a blessing for third parties to do so through expressly prohibited means. Thus, the contract's language itself does not create ambiguity within the User Agreement.

Extrinsic Evidence. hiQ's extrinsic evidence includes an internal email between LinkedIn employees stating that LinkedIn “generally do[es]n't pursue cases of public scraping,” the fact that LinkedIn scraped its competitor's websites, and the fact that LinkedIn's parent company scraped LinkedIn. (hiQ Opp. at 13 (citing HCE Ex. 29, 80, 47).) Also, hiQ argues that LinkedIn's non-enforcement implied that the agreement does not bar scraping.

However, “parol evidence is admissible only to prove a meaning to which the language is ‘reasonably susceptible,’ not to flatly contradict the express terms of the agreement.” Winet, 4 Cal. App. 4th at 1167, 6 Cal.Rptr.2d 554 (emphasis added) (citations omitted). hiQ seeks to introduce parol evidence to prove that it may “[u]se manual or automated software, devices, scripts[,] robots ․ to access, ‘scrape,’ ‘crawl’ or ‘spider’ the [LinkedIn] Services” even though the User Agreement says hiQ “agree[d] that [it] will not” do so. (User Agreement § 8.2.) That extrinsic evidence does not negate or diminish the express terms of the User Agreement. Furthermore, LinkedIn's failure to abide by or enforce the Agreement, which perhaps gives rise to an affirmative defense, does not contradict or render ambiguous the unambiguous terms of the Agreement.

In sum, the relevant language of the User Agreement unambiguously prohibits hiQ's scraping and unauthorized use of the scraped data.

b. Turkers

hiQ argues against liability for the turkers’ conduct because (1) no evidence shows that the turkers ever scraped any profile information, (2) it is not responsible for its independent contractors’ acts, and (3) no evidence shows actual harm.

Undisputed evidence shows that hiQ's turkers registered false LinkedIn identities under its instructions in breach of the User Agreement. Section 8 of the User Agreement (“LinkedIn ‘DOs’ and ‘DON'Ts’ ”) states:

8.2 Don'ts. You agree that you will not:

․

• Create a false identity on LinkedIn;

․

(Id. at § 8.2.) hiQ's turker training document for the Keeper product explicitly instructed, “It is a good idea to make a fake account with a fake email, to deal with the possibility of being banned on LinkedIn.” (LCE 0150-51 (“turk instructions redux”); LCE 0118-19, 0124-25 (Graves Dep. Tr.).) Undisputed evidence suggests that turkers followed this instruction. (See LCE 0124 (Graves Dep. Tr.).) Regardless of whether the turkers scraped LinkedIn's site, they breached the User Agreement's prohibition on creating false identities.

Although turkers were hiQ's independent contractors, hiQ cannot escape liability because they also acted as its agent regarding the log-in process to LinkedIn. See City of Los Angeles v. Meyers Bros. Parking Sys., Inc., 54 Cal. App. 3d 135, 138, 126 Cal.Rptr. 545 (1975) (“[A]n agent may also be an independent contractor.”). “[T]he difference between an agent and an independent contractor is primarily the degree of retained control.” Park v. Burlington N. Santa Fe Ry. Co., 108 Cal. App. 4th 595, 613, 133 Cal.Rptr.2d 757 (2003). “Ordinarily, the question of agency is one of fact,” but “where the evidence is undisputed the issue becomes one of law.” Magnecomp Corp. v. Athene Co., 209 Cal. App. 3d 526, 536, 257 Cal.Rptr. 278 (1989).

Here, undisputed evidence shows that hiQ retained a high degree of control over turkers’ log-in process. Its training documents provided suggestions on how to log into LinkedIn's accounts. (See, e.g., LCE 0151 (“It is a good idea to make a fake account with a fake email, to deal with the possibility of being banned on LinkedIn.”); HCE Ex. 102 (“Log into linkedin.com, you may want to create new accounts for turking to avoid having your account harassed. It is important that you are browsing anonymously.”).) When LinkedIn “cracked down on usage in a way that [wa]s making problems for turking,” hiQ “instruct[ed] turkers on how to proceed.” (LCE 1048 (hiQ email).) hiQ itself considered the turkers their agents. (LCE 0328 (hiQ internal email stating, “by turking we may be violating some of [LinkedIn's] terms of use, such as the requirement that users of LinkedIn don't create false identities and will use their real names on their profiles”) (emphasis added).) hiQ has not pointed to contrary evidence regarding its control over the turkers’ logging in process. There is no dispute then that turkers thus acted as hiQ's agents and their liabilities accrue to it. See Cal. Civ. Code § 2330 (“An agent represents his principal for all purposes within the scope of his actual or ostensible authority, and all the rights and liabilities which would accrue to the agent from transactions within such limit, if they had been entered into on his own account, accrue to the principal.”).

Finally, although hiQ contends LinkedIn suffered no damages from turkers’ actions, “nominal damages are available for breach of contract and can support entry of judgment in favor of a plaintiff who suffered ‘no appreciable harm.’ ” Meta Platforms, Inc. v. BrandTotal Ltd., No. 20-CV-07182-JCS, ––– F.Supp.3d ––––, ––––, 2022 WL 1990225, at *22–23 (N.D. Cal. June 6, 2022) (quoting Elation Sys., Inc. v. Fenn Bridge LLC, 71 Cal. App. 5th 958, 965–66, 286 Cal.Rptr.3d 762 (2021)). hiQ has therefore breached LinkedIn's User Agreement through the turkers’ conduct.

c. Summary

In sum, hiQ breached LinkedIn's User Agreement both through its own scraping of LinkedIn's site and using scraped data, and through turkers’ creation of false identities on LinkedIn's platform.

2. Affirmative Defenses

a. Unclean Hands

hiQ asserts that LinkedIn “dirtied its hands” by bringing “its breach of contract claim as part of its plan to ‘cut off’ hiQ's access to otherwise public data and put hiQ out of business.” (hiQ Opp. at 10 (internal citation omitted).) LinkedIn responds that the alleged misconduct does not directly affect hiQ's obligations under the User Agreement, but merely enforces LinkedIn's bargained-for rights.

In California, “unclean hands is available as an affirmative defense to a contract claim.” Cal-Agrex, Inc. v. Tassell, 258 F.R.D. 340, 348, n.2 (N.D. Cal. 2009) (citing Camp v. Jeffer, Mangels, Butler & Marmaro, 35 Cal. App. 4th 620, 638, 41 Cal.Rptr.2d 329 (1995)). “Whether the particular misconduct is a bar to the alleged claim for relief depends on (1) analogous case law, (2) the nature of the misconduct, and (3) the relationship of the misconduct to the claimed injuries.” E. W. Bank v. Rio Sch. Dist., 235 Cal. App. 4th 742, 751, 185 Cal.Rptr.3d 676 (2015) (citation omitted). hiQ has cited no analogous case law supporting the application of the unclean hands defense to the facts present here. Failing to meet the first prong of the three-part test “alone is sufficient to warrant the denial of the defense.” Id. at 751–52, 185 Cal.Rptr.3d 676 (holding unclean hands does not apply as a matter of law “because there is no analogous case applying the doctrine” to facts there). Therefore, the Court grants summary judgment on hiQ's unclean hands defense in favor of LinkedIn.

b. Waiver

“Under California law, a party to a contract waives a contractual right by ‘indicating an intent to relinquish the right.’ ” Williams v. Apple, Inc., 338 F.R.D. 629, 646 (N.D. Cal. 2021) (quoting Wind Dancer Prod. Grp. v. Walt Disney Pictures, 10 Cal. App. 5th 56, 78, 215 Cal.Rptr.3d 835 (2017)). “The waiver may be either express, based on the words of the waiving party, or implied, based on conduct indicating an intent to relinquish the right.” Wind Dancer, 10 Cal. App. 5th at 78 (internal quotations and citations omitted). “The waiver may be ․ implied ․ when that party's acts are so inconsistent with an intent to enforce the right as to induce a reasonable belief that such right has been relinquished.” Id. Thus, “the pivotal issue in a claim of waiver is the intention of the party who allegedly relinquished the known legal right.” Id. (internal quotations and citations omitted). “Waiver is ordinarily a question of fact unless ‘there are no disputed facts and only one reasonable inference may be drawn.’ ” Id. (quoting DuBeck v. Cal. Physicians’ Serv., 234 Cal. App. 4th 1254, 1265, 184 Cal.Rptr.3d 743 (2015)).

