FOR PUBLICATION

UNITED STATES COURT OF APPEALS

FOR THE NINTH CIRCUIT

IN RE GOOGLE INC. STREET VIEW No. 20-15616

ELECTRONIC COMMUNICATIONS

LITIGATION, D.C. No. 3:10-md-02184-

CRB

BENJAMIN JOFFE; LILLA MARIGZA;

RICK BENITTI; BERTHA DAVIS;

JASON TAYLOR; ERIC MYHRE; JOHN OPINION

E. REDSTONE; MATTHEW BERLAGE;

PATRICK KEYES; KARL H. SCHULZ;

JAMES FAIRBANKS; AARON LINSKY;

DEAN M. BASTILLA; VICKI VAN

VALIN; JEFFREY COLMAN; RUSSELL

CARTER; STEPHANIE CARTER;

JENNIFER LOCSIN,

Plaintiffs-Appellees,

DAVID LOWERY,

Objector-Appellant,

v.

GOOGLE, INC.,

Defendant-Appellee.

Appeal from the United States District Court for the Northern District of California Charles R. Breyer, District Judge, Presiding 2 IN RE GOOGLE INC. STREET VIEW LITIG.

Argued and Submitted February 11, 2021 San Francisco, California

Filed December 27, 2021

Before: Marsha S. Berzon, Morgan Christen, and Bridget S. Bade, Circuit Judges.

Opinion by Judge Bade; Concurrence by Judge Bade

SUMMARY *

Class Actions

The panel affirmed the district court’s order certifying a class, approving a settlement agreement, and awarding attorneys’ fees, in a consolidated class action lawsuit in which plaintiffs alleged, on behalf of an estimated sixty million people, that Google illegally collected their Wi-Fi data through its Street View program.

After a decade of litigation, including a complex, three- year forensic investigation to confirm the standing of the eighteen named plaintiffs, the parties reached a settlement agreement that provided for injunctive relief, cy pres payments to nine Internet privacy advocacy groups, fees for the attorneys, and service awards to class representatives— but no payments to absent class members. The district court

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.

IN RE GOOGLE INC. STREET VIEW LITIG. 3

approved the proposed settlement, finding that it was not feasible to distribute funds directly to class members given the class size and the technical challenges to verifying class members’ claims. David Lowery, one of two objectors to the settlement proposal, appealed the district court’s approval of the settlement and grant of attorneys’ fees.

Federal courts have widely recognized the cy pres doctrine as a tool for distributing unclaimed or non- distributable portions of a class action settlement fund to the “next best” class of beneficiaries.

The panel rejected the suggestion that a district court may not approve a class-action settlement that provides monetary relief only in the form of cy pres payments to third parties.

Lowery argued that, even if permissible in some circumstances, cy pres relief was inappropriate here because it was feasible to distribute settlement funds directly to class members. Because self-identification would be pure speculation, and any meaningful forensic verification of claims would be prohibitively costly and time-consuming, the panel affirmed the district court’s finding that it was not feasible to verify class members’ claims as would be necessary to distribute funds directly to class members. Further, as proof of individual claims would be burdensome and distribution of damages costly, the panel held that the district court did not abuse its discretion by approving the use of cy pres payments in the settlement.

The panel rejected Lowery’s argument that if it was impossible to distribute settlement funds to class members, then class certification was an error of law because the class device was not superior to other available methods for fairly 4 IN RE GOOGLE INC. STREET VIEW LITIG.

and efficiently adjudicating the controversy, as Fed. R. Civ. P. 23(b)(3) requires. Noting that this court, in upholding the validity of cy pres arrangements, has repeatedly recognized that class members do benefit—albeit indirectly—from a defendant’s payment of funds to an appropriate third party, the panel held that the infeasibility of distributing settlement funds to class members does not preclude class certification.

Considering the unique challenges plaintiffs would have faced in proving their claims, the panel held that the district court did not err by concluding that the injunctive relief— which required Google to destroy all acquired payload data, refrain from collecting or storing additional payload data through Street View without notice and consent, and comply with other provisions in an assurance of voluntary compliance entered into with the attorneys general of thirty- eight states and the District of Columbia—together with the indirect benefits conferred by the cy pres provisions, was “fair, reasonable, and adequate” compensation to the class members under Fed. R. Civ. P. 23(e)(2).