As discussed below for hiQ's summary judgment motion, there is a genuine dispute of material fact as to whether LinkedIn knew about hiQ's scraping and use of scraped data as early as October 2014. If it did, a reasonable jury could find that LinkedIn relinquished its enforcement against hiQ's scraping through its conduct—aside from generally deploying anti-scraping technology against all scrapers, it did not take steps to legally enforce against known scraping by hiQ for years, and it allowed its employees to attend hiQ's conferences. Therefore, summary judgment is improper for the waiver defense for hiQ's scraping and unauthorized data usage.

There is no evidence, however, showing that LinkedIn knew about hiQ's use of turkers before this action, so LinkedIn could not have relinquished its right to enforce the User Agreement's provision prohibiting the creation of false identities. The Court therefore grants summary judgment for the waiver defense on that conduct.

c. Estoppel

The doctrine of estoppel generally requires the following elements: (1) “the party to be estopped must be apprised of the facts;” (2) “he must intend that his conduct shall be acted upon, or must so act that the party asserting the estoppel had a right to believe it was so intended;” (3) “the other party must be ignorant of the true state of facts;” and (4) “he must rely upon the conduct to his injury.” Lentz v. McMahon, 49 Cal. 3d 393, 399, 261 Cal.Rptr. 310, 777 P.2d 83 (1989).

Regarding hiQ's scraping and use of scraped data, there remains a genuine dispute of material fact as to all four elements. As explained in hiQ's summary judgment below, LinkedIn might have known about hiQ's conduct in 2014. Viewed in the light most favorable to hiQ, LinkedIn employees’ participation in hiQ's Elevate conferences where hiQ discussed its products led hiQ to believe that LinkedIn acquiesced in its circumvention of LinkedIn's general defense mechanisms to scrape public profiles. A reasonable jury could find that hiQ relied on such acquiescence in developing its business over the years with near total dependency on LinkedIn. The Court thus denies summary judgment on the estoppel defense regarding hiQ's scraping and use of scraped data.

Regarding hiQ's use of turkers, there is no evidence that LinkedIn was appraised of relevant facts as explained above. The Court therefore grants summary judgment on estoppel for that conduct.

d. Unconscionability

hiQ raises the affirmative defense of unconscionability for the first time in its opposition brief, arguing that the User Agreement is procedurally and substantively unconscionable. LinkedIn responds that (1) hiQ has waived this affirmative defense by failing to plead it in hiQ's answer, and that (2) even if not waived, hiQ cannot establish unconscionability.

LinkedIn has not shown prejudice from hiQ's belated assertion of unconscionability, so hiQ has not waived this affirmative defense. See Camarillo v. McCarthy, 998 F.2d 638, 639 (9th Cir. 1993) (absent prejudice, an affirmative defense may be raised for the first time at summary judgment); Mir v. Levine, 745 F. App'x 726, 727 (9th Cir. 2018) (absent showing of prejudice, no waiver of defendants’ collateral estoppel defense even though it was not raised until summary judgment). The Court therefore analyzes hiQ's defense on the merits.

“Unconscionability has both a procedural and a substantive element, the former focusing on oppression or surprise due to unequal bargaining power, the latter on overly harsh or one-sided results.” Lennar Homes of Cal., Inc. v. Stephens, 232 Cal. App. 4th 673, 687, 181 Cal.Rptr.3d 638 (2014) (internal quotations and citations omitted). “The prevailing view is that procedural and substantive unconscionability must both be present in order for a court to exercise its discretion to refuse to enforce a contract or clause under the doctrine of unconscionability;” however, “they need not be present in the same degree.” Id. (emphasis in original). Rather, courts apply a “sliding scale” to determine unconscionability. Id. “The more substantively oppressive the contract term, the less evidence of procedural unconscionability is required to come to the conclusion that the term is unenforceable, and vice versa.” Id. at 687–88, 181 Cal.Rptr.3d 638 (internal quotations and citations omitted). “Substantive unconscionability addresses the fairness of the term in dispute” and “traditionally involves contract terms that are so one-sided as to ‘shock the conscience,’ or that impose harsh or oppressive terms.” Wherry v. Award, Inc., 192 Cal. App. 4th 1242, 1248, 123 Cal.Rptr.3d 1 (2011) (quoting Szetela v. Discover Bank, 97 Cal. App. 4th 1094, 1100, 118 Cal.Rptr.2d 862 (2002)).

hiQ seeks to establish substantive unconscionability by arguing that LinkedIn's prohibition on “copying profiles” is “incompatible with simply visiting the website.” (hiQ Opp. at 12.) It points to evidence that LinkedIn cannot distinguish between a request for profile data from human users and that from scrapers. (HCE Ex. 71 (Bray 7/27/2022 Dep. Tr.) at 233:24-234:1.) That means, after receiving a request from a web browser, LinkedIn's server returns a copy of the HTML code to human users and scrapers alike. Therefore, hiQ reasons, every visitor of LinkedIn's site violates the prohibition against “copy[ing] profiles and information through ․ any ․ technology.” (User Agreement § 8.2.)

hiQ does not contend that the prohibition against the conduct at issue here—scraping, using scraped data without consent, and creating false identities—is unconscionable as a general matter. Its attack on the conscionability is based on a circumstance not applicable here—merely visiting LinkedIn's site. hiQ would have the Court construe LinkedIn's User Agreement as prohibiting anyone from ever using its own service. No reasonable user would so interpret the User Agreement. hiQ thus fails to offer any evidence supporting substantive unfairness of the User Agreement.

Absent substantive unconscionability, the Court needs not consider procedural unconscionability. The Court grants LinkedIn's motion for summary judgment on this affirmative defense.

e. Other Affirmative Defenses

LinkedIn moves for summary judgment on hiQ's affirmative defenses of in pari delicto, ratification, and consent, and hiQ does not oppose in its response brief. hiQ has thus waived any argument for those affirmative defenses. See Pac. Dawn LLC v. Pritzker, 831 F.3d 1166, 1178 n.7 (9th Cir. 2016) (deeming argument waived where party failed to raise argument in opposition to motion for summary judgment). The Court therefore grants LinkedIn's motion for summary judgment on hiQ's affirmative defenses of in pari delicto, ratification, and consent.

3. Conclusion

In sum, the Court therefore DENIES LinkedIn's motion for summary judgment on the breach of contract claim as to hiQ's scraping and unauthorized use of data because there remains a genuine dispute of material facts for hiQ's waiver and estoppel defenses, but GRANTS the motion as to hiQ turkers’ conduct.

B. California Litigation Privilege

LinkedIn argues that hiQ's tort and UCL claims are barred by Cal. Civ. Code § 47(b) because hiQ's alleged damages were caused by LinkedIn's cease-and-desist letter (“C & D letter”) that was a communication made in anticipation of potential litigation. hiQ responds that the privilege does not apply because the C & D letter was not the sole cause of its damages, but rather only part of LinkedIn's broader anti-competitive scheme to “shut down” hiQ. hiQ further urged this Court to apply the “sham” analysis under the Noerr-Pennington doctrine and find that the “C & D letter was merely the ‘enforcement mechanism ․ [of LinkedIn's] anticompetitive scheme.’ ”

1. Legal Standard

“The essence of the Noerr-Pennington doctrine is that those who petition any department of the government for redress are immune from statutory liability for their petitioning conduct. The doctrine derives from two Supreme Court cases holding that the First Amendment Petition Clause immunizes acts of petitioning the legislature from antitrust liability. The doctrine has since been applied to actions petitioning each of the three branches of government, and has been expanded beyond its original antitrust context.” Theme Promotions, Inc. v. News Am. Mktg. FSI, 546 F.3d 991, 1006-07 (9th Cir. 2008). Noerr-Pennington thus applies to “petitions sent directly to the court in the course of litigation” as well as “conduct incidental to the prosecution of the suit.” Sosa v. DIRECTV, Inc., 437 F.3d 923, 934 (9th Cir. 2006).

Although similar to the Noerr-Pennington doctrine, the California litigation privilege is broader. See Vinson v. Cal. Dep't of Corrs. & Rehab., No. 13-CV-00699-JST, 2014 WL 4594208, at *6 (N.D. Cal. Sept. 15, 2014) (comparing the two privileges, stating that the California litigation privilege is “[a] similar, though broader, rule”). The relevant statute reads in pertinent parts: “A privileged publication or broadcast is one made: [¶] ․ [¶] (b) In any ․ (2) judicial proceeding․” Cal. Civ. Code § 47(b). The California litigation privilege applies to “any communication (1) made in judicial or quasi-judicial proceedings; (2) by litigants or other participants authorized by law; (3) to achieve the objects of the litigation; and (4) that have some connection or logical relation to the action.” Silberg v. Anderson, 50 Cal. 3d 205, 212, 266 Cal.Rptr. 638, 786 P.2d 365 (1990). “[T]here is no ‘sham’ exception to the California litigation privilege comparable to the exception for Noerr-Pennington immunity.” Dairy, LLC v. Milk Moovement, Inc., No. 2:21-CV-02233 WBS AC, 2022 WL 2392622, at *5 (E.D. Cal. July 1, 2022); see also NextG Networks, Inc. v. NewPath Networks, LLC, No. C 08-1565 VRW, 2008 WL 11399757, at *3 (N.D. Cal. Oct. 15, 2008) (noting that lack of “sham exception” to California's litigation privilege is “unsurprising given the California Supreme Court's characterization of the privilege as absolute”).

California courts have given § 47(b) “expansive application” and “the privilege has been extended to any communication, whether or not it is a publication, and to all torts other than malicious prosecution.” Edwards v. Centex Real Est. Corp., 53 Cal. App. 4th 15, 29, 61 Cal.Rptr.2d 518 (1997) (emphasis in original). And when applicable, the privilege is “absolute” regardless of the communicator's “motives, morals, ethics, or intent.” Silberg, 50 Cal. 3d at 220, 266 Cal.Rptr. 638, 786 P.2d 365. “Any doubt about whether the privilege applies is resolved in favor of applying.” Kashian v. Harriman, 98 Cal. App. 4th 892, 913, 120 Cal.Rptr.2d 576 (2002).

Prelitigation communications may fall within the scope of the absolute privilege. Edwards, 53 Cal. App. 4th at 30, 61 Cal.Rptr.2d 518 (“[C]ourts have applied the judicial privilege to certain discrete categories of communications made in advance of actual litigation.”). However, such statements must meet additional requirements. Specifically, the “communication must have some relation to an imminent lawsuit or judicial proceeding which is actually contemplated seriously and in good faith to resolve a dispute, and not simply as a tactical ploy to negotiate a bargain.” Id. at 36, 61 Cal.Rptr.2d 518 (emphasis in original) (citing Silberg 50 Cal. 3d at 212–14, 266 Cal.Rptr. 638, 786 P.2d 365). The “mere potential or ‘bare possibility’ that judicial proceedings ‘might be instituted’ in the future is insufficient to invoke the litigation privilege.” Id. (quoting Rest.2d Torts, §§ 586–588, com. e, pp. 247–251); see also Action Apartment Assn., Inc. v. City of Santa Monica, 41 Cal. 4th 1232, 1251, 63 Cal.Rptr.3d 398, 163 P.3d 89 (2007). Whether a party acts in good faith is a question of fact. Action Apartment, 41 Cal. 4th at 1251, 63 Cal.Rptr.3d 398, 163 P.3d 89. However, “if there is no dispute as to the operative facts, the application of the litigation privilege is a question of law.” Kashian, 98 Cal. App. 4th at 913, 120 Cal.Rptr.2d 576.

2. The Good Faith Requirement

Because the C & D letter herein was prelitigation communications, “the factual question of whether [such] prelitigation communication [were] made in good faith must be resolved prior to the application of the privilege.” Intermarketing Media, LLC v. Barlow, No. 820CV00889JLSDFMX, 2021 WL 5990190, at *4 (C.D. Cal. May 4, 2021); see also Edwards, 53 Cal. App. 4th at 35 n.10, 61 Cal.Rptr.2d 518 (noting that even in the case of the “classic example” of prelitigation communication, i.e., “the threat of an attorney to file suit if a claim is not settled,” the factual determination as to good faith must precede application of the privilege). Importantly, under the California litigation privilege the communicator's “motives, morals, ethics, or intent” are irrelevant, so long as there was a good faith intent to pursue legal action. Silberg, 50 Cal. 3d at 220, 266 Cal.Rptr. 638, 786 P.2d 365. Thus, so long as LinkedIn had an intent to bring forth litigation when it sent the C & D letter, the underlying motivation in doing so is irrelevant, i.e., any anti-competitive motivations are irrelevant.

hiQ has not presented any evidence that LinkedIn sent the C & D letter with bad faith intent to not commence litigation. In fact, the evidence suggests the contrary. Specifically, the C & D letter was sent by LinkedIn's attorneys who contemplated and understood the legal implications of the letter and the repercussions of any subsequent legal action. (See Docket No. 336-2 at 2 (Bajoria Decl.) (stating that he “determined that there appeared to be a sufficient factual and legal basis to send a cease-and-desist letter ․” and “believed the assertions of that letter were both factually and legally justified.”); see also Lawit Decl. at 8 (stating that “[w]hen sending a cease-and-desist letter, we hope that litigation will not be necessary but we always know that litigation is a possibility and, accordingly, make sure that we have a legal basis to proceed if the factual allegations prove to be true and we cannot reach an acceptable outcome ․”).) Additionally, the record indicates that LinkedIn has initiated litigation after sending C & D letters in the past, supporting the notion that litigation was seriously contemplated when sending the C & D letter. (See Lawit Decl. at 9 (indicating that LinkedIn has subsequently brought suit six times after sending a C & D letter).) Certainly, LinkedIn's vigorous litigation herein underscores the seriousness of its intent to pursue its rights. These facts indicate more than a “bare possibility” that a judicial proceeding “might be instituted” in the future. Therefore, no reasonable jury could find that litigation was contemplated in bad faith. LinkedIn has thus satisfied the prelitigation requirement under Cal. Civ. Code § 47.

3. Nature of LinkedIn's Conduct

“[B]ecause the litigation privilege applies only to communications, the ‘threshold issue’ is whether [LinkedIn's] conduct was communicative”—it was. Mireskandari v. Gallagher, 59 Cal. App. 5th 346, 368, 273 Cal.Rptr.3d 371 (2020) (citations omitted). “ ‘The distinction between communicative and noncommunicative conduct hinges on the gravamen of the action ․ [T]he key in determining whether the privilege applies is whether the injury allegedly resulted from an act that was communicative in its essential nature.’ ” Mireskandari v. Gallagher, 59 Cal. App. 5th 346, 369, 273 Cal.Rptr.3d 371 (2020) (quoting Rusheen v. Cohen, 37 Cal. 4th 1048, 1058, 39 Cal.Rptr.3d 516, 128 P.3d 713 (2006)). Importantly, “if the gravamen of the action is communicative, the litigation privilege extends to noncommunicative acts that are necessarily related to the communicative conduct[.]” Id. (quotation omitted). “Stated another way, unless it is demonstrated that an independent, noncommunicative, wrongful act was the gravamen of the action, the litigation privilege applies.” Rusheen, 37 Cal. 4th at 1065, 39 Cal.Rptr.3d 516, 128 P.3d 713.

hiQ argues that C & D letter was not the sole gravamen of its complaint. It claims that the C & D was merely part of LinkedIn's broader anti-competitive scheme to “shut down” hiQ. In its view, LinkedIn “gathered competitive intelligence on hiQ's customers, partners, and product features” before ultimately “driv[ing] hiQ out of the market.” (hiQ Opp. at 1.) But the facts show that hiQ's alleged damages hinge entirely on the C & D letter. Specifically, when asked what “monetary remedies” hiQ was seeking in the action against LinkedIn, hiQ's CEO and 30(b)(6) witness, Mr. Mark Weidick, stated that the company had a massive milestone and “opportunity to be a billion dollar business” but that “stopped after [LinkedIn's] accusations and threats were put in the market from the cease and desist letter.” (LCE 1207–08 (Weidick Dep. Tr.).) Further, hiQ's damages expert, Mr. Benjamin Sacks, stated that his damages analysis stops on the day the C & D letter was sent because he “assumed that there was no value in the company” after that date. (LCE 0998–1004 (Sacks Dep. Tr.); see also HCE Ex. 90 (Sacks Dep. Tr. at 245:20-247:10); HCE Ex. 88 (Sacks Second Amended Expert Report) at 2 (“I am instructed to assume that the C & D interfered with hiQ's existing and prospective contracts and business relationships, effectively destroying hiQ's business[.]”).) Such evidence indicates that hiQ's damages resulted from the C & D letter and that the letter is the gravamen of the complaint.