Lowery argued that the settlement violates the First Amendment’s prohibition on compelled speech by distributing class settlement funds to organizations “that take lobby positions adverse to” his own interests and beliefs. The panel did not decide whether, or under what circumstances, a district court’s approval of a class action settlement agreement is “state action” for purposes of the First Amendment. Instead, the panel held that the settlement agreement does not compel class members to subsidize third-party speech because any class member who does not wish to subsidize speech by a third party that he or she does not wish to support, can simply opt out of the class.

IN RE GOOGLE INC. STREET VIEW LITIG. 5

Lowery argued that the district court abused its discretion by approving cy pres recipients who had a “significant prior affiliation” with defense counsel and class counsel. The panel noted that this court has never held that merely having previously received cy pres funds from a defendant, let alone other defendants in unrelated cases, disqualifies a proposed recipient for all future cases; and that this court has affirmed cy pres provisions involving much closer relationships between recipients and parties than anything Lowery alleges here. The panel concluded that the district court’s approval of the cy pres recipients comported with the applicable standards, and found no abuse of discretion.

Lowery argued that the district court abused its discretion by “blindly applying” a 25% benchmark for attorneys’ fees without regard for the actual benefit the settlement conferred on the class. The panel wrote that the district court’s reasoning makes clear that this was not a “blind” application of a benchmark to the circumstances of the case. The panel also explained that there is no uniform rule that district courts must discount the value of any cy pres relief, regardless of the feasibility of distribution to class members or other relevant circumstances. Affirming the fee award, the panel wrote that the district court properly considered all relevant circumstances, including the value to class members.

Because the panel affirmed the district court’s finding that the settlement provides adequate value to the class, and because there is no indication that counsel accepted attorneys’ fees or favored third parties over class members, the panel rejected Lowery’s argument that class counsel and their class representatives breached their fiduciary duties by entering the settlement. 6 IN RE GOOGLE INC. STREET VIEW LITIG.

Concurring, Judge Bade wrote separately to express some general concerns about cy pres awards. She wrote that she is not convinced that cy pres awards to uninjured third parties should qualify as an indirect benefit to injured class members, and that she is concerned that the cy pres remedy is purely punitive, with defendants paying millions of dollars in what are essentially civil fines to class counsel and third parties while providing no compensation to injured class members. She further questioned whether cy pres awards are inherently unfair when the class receives no meaningful relief in exchange for their claims, and whether such awards can be justified given the serious ethical, procedural, and constitutional problems that others have identified.

COUNSEL

Adam E. Schulman (argued) and Theodore H. Frank, Hamilton Lincoln Law Center, Center for Class Action Fairness, Washington, D.C., for Objector-Appellant.

Daniel A. Small (argued) and Robert W. Cobbs, Cohen Milstein Sellers & Toll PLLC, Washington, D.C.; Elizabeth L. Cabraser, Michael W. Sobol, and Melissa Gardner, Leiff Cabraser Heimann & Bernstein LLP, San Francisco, California; Jeffrey L. Kodroff, John A. Macoretta, and Mary Ann Geppert, Spector Roseman & Kodroff P.C., Philadelphia, Pennsylvania; for Plaintiffs-Appellees.

Brian M. Willen (argued) and Eli B. Richlin, Wilson Sonsini Goodrich & Rosati, New York, New York; David H. Kramer, Wilson Sonsini Goodrich & Rosati, Palo Alto, California; Paul N. Harold, Wilson Sonsini Goodrich & Rosati, Washington, D.C.; for Defendant-Appellee.

IN RE GOOGLE INC. STREET VIEW LITIG. 7

Kate B. Sawyer (argued), Assistant Solicitor General; Keena Patel, Assistant Attorney General; Oramel H. Skinner, Solicitor General; Mark Brnovich, Attorney General; Office of the Attorney General, Phoenix, Arizona; Steve Marshall, Attorney General, State of Alabama; Kevin G. Clarkson, Attorney General, State of Alaska; Leslie Rutledge, Attorney General, State of Arkansas; Lawrence G. Wasden, Attorney General, State of Idaho; Curtis T. Hill Jr., Attorney General, State of Indiana; Derek Schmidt, Attorney General, State of Kansas; Jeff Landry, Attorney General, State of Louisiana; Eric Schmitt, Attorney General, State of Missouri; Aaron D. Ford, Attorney General, State of Nevada; Wayne Stenehjem, Attorney General, State of North Dakota; Dave Yost, Attorney General, State of Ohio; Mike Hunter, Attorney General, State of Oklahoma; for Amici Curiae Thirteen Attorneys General for the States of Arizona, Alabama, Alaska, Arkansas, Idaho, Indiana, Kansas, Louisiana, Missouri, Nevada, North Dakota, Ohio, and Oklahoma.