Because the C & D letter was the gravamen of hiQ's complaint, it follows that the four elements of the privilege are satisfied, barring hiQ's tortious interference and UCL claims. Specifically, the C & D letter was a communication made in anticipation of litigation, i.e., a judicial proceeding. See Edwards, 53 Cal. App. 4th at 35 n.10, 61 Cal.Rptr.2d 518 (“The classic example of an instance in which the privilege would attach to prelitigation communications is the attorney demand letter threatening to file a lawsuit if a claim is not settled.”). The letter was sent by attorneys on behalf of LinkedIn. See Silberg, 50 Cal. 3d at 219, 266 Cal.Rptr. 638, 786 P.2d 365 (holding that “[d]efendant's statements ․ suitability were made by a participant, i.e., the attorney for a party”). The object of the letter was to assert rights. And the letter bears “some relation” to the action (indeed it was the catalyst for this action) as it sparked hiQ's lawsuit against LinkedIn as well set forth the claims with which LinkedIn has now asserted against hiQ.

4. Conclusion

Having found that the California litigation privilege applies, the Court GRANTS summary judgment for LinkedIn on hiQ's UCL and tortious interference claims. LinkedIn's motions to exclude hiQ's expert opinions supporting those claims are therefore DENIED as moot.

V. HIQ'S MOTION FOR SUMMARY JUDGMENT (DOCKET NO. 355)

hiQ moves for summary judgment on LinkedIn's claim under the CFAA because it is time-barred by the statute's two-year statute of limitations. Based on two email chains ³ among LinkedIn employees, hiQ argues that LinkedIn had reasonable notice of hiQ's scraping activity in 2014—more than two years before the filing of this action on June 7, 2017. LinkedIn disputes whether the employees in question had notice and, even if they did, whether their knowledge can be imputed to LinkedIn.

A. Legal Standard

A claim under the CFAA is timely if brought “within 2 years of the date of the act complained of or the date of the discovery of the damage.” 18 U.S.C. § 1030(g). “When ‘legislators have written the word “discovery” directly into the statute ․ state and federal courts have typically interpreted the word to refer not only to actual discovery, but also to the hypothetical discovery of facts a reasonably diligent plaintiff would know.’ ” West v. Ronquillo-Morgan, No. CV 20-2711 DSF (EX), 2021 WL 2953160, at *3 (C.D. Cal. May 10, 2021) (quoting Merck & Co. v. Reynolds, 559 U.S. 633, 645, 130 S.Ct. 1784, 176 L.Ed.2d 582 (2010)); see also Maddalena v. Toole, No. 2:13-CV-4873-ODW, 2013 WL 5491869, at *4 (C.D. Cal. Oct. 1, 2013) (“Like many statutes of limitation, the statute[ ] at issue in this action do[es] not require that the claimant have actual knowledge of the violation. Rather, ․ 1030(g) demands only that the claimant have had a reasonable notice to discover the violation.”).

Because LinkedIn has the burden of proof at trial to establish that it is entitled to the benefit of the discovery rule, “to defeat summary judgment [it is] required to come forward with evidence establishing a triable issue of fact with regard to whether the discovery rule applies.” O'Connor v. Boeing N. Am., Inc., 311 F.3d 1139, 1150 (9th Cir. 2002) (citation omitted). Summary judgment is improper “unless the only reasonable inference that can be drawn is that [LinkedIn] knew or should have known” its damage from hiQ's scraping before June 7, 2015. Id.; 18 U.S.C. § 1030(g).

“A two-part analysis determines whether [LinkedIn] reasonably should have known of [its] claim.” O'Connor, 311 F.3d at 1150 (citations omitted). First, courts consider “whether a reasonable person in [LinkedIn's] situation would have been expected to inquire about the cause of [its] injury.” Id. Second, if LinkedIn was on inquiry notice, courts must next determine whether an inquiry “would have disclosed the nature and cause of [LinkedIn's] injury so as to put [it] on notice of [its] claim.” Id. LinkedIn is “charged with knowledge of facts that [it] would have discovered through inquiry.” Id. (citations omitted).

B. The October 2014 Email Chain

On October 8, 2014, several LinkedIn employees exchanged emails regarding an upcoming hiQ Elevate conference. All the LinkedIn employees on the email chain were members of the Talent Analytics Team that supported LinkedIn's Human Resources group. In the chain, one employee, Daniel Maurath, initially emailed the Talent Analytics Team's group alias address and asked if anyone had heard of hiQ. (Docket No. 356 Ex. C.) Another employee, William Gaker, responded, “This looks like a sales conference for a company presenting a new product. I would love to see what data they are using. It sounds like they might be scraping our site to see who has updated profiles and using that as a signal to predict turnover.” (Id.) He later followed up, “It might be good for someone (or at least alert someone who handles how our data gets used) to go to make sure this vendor isn't accessing our data without our permission.” (Id.) The email chain concluded with Mr. Maurath agreeing to “request to attend” hiQ's conference and “at least find out more info.” (Docket No. 356 Ex. D.) However, no one attended the conference, investigated hiQ, or reported hiQ to LinkedIn's legal department. (Docket No. 358-3 Ex. 33 (Lawit Dep. Tr. (8/26/22)) at 72:18-73:3; Docket No. 358-1 (Maurath Decl.) at ¶ 4; Docket No. 358-2 (Rigano Decl.) at ¶ 4.)

No matter what the employees on the email chain knew, a genuine dispute of material fact remains regarding whether their knowledge can be imputed to LinkedIn. In the employment context, “[k]nowledge acquired by an agent while acting within the course and scope of his employment is chargeable to his principal, his employer.” Mountain Copper Co. v. Welcome Growers Gin Co., 197 Cal. App. 2d 253, 256, 17 Cal.Rptr. 355 (1961). The parties agree that the employee must have a duty to disclose that knowledge. (Docket No. 360-2 (“LinkedIn Opp.”) at 22; Docket No. 385-3 (“hiQ Reply”) at 4 (“[T]he question is whether the facts were acquired within the scope and course of employment, and accordingly, whether the LinkedIn employees who concededly had knowledge of, or suspected, hiQ's scraping activity had a duty to disclose that knowledge.”) (emphasis added).) See Cal. Civ. Code § 2332 (“As against a principal, both principal and agent are deemed to have notice of whatever either has notice of, and ought, in good faith and the exercise of ordinary care and diligence, to communicate to the other.”); Triple A Mgmt. Co., Inc. v. Frisone, 69 Cal. App. 4th 520, 534–35, 81 Cal.Rptr.2d 669 (1999) (“The basis for imputing knowledge to the principal is that the agent has a legal duty to disclose information obtained in the course of the agency and material to the subject matter of the agency, and the agent will be presumed to have fulfilled this duty.”); United States v. Georgia-Pacific Co., 421 F.2d 92, 98 n.9 (9th Cir. 1970) (“[A] principal is bound by the knowledge of its agent concerning a matter upon which it is the agent's duty to give the principal information.”) (citing Restatement, Agency 2d § 272).

The parties dispute whether LinkedIn's employees had a duty to disclose their knowledge of hiQ's suspected scraping to LinkedIn, and that is a factual dispute for the jury. See Maron v. Swig, 115 Cal. App. 2d 87, 90, 251 P.2d 770 (1952) (holding whether it was within the scope of agency to take certain actions was a question of fact for the jury).

On one hand, the LinkedIn employees did not seem to explicitly have anti-scraping responsibilities. Two employees on the October 2014 email chain submitted declarations stating that their “work supported only LinkedIn's own hiring and retention efforts,” that and “did not support or work on LinkedIn's external facing products.” (Maurath Decl. at ¶ 3; Docket No. 358-2 (Rigano Decl.) at ¶ 3.)