Ellen Bronchetti and Ron Holland, McDermott Will & Emery LLP; San Francisco, California; Wilber H. Boies and Timothy M. Kennedy, McDermott Will & Emery LLP, Chicago, Illinois; for Amici Curiae Legal Aid Organizations.

Stuart T. Rossman, National Consumer Law Center, Boston, Massachusetts; Michael Landis, Center for Public Interest Research, Denver, Colorado; for Amici Curiae United States Public Interest Research Group Education Fund and National Consumer Law Center. 8 IN RE GOOGLE INC. STREET VIEW LITIG.

OPINION

BADE, Circuit Judge:

In this consolidated class action lawsuit, plaintiffs alleged, on behalf of an estimated sixty million people, that Google illegally collected their Wi-Fi data through its Street View program. After a decade of litigation, including a complex, three-year forensic investigation to confirm the standing of the eighteen named plaintiffs, the parties reached a settlement agreement that provided for injunctive relief, cy pres payments to nine Internet privacy advocacy groups, fees for the attorneys, and service awards to class representatives—but no payments to absent class members. The district court approved the proposed settlement, finding that it was not feasible to distribute funds directly to class members given the class size and the technical challenges to verifying class members’ claims.

David Lowery, one of two objectors to the settlement proposal, appeals the district court’s approval of the settlement and grant of attorneys’ fees. He argues that the district court should not have approved the settlement because it was feasible to distribute funds to class members, and that if it truly was not feasible to do so, then the district court should not have certified the class. He also asserts that the settlement violated the First Amendment’s prohibition on compelled speech, that the cy pres recipients had improper relationships with the parties and class counsel, that the district court awarded excessive attorneys’ fees, and that class counsel and the class representatives breached their fiduciary duties. We conclude that the district court did not abuse its discretion in approving the settlement, certifying the class, or in its award of attorneys’ fees, and that it did not commit legal error by rejecting Lowery’s First Amendment argument. We affirm.

IN RE GOOGLE INC. STREET VIEW LITIG. 9

I

In 2007, Google launched Street View, a web-based technology that would eventually provide users with panoramic street-level images from numerous points along roads throughout the world. To obtain the images for Street View, Google deployed a fleet of specially adapted cars (“Street View Vehicles”). As it turned out, however, these vehicles did not simply take photographs; they were also equipped with Wi-Fi antennas and software designed to collect, decode, and analyze various kinds of data commonly transmitted over Wi-Fi networks. The Street View Vehicles collected basic identifying information—such as signal strength, broadcasting channel, data transmission rate, media access control (“MAC”) address, and Service Set Identifier (“SSID”)—from Wi-Fi networks along the roads they travelled, apparently for the purpose of providing enhanced, “location-aware” services to Street View users. 1

In May 2010, Google revealed that its Street View Vehicles had been collecting not just network identifying information, but also payload data—that is, substantive information such as emails, usernames, passwords, videos, photographs, and documents—that Internet users transmitted over unencrypted Wi-Fi networks when the Street View Vehicles were nearby. See Joffe v. Google, Inc., 729 F.3d 1262, 1264 (9th Cir.), amended and superseded on reh’g by 746 F.3d 920 (9th Cir. 2013). In total, the Street View Vehicles apparently collected around three billion

1 As Google explains it, this identifying information for Wi-Fi networks would allow Street View to utilize these networks as “unique geographical landmark[s]” for users to pinpoint their location when satellite-based GPS is unavailable. 10 IN RE GOOGLE INC. STREET VIEW LITIG.

frames of raw data from wireless networks, including approximately 300 million frames containing payload data.