On the other hand, a reasonable factfinder could find that the LinkedIn employees implicitly had a duty to report suspected scraping, especially when LinkedIn appears to have no functioning anti-scraping program or enforcement team as of October 2014. (See LinkedIn's Supp. Resp. to Interrogatory 8 (“When hiQ was subsequently identified by email to Scraper Council in October 2015, at the time the cross-functional group was still working out its internal processes and there were not yet clear lines of responsibility or a playbook on how to proceed with reports made to the Scraper Council.”); Docket No. 386 Ex. D (LinkedIn April 2015 slide deck on third party scraping stating that it was “investing very little into stopping 3rd party scraping” and that the team working on scraping prevention consisted of “1 data scientist, 1 infra engineer, 0.25 of a person in legal, and very minimal support from TnS/BD”).) Given such limited resources, it is reasonable to infer that employees not explicitly designated to report scraping nonetheless had such a responsibility. A factfinder may also infer such a duty from the fact that Mr. Maurath volunteered to “request to attend [hiQ's conference]” when Mr. Gaker recommended someone “to go to make sure [hiQ] isn't accessing [LinkedIn's] data without [its] permission,” even though no one did so. (Docket No. 356 Exs. C, D; Lawit Dep. (8/26/22) at 72:18-73:3; Maurath Decl. ¶ 4.)

Because a genuine dispute exists as to the scope of employment of the LinkedIn's employees on the October 2014 email chain, a factfinder may draw more than one reasonable inference regarding whether LinkedIn knew or should have known about hiQ's scraping then. Summary judgment in hiQ's favor based on the October 2014 LinkedIn email chain thus would be improper. See Miniace v. Pac. Mar. Ass'n, No. C 04-03506 SI, 2006 WL 648732, at *2 (N.D. Cal. Mar. 13, 2006) (denying summary judgment on statute of limitations defense where, under California agency law, factual questions remained as to whether Director of Human Resources had obligation to report amendment to executives’ life insurance to impute knowledge to corporation).

C. The December 2014 Email Chain

On December 15, 2014, Bob Rosin, LinkedIn's Vice President of Business Development, forwarded an email to Lee Womer, LinkedIn's then Director of Business Development. (Docket No. 356-7 Ex. H at 3.) The email was from a third-party venture capital investor who had looked into hiQ and “[wasn't] sure that LinkedIn would look favorably at a company that is analyzing profiles.” (Id.) In response, Mr. Womer expressed concern that hiQ might not be a “members’ first use case.” (Id. at 2.) He defined “members’ first use” as “one of LinkedIn's corporate values that guides decision making, which is that the interest of LinkedIn members on the platform need to take first priority amongst a range of priorities.” (Docket No. 356-8 Ex. I ⁴ at 55:11-15.)

There is a material factual dispute as to whether the LinkedIn employees on the December 2014 email chain had notice of hiQ's scraping activity. The email did not expressly mention the term “scraping.” And both Mr. Womer and Mr. Rosin testified that they could not specifically remember if they suspected hiQ of scraping when they received the email. (Docket No. 358-3 Ex. 7 (Womer Dep. Tr.) at 54:11-16; Docket No. 356-5 Ex. F (Rosin Dep. Tr.)⁵ at 126:25-127:4.) hiQ argues that a January 12, 2015, email that Mr. Rosin received shows that he knew about third parties scraping LinkedIn and that LinkedIn did not approve of it. (HCE Ex. 22.) That 2015 email involves a discussion about another company that was scraping LinkedIn data, but it makes no mention of hiQ. (Id.) Thus, even if Mr. Rosin knew about third parties scraping LinkedIn generally, there is a genuine dispute as to whether Messrs. Womer and Rosin knew hiQ was “analyzing profiles” through scraping.

Additionally, there is a material factual dispute as to whether Messrs. Womer and Rosin's knowledge was imputable to LinkedIn. Mr. Rosin testifies that he was not involved with anti-scraping in 2014. (Rosin Dep. Tr. at 108:2-109:1, 117:15-118:1, 152:16-25.) And Mr. Womer did not take on responsibility for scraping-related issues until 2015. (Womer Dep. Tr. at 30:11-31:1.) For the same reasons as explained for the October 2014 email chain, it is for the jury to decide whether they had a duty to report suspected scraping to LinkedIn.

Even if Messrs. Womer and Rosin knew hiQ was scraping LinkedIn and their knowledge can be imputed to LinkedIn, there is still a factual dispute as to “whether a reasonable person in [LinkedIn's] situation would have been expected to inquire about the cause of [its] injury.” O'Connor, 311 F.3d at 1150. LinkedIn provided evidence that it received over one hundred recorded reports of scraping every year between 2015 and 2021 and that it “did not have the resources to investigate every single report of potential abuse.” (Lawit Decl. at ¶¶ 14-15.) Mr. Womer testified that there was nothing about the December 2014 email that suggested he should prioritize an investigation into hiQ's potential scraping activity. (Womer Dep. Tr. at 57:13-24.) Thus, the Court cannot conclude as a matter of law that LinkedIn had or should have had notice of hiQ's scraping from the December 2014 email chain.

In sum, because more than one reasonable inference can be drawn as to whether LinkedIn knew or should have known that hiQ scraped LinkedIn's site before June 7, 2015, the Court DENIES hiQ's motion for summary judgment.

VI. LINKEDIN'S MOTION FOR SPOLIATION SANCTIONS (DOCKET NO. 337)

LinkedIn requests evidentiary sanctions applied coincident with its pending summary judgment and Daubert motions, and to the extent necessary thereafter, as jury instructions against hiQ for the destruction of two categories of electronically stored information. Besides monetary sanctions, it seeks (1) various adverse inference jury instructions, and (2) dismissal of all of hiQ's claims that involve its relationships with its customers and prospective, or in the alternative, an order precluding hiQ from offering any evidence or argument attributing its loss of customers or potential customers to any action by LinkedIn, including preclusion of its expert report calculating “business value” based on implied customer relationships.

A. The Evidence At Issue

This sanctions motion concerns the evidence hiQ stored with three cloud service providers—Salesforce, Splunk, and Amazon Web Services (“AWS”). Salesforce hosted hiQ's “Customer Relationship Management Database” or “CRM Database.” Splunk and AWS both hosted hiQ's scraping activity records. Because the data stored in Salesforce CRM Database only relate to hiQ's UCL and tortious interference claims on which the Court has granted motion for summary judgment for LinkedIn, the Court only analyzes data stored with Splunk and AWS.

hiQ issued a company-wide litigation hold notice on June 20, 2017, about a month after LinkedIn's C & D letter and two weeks after hiQ's filing of this action. (LCE Ex. 93.) After that, hiQ's employees “started to leave the business rather quickly.” (Docket No. 368-1 Ex. P (Weidick Dep. Tr.) at 51:11.) In 2018, hiQ had 13 individuals left on the payroll, “barely half the personnel on its payroll the year prior.” (Docket No. 368 at 3.) hiQ went into hibernation that year. (Docket No. 335-3 at 5.) By January 2019, no personnel remained on hiQ's payroll. (Id.) hiQ went in debt. (Id.) At times, hiQ's CEO and its employees put their personal credit cards on accounts to keep the accounts active. (Docket No. 368-1 Ex. P (Weidick 3/8/22 Dep. Tr.) at 23:12-20.)

1. hiQ's AWS Account

hiQ lost the data stored in its AWS account in or around September 2020 after AWS terminated the account earlier that year for non-payment. (Docket No. 368 at 4.) AWS housed a computer database program called MongoDB that stored several collections of hiQ data. (Docket No. 355-3 at 3–4.) The “Raw Scrape” collection contained the actual information hiQ scraped from LinkedIn member profiles. (Id. at 4.) The “scrapus” and “scrapus2” collections (collectively, the “Scrapus Collections”) contained scraping information for each target LinkedIn profile URL, including whether the last request that hiQ's scrapers made on LinkedIn's servers for each profile was successful or blocked. (Id.) The “Proxies Collection” organized data around each IP address that hiQ's scrapers were using to evade LinkedIn's general technical defenses. (Id. at 4–5.) MongoDB also housed data related to hiQ's use of “mechanical turkers”— contractors who logged into LinkedIn's platform, sometimes using fake accounts, to collect profile data not publicly available. (Id.; Docket No. 368 at 4.)