Google publicly apologized for collecting payload data, suspended operation of the Street View Vehicles, and stated that it had segregated the data and rendered it inaccessible. It insisted (as it still maintains) that it never intended to collect payload data. Nevertheless, the revelations led to state and federal investigations, including a joint investigation by the attorneys general of thirty-eight states and the District of Columbia. In March 2013, Google entered an Assurance of Voluntary Compliance (“AVC”) with these attorneys general regarding its collection of Wi- Fi data from Street View Vehicles. Among other provisions, the AVC stated that Google would destroy all payload data it had acquired, refrain from collecting or storing any additional payload data through Street View without notice and consent, maintain a “privacy program” as described in the AVC, and undertake a public service and education campaign. 2 The AVC also required Google to pay a total of $7 million to the attorneys general.

But Google’s legal troubles related to the Street View Vehicles did not end with the AVC. Shortly after Google’s May 2010 admission, at least thirteen putative class action lawsuits were brought based on the Street View Vehicles’ collection of payload data. In August 2010, the Judicial

2 The public service campaign was required to include several components, including “[d]evelop[ing] and promot[ing] a video on YouTube that explains how users can encrypt their wireless networks,” keeping the video on YouTube for at least two years, writing “a blog post . . . explaining the value of encrypting a wireless network,” and running “at least one half-page educational newspaper ad in a newspaper of national circulation and at least one half-page educational ad in the newspaper with the greatest circulation rate in each State.”

IN RE GOOGLE INC. STREET VIEW LITIG. 11

Panel on Multidistrict Litigation consolidated eight of these cases and transferred them to the Northern District of California.

In November 2010, Plaintiffs filed a Consolidated Class Action Complaint asserting various state and federal claims, including violations of the Wiretap Act, see 18 U.S.C. § 2511, and seeking statutory and punitive damages as well as injunctive relief. Google moved to dismiss the complaint, and the district court dismissed the state law claims on pre- emption and standing grounds but held that Plaintiffs had adequately alleged violations of the Wiretap Act. See In re Google Inc. St. View Elec. Commc’ns Litig., 794 F. Supp. 2d 1067, 1073–87 (N.D. Cal. 2011). We affirmed in an interlocutory appeal. Joffe, 746 F.3d 920.

On remand, Google disputed the named plaintiffs’ standing, and the district court appointed a special master to determine “whether any communications from [named] Plaintiffs’ unencrypted Wi-Fi networks were actually acquired by Google.” This investigation first required the eighteen named plaintiffs to provide “personal information and forensic evidence of their wireless network equipment,” including MAC addresses, email addresses, and SSIDs, to the special master. Then, as the district court described it, the special master organized the massive troves of Street View data “into a searchable database,” developed custom software to process the data, and “conducted complex technical searches” to identify whether the data contained any transmissions intercepted from the named plaintiffs. This process took three years and culminated in a report, filed under seal with the district court in 2017, which was apparently still not entirely conclusive on whether Google had intercepted payload data from the named plaintiffs. 12 IN RE GOOGLE INC. STREET VIEW LITIG.

In June 2018, the parties reached a settlement agreement for a class consisting of “all persons who used a wireless network device from which Acquired Payload Data was obtained” from January 1, 2007 through May 15, 2010. Class counsel estimated that this class included approximately sixty million members. The settlement agreement provided that Google would establish a $13 million settlement fund. The agreement did not provide for any direct payments to absent class members. Instead, after attorneys’ fees, litigation expenses, service awards for the class representatives, notice and claims administration costs, and escrow account charges and taxes, the remainder of the fund was to be divided equally among “one or more Proposed Cy Pres Recipient(s).” Plaintiffs would select the proposed recipients and, after disclosing the list to Google and consulting “in good faith regarding any concerns Google may have,” would recommend them to the district court for approval. Each cy pres recipient would have to “commit to use the funds to promote the protection of Internet privacy.”

Plaintiffs proposed eight cy pres recipients without objection from Google: the Center on Privacy & Technology at Georgetown Law, the Center for Digital Democracy, Massachusetts Institute of Technology’s Internet Policy Research Initiative, World Privacy Forum, Public Knowledge, the Rose Foundation for Communities and the Environment, the American Civil Liberties Union Foundation (ACLU), and Consumer Reports. The Electronic Privacy Information Center (EPIC) also successfully petitioned the district court to be included as a cy pres recipient without objection from Google or Plaintiffs.