For all the data stored in its AWS account, hiQ archived only the “Raw Scrape” collection in 2018, the year when hiQ started receiving suspension notices from AWS due to non-payment. (Docket No. 368 at 4.) It maintains that the other collections remained in hiQ's AWS account for several years thereafter (id.), but LinkedIn points to hiQ internal communications between 2017 and 2018 referencing plans to affirmatively delete AWS data as part of a migration of its data to save money (Docket No. 335-3 at 6). In March 2020, a month before AWS permanently suspended hiQ's account, hiQ's CEO sent an email to its litigation funder that hiQ was $895,000 in debt, $22,600 of which was owed to AWS. (LCE 1255 (3/2/20 Weidick email).) The email chain mentioned a few potential options to preserve AWS and the fact that hiQ's AWS's account would “shut down and delete immediately” if not paid. (Id.) After AWS's deletion and after LinkedIn sought information from AWS beyond what had been achieved, hiQ, through counsel, attempted to recover the lost Salesforce data without success. (Docket No. 368 at 4.)

2. hiQ's Splunk Account

hiQ lost most of its data stored with Splunk, a company that provided specialized software to hiQ. (Docket No. 335-3 at 7.) Splunk logged and kept for analysis data about every hiQ scrape attempt and the failure rate of those attempts. (Id.) In July 2018, when discussing hiQ's winding down of storage management services, a hiQ software developer noted that hiQ should not delete the 580 gigabytes Splunk data as it could “be important for legal issues.” (Id.) hiQ did not affirmatively delete those data, but did let its Splunk account become inactivated. (Docket No. 335-3 (“The lost data was completely accessible to hiQ until the point at which Splunk discontinued its services to hiQ.”) (citing Devorakonda dep.).) hiQ initially did not disclose to LinkedIn the existence of Splunk when the parties had detailed discussions regarding ESI from August to November 2021. (Id. at 8.) After LinkedIn discovered the existence of Splunk and inquired about it, hiQ “learned that it was unable to access its Splunk account.” (Id.; Docket No. 368 at 4.) Splunk advised hiQ that all of its data would remain available in the event of reactivation, but upon reactivation, hiQ discovered that only less than one month of data in 2020 remained in the account. (Id.)

hiQ has produced over 600 daily Splunk scraping dashboards and other documentation concerning hiQ's scraping efforts. (Docket No. 368 at 22.) The scraping dashboards were created in the normal course of business between November 11, 2016, and July 7, 2020. (Id.) These dashboards showed various statistics relating to hiQ's technology, including the general trend of requests hiQ made to LinkedIn's servers and ban rates of those requests. (Docket No. 368-1 Exs. A-H.) LinkedIn observes that the produced dashboards consist of barely legible line graphs. (Docket No. 387 at 9.) And its expert testifies that it is difficult to discern the precise number of requests made. (Schmidt Decl. at ¶ 13.) Examples of the dashboards are shown below.

(Docket No. 368-1 Ex. B.)

(Id. Ex. C.)

hiQ has also produced communications among its engineers regarding the success rate of hiQ's scraping attempts of LinkedIn during the relevant time period. (See LCE 0051 (“Ban rate continues to fluctuate between .90 and 1.0. Often[,] we see ban rate stick at close to 1.0 for several days in a row.”), 0770 (“we can no longer scan LinkedIn, ban rate is > 99%.”), 1173 (“I am shutting down scraping for the time being. We're only getting a trickle of data, and we could be in danger of hitting our proxies too hard and creating a situation.”), 1174 (“Need path to scrape [LinkedIn] better & faster (e.g.[,] from 100K per day to 500K per day).”), 1258 (“The first order of business is to get scraping back up and running to some viable pace.”), 1259 (“The ban rate has increased to the point where it [is] no longer working again.”), 1616 (“unfortunately, Luminati has told me they can't deliver any help for ~2 weeks.”), 1626–27 (discussing scraping LinkedIn via proxy).)

B. Legal Standard

Federal Rule of Civil Procedure 37(e), as amended in 2015, provides in full as follows:

(e) Failure to Preserve Electronically Stored Information. If electronically stored information that should have been preserved in the anticipation or conduct of litigation is lost because a party failed to take reasonable steps to preserve it, and it cannot be restored or replaced through additional discovery, the court:

(1) upon finding prejudice to another party from loss of the information, may order measures no greater than necessary to cure the prejudice; or

(2) only upon finding that the party acted with the intent to deprive another party of the information's use in the litigation may:

(A) presume that the lost information was unfavorable to the party;

(B) instruct the jury that it may or must presume the information was unfavorable to the party; or

(C) dismiss the action or enter a default judgment.

Fed. R. Civ. P. 37(e).

Rule 37(e) therefore provides two levels of sanctions. Under Rule 37(e)(1), upon finding that the loss of ESI that the offending party had a duty to preserve has prejudiced the moving party, a court “may order measures no greater than necessary to cure the prejudice.” If the court further finds that the offending party “acted with the intent to deprive another party of the information's use in the litigation,” the court may presume (or instruct the jury to presume) that the lost information was unfavorable to the offending party, dismiss the action, or enter a default judgment. Fed. R. Civ. P. 37(e)(2).

The Advisory Committee Notes emphasize that the “remedy should fit the wrong.” Fed. R. Civ. P. 37(e) advisory committee's note to 2015 amendment. A finding of breach and prejudice under Rule 37(e)(1) is sufficient for the court to “allow[ ] the parties to present evidence to the jury concerning the loss and likely relevance of information and instructing the jury that it may consider that evidence, along with all the other evidence in the case, in making its decision,” as long as this is “no greater than necessary to cure prejudice.” Id.

C. Analysis

hiQ does not dispute that it had a duty to preserve the ESI evidence at issue here. (Docket No. 368 at 3 (“hiQ understood that data on its MongoDB and AWS accounts should be preserved for this litigation.”); id. at 4 (“Splunk data should not be deleted as it ‘can be important for legal issues.’ ”) (quoting hiQ employee); id. at 5 (“hiQ understood that it should maintain the data from its Salesforce account for the purposes of litigation.”).)

Nor does hiQ dispute that it has permanently lost the evidence at issue. (Docket No. 368-1 Ex. Z (Weidick Decl.) at ¶ 3 (“With respect to AWS and Salesforce, despite our best efforts, we were not able to recover any data from those accounts and were advised that the data had been permanently deleted after hiQ's accounts had been closed due to non-payment.”), ¶ 4 (“[O]nly a subset of the data still existed in hiQ's Splunk account, and that older data was no longer available.”).)

On the merits, the Court therefore analyzes (1) whether hiQ “acted with the intent to deprive [LinkedIn] of the information's use in the litigation” under Federal Rule of Civil Procedures 37(e)(2), (2) whether the loss of the evidence prejudiced LinkedIn under Rule 37(e)(1), and (3) what sanctions, if any, are appropriate.

1. There Is Insufficient Evidence To Find hiQ Acted With The Requisite Intent To Warrant Rule 37(e)(2) Sanctions

Unlike Rule 37(e)(1), Rule 37(e)(2) “does not include a requirement that the court find prejudice to the party deprived of the information.” Fed. R. Civ. P. 37(e) advisory committee's note to 2015 amendment. “This is because the finding of intent required by the subdivision can support not only an inference that the lost information was unfavorable to the party that intentionally destroyed it, but also an inference that the opposing party was prejudiced by the loss of information that would have favored its position.” Id.

a. A Knowing Failure To Preserve Per Se Does Not Establish An “Intent To Deprive”

The parties do not dispute that hiQ knowingly failed to preserve the evidence at issue, but disagree whether that amounts to an “intent to deprive [LinkedIn] of the information's use in the litigation.” Fed. R. Civ. P. 37(e)(2). Rule 37(e)(2) requirement of a specific “intent to deprive” may be inferred from a knowing failure to preserve when combined with other circumstances.

LinkedIn argues that a knowing failure to preserve qualifies as an “intent to deprive” per se, primarily relying on Fourth Dimension Software v. DER Touristik Deutschland GmbH, No. 19-cv-05561-CRB, 2021 WL 5919821, at *10 (N.D. Cal. Dec. 15, 2021). In that case, the court cited the standard that “a party's deletion of information qualifies as intentional ‘if the party has some notice that the documents were potentially relevant to the litigation before they were destroyed.’ ” Id. (quoting Leon v. IDX Sys. Corp., 464 F.3d 951, 959 (9th Cir. 2006)). But the quoted authority, Leon, concerns willful spoliation under a court's inherent authority. 464 F.3d at 958. The Ninth Circuit has indicated that Rule 37(e) “foreclose[s] reliance on inherent authority.” Newberry v. Cty. of San Bernardino, 750 F. App'x 534, 537 (9th Cir. 2018) (quoting Fed. R. Civ. P. 37 Advisory Committee Notes to the 2015 Amendment); but see Meta Platforms, ––– F.Supp.3d at ––––, 2022 WL 1990225, at *7 (holding party's negligence in failing to preserve ESI could support adverse inference instruction under Court's inherent authority). And the Fourth Dimensions court did not seem to have actually relied on the Leon standard. Instead, the court found the plaintiff to have acted with an intent to deprive under Rule 37(e)(2) “based on the timing and circumstances of the deletion.” Id., 2021 WL 5919821, at *10. Ottoson v. SMBC Leasing and Fin., Inc. which relies on inherent authority therefore is also inapplicable. 268 F. Supp. 3d 570, 580 (S.D.N.Y. 2017).