In addition to the provisions regarding the $13 million settlement fund, Google agreed to the following injunctive

IN RE GOOGLE INC. STREET VIEW LITIG. 13

relief for a period of five years after final approval of the settlement agreement:

• To “destroy all Acquired Payload Data . . . within forty-five (45) days of Final Approval” of the settlement agreement;

• Not to “collect and store for use in any product or service Payload Data via Street View vehicles, except with notice and consent”;

• To “comply with all aspects of the Privacy Program described in . . . the [AVC] and with the prohibitive and affirmative conduct described in [the AVC],” and to “confirm to Plaintiffs in writing on an annual basis that it remains in compliance”; and

• To “host and maintain educational webpages that instruct users on the configuration of wireless security modes and the value of encrypting a wireless network.”

After the district court granted preliminary approval of the settlement agreement, two putative class members— David Lowery and David Franco—objected, and a group of state attorneys general, led by the Arizona Attorney General, filed an amicus brief objecting to the settlement agreement. At a fairness hearing in February 2020, Lowery’s attorney and a representative from the Arizona Attorney General’s Office both argued that cy pres relief was inappropriate and that the $13 million fund should instead be distributed to class members through either a claims process or a lottery distribution to class members who self-identified. Alternatively, Lowery argued that if it truly was not feasible to distribute the funds to class members, then class 14 IN RE GOOGLE INC. STREET VIEW LITIG.

certification was inappropriate based on Federal Rule of Civil Procedure 23(b)(3)’s requirement that the class device be superior to other forms of adjudication. Lowery also argued that distribution of settlement funds to cy pres recipients constituted compelled speech in violation of the First Amendment, that the proposed recipients had improper pre-existing relationships with counsel and the parties, and that the requested 25% fee was excessive.

In March 2020, the district court certified the class for settlement purposes under Rule 23(b)(3), granted attorneys’ fees of 25% of the net settlement fund, and approved the settlement after considering the fairness factors of Rule 23(e)(2) and the reaction of the class members. The district court rejected Lowery’s arguments about the feasibility of distribution and concluded that the inability to distribute funds did not preclude class certification. It also rejected Lowery’s First Amendment argument, his objections to the cy pres recipients, and his objection to the fee award. Lowery timely appealed.

II

“We review a district court’s approval of a proposed class action settlement, including a proposed cy pres settlement distribution, for abuse of discretion.” Nachshin v. AOL, LLC, 663 F.3d 1034, 1038 (9th Cir. 2011). “[W]e will affirm if the district judge applies the proper legal standard and his findings of fact are not clearly erroneous.” In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d 935, 940 (9th Cir. 2011) (internal quotation marks and citation omitted). We also “review a district court’s class certification decision for abuse of discretion.” Sali v. Corona Reg’l Med. Ctr., 909 F.3d 996, 1002 (9th Cir. 2018). We review a First Amendment challenge to the district court’s approval of a settlement agreement de novo. See

IN RE GOOGLE INC. STREET VIEW LITIG. 15

Pac. Coast Horseshoeing Sch., Inc. v. Kirchmeyer, 961 F.3d 1062, 1067 n.3 (9th Cir. 2020). “We also review for abuse of discretion a district court’s award of fees and costs to class counsel, as well as its method of calculation.” Bluetooth Headset, 654 F.3d at 940. “Findings of fact underlying an award of fees are reviewed for clear error.” Id.

III

Before turning to Lowery’s specific objections to the settlement, we first review the development of cy pres provisions as a tool to address unclaimed or non- distributable funds from class action settlements, and our precedent addressing such provisions. As one court has explained, “[w]hen modern, large-scale class actions are resolved via settlement, money often remains in the settlement fund even after initial distributions to class members have been made because some class members either cannot be located or decline to file a claim.” Klier v. Elf Atochem N. Am., Inc., 658 F.3d 468, 473 (5th Cir. 2011); see Six (6) Mexican Workers v. Ariz. Citrus Growers, 904 F.2d 1301, 1307 (9th Cir. 1990). Courts have recognized a few possible solutions to the problem of unclaimed settlement funds. One option is to permit such funds to escheat to the government. Hodgson v. YB Quezada, 498 F.2d 5, 6 (9th Cir. 1974); see 28 U.S.C. § 2042 (providing that funds “unclaimed by the person entitled thereto” for five years revert to the federal treasury). In other cases, courts have permitted additional pro rata distributions to those class members who did claim funds. See, e.g., Klier, 658 F.3d at 475. “[I]n exceptional circumstances,” courts have even recognized that “it may be proper to permit unclaimed sums to revert to the [defendant].” YB Quezada, 498 F.2d at 6; see also, e.g., Van Gemert v. Boeing Co., 739 F.2d 730, 736–37 (2d Cir. 1984). 16 IN RE GOOGLE INC. STREET VIEW LITIG.

Beginning in the 1970s, some federal courts began to recognize another option for disbursing unclaimed settlement funds. In Miller v. Steinbach, the district court for the Southern District of New York considered “a somewhat unorthodox settlement” in a stockholders’ derivative suit. No. 66 Civ. 356, 1974 WL 350, at *1 (S.D.N.Y. Jan. 3, 1974). “In view of the very modest size of the settlement fund” in that case “and the vast number of shares among which it would have to be divided,” the parties agreed to, and the district court approved, an arrangement by which settlement funds would be paid to an employee retirement plan rather than class members. Id. The district court described this arrangement as “a variant of the cy pres doctrine at common law.” Id. That doctrine, which “takes its name from the Norman French expression cy pres comme possible (or ‘as near as possible’), is an equitable doctrine that originated in trusts and estates law as a way to effectuate the testator’s intent in making charitable gifts.” In re Google Referrer Header Priv. Litig., 869 F.3d 737, 741 (9th Cir. 2017), vacated and remanded, Frank v. Gaos, 139 S. Ct. 1041 (2019).

In the years since Miller, federal courts have widely recognized the cy pres doctrine as a tool for “distribut[ing] unclaimed or non-distributable portions of a class action settlement fund to the ‘next best’ class of beneficiaries.” Nachshin, 663 F.3d at 1036 (citation omitted). It is well established in this circuit that district courts may approve settlements with cy pres provisions that affect only a portion of the total settlement fund. See, e.g., Molski v. Gleich, 318 F.3d 937, 954 (9th Cir. 2003), overruled on other grounds by Dukes v. Wal-Mart Stores, Inc., 603 F.3d 571 (9th Cir. 2010) (en banc). Moreover, although no binding Ninth Circuit precedent specifically addresses the propriety of settlements where, as here, the only monetary relief comes

IN RE GOOGLE INC. STREET VIEW LITIG. 17

in the form of cy pres payments to third parties, we upheld such a settlement in Lane v. Facebook, Inc., 696 F.3d 811, 820–21 (9th Cir. 2012), and have repeatedly indicated that such settlements are permissible under appropriate circumstances.

For example, in Nachshin v. AOL, LLC, we reversed approval of a settlement that included cy pres payments “on behalf of a nationwide plaintiff class” to “four charities of the class representatives’ choice” and three other agreed- upon charities, including the Boys and Girls Club of America, the New Roads School of Santa Monica, Oklahoma Indian Legal Services, the Federal Judicial Center Foundation, and the Friars Foundation. 663 F.3d at 1036– 37. The district court approved cy pres payments to these charities, whose work had little to do with the plaintiffs’ claims (unjust enrichment based on AOL’s wrongful insertion of promotional messages into subscribers’ emails), after the parties concluded that monetary damages “were small and difficult to ascertain,” and “they could not identify any charitable organization that would benefit the class or be specifically germane to the issues in the case.” Id. at 1037.

We reversed, not because the monetary relief went only to cy pres recipients instead of class members, but because the chosen recipients were unsuitable given the composition and injuries of the plaintiff class. The diverse assortment of cy pres recipients, we held, “fail[ed] to meet any of the guiding standards” for such settlements, id. at 1040, which require that cy pres disbursements “account for the nature of the plaintiffs’ lawsuit, the objectives of the underlying statutes, and the interests of the silent class members, including their geographic diversity,” id. at 1036. We explained: 18 IN RE GOOGLE INC. STREET VIEW LITIG.