The court in Moody v. CSX Transp., Inc. inferred a specific “intent to deprive” under Rule 37(e)(2) from a knowing failure to preserve when “mere negligence” was “implausible.” 271 F. Supp. 3d 410, 424, 427 (W.D.N.Y. 2017). That case concerns the loss of data from a train data recorder, “highly relevant—if not the most important objective evidence”—to determine liability in a train accident where plaintiff lost a limb. Id. at 422, 431. The train company's foreman downloaded the data to a computer and was supposed to then transmit the data to a centralized server, but appeared to have transmitted the wrong file. Id. at 422. His computer then crashed and the company lost the computer without ever trying to retrieve the data. Id. at 423. The court held that the company acted with “intent to deprive” because it could not “credibly explain[ ]” its “failure to make any effort over the course of four years to confirm that the data was properly preserved.” Id. at 431. Additionally, had the defendants made a reasonable inquiry as required by Rules 11 and 26(a), they would have noticed that the data was missing at a time when it was still capable of being retrieved. Id. at 427. Hence, the court found more than mere negligence.

LinkedIn's other cases suggest that courts require conduct akin to bad faith to infer a specific “intent to deprive.” See Sines v. Kessler, No. 3:17-CV-00072, 2021 WL 4943742, at *11 (W.D. Va. Oct. 22, 2021), order approved, No. 3:17-CV-00072, 2021 WL 5492826 (W.D. Va. Nov. 19, 2021) (finding intent to deprive based on defendant's “inability to recall almost any fact about the steps he took (or did not take) to preserve or recover this ESI, combined with his occasionally flippant or dismissive answers to opposing counsel's questions at his court-ordered deposition, demonstrates a contempt for his discovery obligations ․ that further reinforces this conclusion”) (emphasis in original); O'Berry v. Turner, No. 7:15-CV-00064-HL, 2016 WL 1700403, at *4 (M.D. Ga. Apr. 27, 2016) (finding defendants’ “irresponsible and shiftless behavior,” including having no written policy on proper procedure to preserve evidence and failing to collect the documents at issue after numerous requests by the Plaintiff to do so, “can only lead to one conclusion—that [they] acted with the intent to deprive Plaintiff of the use of this information at trial”). In sum, these authorities appear to require a specific intent to deprive, rather than a mere knowing failure to preserve.

b. Direct And Circumstantial Evidence Suggests That hiQ Lost The Evidence At Issue Negligently Or With An Intent To Cut Cost

The circumstances in this case suggest that hiQ destroyed, or allowed to be destroyed, the evidence at issue either negligently or with an intent to cut cost, but not with a specific “intent to deprive [LinkedIn] of the information's use in the litigation.” Fed. R. Civ. P. 37(e)(2). First, LinkedIn seems to agree that any affirmative deletion of the evidence by hiQ is “part of a migration of [hiQ's] data in order to save money.” (Docket No. 335-3 at 6.) hiQ attempted to cut costs in 2018 by migrating all of its MongoDB operations from California and Oregon to Virginia. (Id.; LCE 1535 (2/02/18 hiQ email regarding “All-hands AWS cleanup” to “strip down any systems that [we]re defunct or should be ․ to reduce [its] AWS costs by several thousand dollars”).) In that process, hiQ's employees discussed deleting some AWS data, including data that made up the Scrapus Collections and the Proxies Collection. (LCE 1614 (7/6/18 hiQ Chat Tr.) (discussing deletion of three “AWS volumes”); LCE 1618 (5/16/18 hiQ Chat Tr.) (“Scrapus2 we dont [sic] need to backup”); LCE 1634 (8/22/17 hiQ Chat Tr.) (discussing deleting “Mongo-Scraper”).)

Second, there is no dispute that hiQ has struggled financially and that it lost its Salesforce and AWS accounts because it was unable to pay the bills. (See LCE 1391 (hiQ Second Supp. Resp. to LinkedIn's Interrogatory No. 16) (“Salesforce deleted the raw data from hiQ's Salesforce instance at some point in or around January 2020 after hiQ had been unable to pay its Salesforce bills ․ AWS delet[ed] hiQ's data for non-payment.”).) Although hiQ received repeated suspension notices from AWS and consciously let its AWS account lapse, the decision was driven by financial considerations. In March 2020, a month before AWS permanently suspended hiQ's account, hiQ's CEO wrote to its litigation funder detailing the accounts receivable to “prevent any creditor from pushing the company into a receivership situation and thereby take control of [its] litigation.” (LCE 1255 (3/2/20 Weidick email).) To be sure, hiQ's financial situation does not absolve its duty of preservation, but it does suggest the driving factor was not an intent to deprive LinkedIn of evidence.

Third, the timing of the evidence's loss indicates a lack of such intent. “The most decisive factor [in the ‘intent’ analysis] is the timing” of the offending conduct. Fed. Trade Comm. v. Noland, No. CV-20-00047-PHX-DWL, 2021 WL 3857413, at *12 (D. Ariz. Aug. 30, 2021) (finding “intent to deprive” where defendants installed “elaborate encrypted privacy-focused apps immediately after discovering they were the subject of an FTC investigation”); accord Fourth Dimensions, 2021 WL 5919821, at *10 (finding party to have intentionally destroyed records where it affirmatively destroyed records “shortly after receiving notice that [plaintiff] was prepared to file suit”); Brittney Gobble Photography, LLC v. Sinclair Broad. Grp., Inc., No. SAG-18-3403, 2020 WL 1809191, at *10 (D. Md. Apr. 9, 2020) (“more than a month” delay of routine deletion of emails after receiving subpoena suggested that defendant “did not intend to deprive [plaintiff] of evidence for litigation but merely allowed the routine e-mail deletion to proceed”). Here, hiQ deleted, or allowed to be deleted, the evidence at issue years after the litigation began. There was no rush to delete relevant evidence.

Fourth, there was no selective preservation of evidence from the lost hiQ accounts aimed at depriving LinkedIn of the relevant information. LinkedIn argues that hiQ selectively archived some “raw scrape” data from AWS that “supported [hiQ's] story,” but discarded the remaining scraping activity records that supported LinkedIn's defense or undermined hiQ's claims. (Docket No. 335-3 at 17–18.) However, hiQ's CEO testified that he instructed and expected hiQ's CTO to archive on a hard drive all the raw data within MongoDB in its existing form. (Docket No. 368-1 Ex. P at 94:24-95:7.) LinkedIn did not point to any evidence that the CTO's failure to carry out the instruction was due to an intent to deprive LinkedIn of that information. Amidst hiQ's dire financial situation and intense employee turnover, negligence is the more likely explanation to its failure to preserve evidence.

Finally, hiQ did attempt to recover the data from AWS, Splunk, and Salesforce. (Docket No. 368 at 16.) Although largely unsuccessful, such attempts, coupled with the partial production of Salesforce exports, the Splunk dashboards, and hiQ internal communications on scraping, indicate a lack of intent to deprive LinkedIn of evidence. See Gaina v. Northridge Hosp. Med. Ctr., No. CV 18-00177-DMG (RAOx), 2018 WL 6258895, at *5 (C.D. Cal. Nov. 21, 2018). LinkedIn argues that hiQ is not an unsophisticated individual who spoliated evidence, which goes to show that hiQ should not have lost the evidence in the first place. (Docket No. 387 at 7.) But it does not diminish the inference from hiQ's salvation efforts that hiQ did not intend to hide the evidence from LinkedIn.

In sum, although hiQ should have informed LinkedIn and the Court about the impending closure of its accounts—particularly the AWS account, hiQ's loss of evidence was not driven by an intent to deprive LinkedIn of evidence. It was more akin to negligence. “Negligence—even gross negligence—in failing to retain relevant evidence is not sufficient to support an adverse inference under Rule 37(e)(2).” Meta Platforms, Inc. v. BrandTotal Ltd., ––– F. Supp. 3d. ––––, No. 20-CV-07182-JCS, 2022 WL 1990225, at *6 (N.D. Cal. June 6, 2022). Thus, there is insufficient evidence to find hiQ acted with the requisite intent to warrant harsh Rule 37(e)(2) sanctions.

2. hiQ's Loss Of The Evidence At Issue Has Prejudiced LinkedIn Under Rule 37(e)(1)

“Prejudice exists where ‘the [spoiling party's] actions impaired [the moving party's] ability to go to trial or threatened to interfere with the rightful decision of the case.’ ” RG Abrams Ins. v. L. Offs. of C.R. Abrams, 2022 WL 3133293, at *29 (C.D. Cal. July 1, 2022) (quoting United States ex rel. Wiltec Guam, Inc. v. Kahaluu Constr. Co., 857 F.2d 600, 604 (9th Cir. 1988)).

Rule 37(e)(1) “does not place a burden of proving or disproving prejudice on one party or the other,” but instead “leaves judges with discretion to determine how best to assess prejudice in particular cases.” Fed. R. Civ. P. 37(e)(1) advisory committee's note to 2015 amendment. The Committee Notes acknowledge that, in some circumstances, determining the content of lost ESI will be difficult and placing the burden of proving prejudice on the moving party may be unfair. Id. Courts have discretion under Rule 37(e)(1) to determine how to assess prejudice on a case-by-case basis. Id. “Proving that lost evidence is relevant can be a difficult task, however, because the evidence no longer exists. To show prejudice resulting from the spoliation, therefore, courts have held that a party must only come forward with plausible, concrete suggestions as to what [the destroyed] evidence might have been.” Fast v. GoDaddy.com LLC, 340 F.R.D. 326, 339 (D. Ariz. 2022) (citations omitted, internal quotation marks omitted).

LinkedIn argues that the lost AWS and Splunk data inform the factual questions of (1) “how many requests hiQ's scraping bots made to LinkedIn's servers” that “goes directly to hiQ's argument that LinkedIn cannot show the requisite harm to its system,” and (2) the failure rate (or “ban rate”) of hiQ's scraping attempts that rebut hiQ's claim that LinkedIn's C & D caused hiQ's business’ downfall. (Docket No. 355-3 at 19, 21.) LinkedIn's expert explained that “[d]uring the time that hiQ was using Splunk, the deleted data showed the full scope of hiQ's scraping activity.” (Id. (citing Schmidt Decl. at ¶ 13).) And based on hiQ's source code, the deleted AWS “would have provided enough information to make a much more precise estimate of all the scraping requests from the period before hiQ started using Splunk.” (Id. at 19–20 (citing Schmidt Decl. at ¶ 9).) hiQ responds that hiQ's produced daily Splunk scraping dashboards and internal communications provide the evidence LinkedIn seeks. (Docket No. 368 at 22.)

Prejudice. The loss of hiQ's scraping data prejudiced LinkedIn because it interfered with the accurate determination of the number of hiQ's scraping requests and the ban rates. Without the lost scraping activity records, LinkedIn had to estimate those numbers by (1) extracting into a table the timestamp, URL, and scrape ID from the raw scrape data produced by hiQ (Schmidt Decl. at ¶ 6), (2) comparing the URL and timestamp to LinkedIn's activity logs to identify the IP addresses that made the scraping requests, and (3) locating requests in LinkedIn's server logs associated with the IP addresses. (Docket No. 335-3 at 20.) “The results showed that activity associated with the hiQ IP addresses totaled over 50 billion requests in an 18-month period.” (Id.) hiQ avers that LinkedIn could have determined the number of hiQ bot requests to LinkedIn's server from existing data, but does not explain how to do so. The Court therefore credits LinkedIn's expert's testimony that “he would have been able to make a much more precise estimate of the number of scrape attempts, and the success rate of those attempts” from the lost data “than [he] could with the data that was produced.” (Schmidt Decl. at ¶ 9.)

The lost scraping records also prejudiced LinkedIn financially. Instead of directly assessing hiQ's scraping requests from its AWS and Splunk data, LinkedIn had to pay its outside experts and consultants to analyze hiQ's raw data and use its internal employee time and resources to estimate the numbers. (Docket No. 335-3 at 21.)

Sanctions. For the destruction of the scraping activity records relating to the number of hiQ scraping attempts, LinkedIn requests that the Court (1) presume that LinkedIn's expert Xiaofeng Wu's analysis is correct and that hiQ's scrapers made at least fifty billion requests on LinkedIn's servers, (2) issue mandatory adverse inference jury instructions to that effect, and (3) preclude hiQ from offering any evidence to the contrary or otherwise disputing that analysis in connection with summary judgment, Rule 702 motions, and trial. (Docket No. 335-3 at 24.) LinkedIn also seeks costs associated with analysis directed by Mr. Wu, which would have been unnecessary had hiQ retained its scraping activity records and produced them. (Id. at 25.)

Although LinkedIn was not able to directly assess the number of hiQ's scraping attempts and the ban rates, it was able to estimate those statistics using other means. Therefore, the prejudice likely does not warrant “the quite extraordinary relief ․ of mandatory adverse inference instructions.” Chinitz v. Intero Real Est. Servs., No. 18-cv-05623-BLF, 2020 WL 7389417, at *3 (N.D. Cal. May 13, 2020). The Court therefore will issue permissive adverse inference instructions that the jury may presume that (1) LinkedIn's expert Xiaofeng Wu's analysis is correct and that hiQ's scrapers made at least fifty billion requests on LinkedIn's servers, and that (2) prior to hiQ's receipt of the C & D letter, LinkedIn's general technical defenses had blocked hiQ's anonymous scraping from collecting data to such a degree that hiQ could no longer effectively scrape LinkedIn profiles. LinkedIn's request for the Court to presume the same in the context of summary judgment is moot as these facts are not necessary to the Court's ruling on the motions for summary judgment.

Additionally, the Court grants LinkedIn's requests for costs associated with the analysis directed by Mr. Wu that would have been unnecessary but for hiQ's loss of evidence. The Court also grants LinkedIn's attorney fees for bringing this motion that would not have been necessary had hiQ carried out its duty to preserve. hiQ opposes based on its “precarious financial position” and that an attorney fee award would be a case-ending sanction—a windfall not appropriate under Rule 37(e)(1). (Docket No. 369 at 24.) But hiQ could have satisfied its duty by notifying LinkedIn of the closure of its various accounts at issue.

At this juncture, the Court instructs the Clerk of the Court to file this order, in its entirety, under seal. The Court orders the parties to meet and confer to determine which portions of this order may be publicly filed. The parties shall jointly file their request to file under seal within a week of the date of this order.

This order disposes of Docket Nos. 336, 337, 338, 339, and 355.

IT IS SO ORDERED.

FOOTNOTES

2.   “HCE” and “LCE” refer to hiQ's and LinkedIn's Compendium of Evidence, respectively, filed concurrently with the motions subject to this order.

3.   The Court overrules LinkedIn's evidentiary objection that hiQ's attempt to impute the knowledge or statements of participants in the email chains to LinkedIn lacks foundation. “At the summary judgment stage, we do not focus on the admissibility of the evidence's form. We instead focus on the admissibility of its contents.” See Fraser v. Goodale, 342 F.3d 1032, 1036 (9th Cir. 2003). hiQ can lay the requisite foundation at trial by, for example, calling those participants as witnesses.

4.   The Court overrules LinkedIn's objection to this exhibit because it is not properly authenticated. For the purpose of this motion, the Court considers this exhibit because it can be presented in a form admissible at trial. See Fraser, 342 F.3d at 1036 (“At the summary judgment stage, we do not focus on the admissibility of the evidence's form. We instead focus on the admissibility of its contents.”); see also Faulks v. Wells Fargo & Co., 231 F. Supp. 3d 387 (N.D. Cal. 2017).

5.   LinkedIn objects to hiQ's Exhibit F because it is not properly authenticated. The Court overrules this objection for the same reason as above.

EDWARD M. CHEN, United States District Judge