We are also not persuaded by the parties’ claims that the size and geographic diversity of the plaintiff class make it “impossible” to select an adequate charity. It is clear that all members of the class share two things in common: (1) they use the internet, and (2) their claims against AOL arise from a purportedly unlawful advertising campaign that exploited users’ outgoing e-mail messages. The parties should not have trouble selecting beneficiaries from any number of non-profit organizations that work to protect internet users from fraud, predation, and other forms of online malfeasance. If a suitable cy pres beneficiary cannot be located, the district court should consider escheating the funds to the United States Treasury.

Id. at 1040–41.

We again considered a settlement that provided no monetary relief directly to absent class members in Lane, where a district court approved a settlement agreement in which Facebook would pay $9.5 million in exchange for a release of all the plaintiffs’ class claims. 696 F.3d at 816. After attorneys’ fees, administrative costs, and class representative payments, “Facebook would use the remaining $6.5 million or so in settlement funds to set up a new charity organization” “to educate users, regulators[,] and enterprises regarding critical issues relating to protection of identity and personal information online.” Id. at 817 (alteration in original).

IN RE GOOGLE INC. STREET VIEW LITIG. 19

On appeal, objectors argued that the settlement was unfair because its cy pres provision gave Facebook too much control over the charity and because the settlement amount was too small. Id. at 820, 822. We affirmed the district court’s approval of the settlement, reasoning that “[t]he cy pres remedy the settling parties here have devised bears a direct and substantial nexus to the interests of absent class members and thus properly provides for the ‘next best distribution’ to the class.” Id. at 821. While we did not explicitly analyze the propriety of so-called “cy pres-only” settlements as a general matter, 3 we indicated that such arrangements can be appropriate provided they have “the requisite nexus between the cy pres remedy and the interests” of the class members. Id. at 822.

In In re Google Referrer Header Privacy Litigation, we reviewed a district court’s approval of a settlement involving “a cy pres-only distribution of the [amount] that remain[ed] in the settlement fund after attorneys’ fees, administration costs, and incentive awards for the named plaintiffs.” 869 F.3d at 741. “As an initial matter, we quickly dispose[d] of the argument that the district court erred by approving a cy pres-only settlement.” Id. While recognizing that such “settlements are considered the exception, not the rule,” we

3 The term “cy pres-only settlement” is a misnomer. As in Nachshin, Lane, and Google Referrer, the settlement here does not only provide cy pres payments to third parties; it also includes injunctive relief. While “cy pres only” may be a convenient shorthand for settlements that provide for monetary payments to third parties but not to absent class members, we apply the same standards when reviewing these settlements that we would for any class action settlement, asking whether the total relief afforded by the settlement—whether in the form of injunctive relief, cy pres payments, or direct monetary payments—adequately compensates class members for relinquishing their claims. See Koby v. ARS Nat’l Servs., Inc., 846 F.3d 1071, 1079 (9th Cir. 2017). 20 IN RE GOOGLE INC. STREET VIEW LITIG.

held that “they are appropriate where the settlement fund is ‘non-distributable’ because ‘the proof of individual claims would be burdensome or distribution of damages costly.’” Id. (quoting Lane, 696 F.3d at 819).

The Supreme Court granted certiorari in Google Referrer on the issue of “whether a class action settlement that provides a cy pres award but no direct relief to class members satisfies the requirement that a settlement binding class members be ‘fair, reasonable, and adequate.’” Frank v. Gaos, 139 S. Ct. 1041, 1045 (2019) (quoting Fed. R. Civ. P. 23(e)(2)). Ultimately, however, the Supreme Court did not reach this question; instead, it vacated and remanded on standing grounds. Id. at 1046. Our analysis of the cy pres issue in Google Referrer, while no longer binding, is still persuasive authority. See Rosenbloom v. Pyott, 765 F.3d 1137, 1154 n.14 (9th Cir. 2014).

IV

Turning to Lowery’s arguments, we reiterate at the outset that strictly speaking, the settlement here is not, as Lowery describes it, a “cy pres-only settlement.” Instead, it involves cy pres payments to third-party organizations and injunctive relief. Nonetheless, in evaluating whether the settlement was “fair, reasonable, and adequate” under Rule 23(e)(2), we first consider the district court’s finding that it was not feasible to distribute funds directly to class members. Second, we consider Lowery’s argument that if it was infeasible to distribute funds directly to class members, the district court should not have certified the class. Third, we ask whether the total value of the settlement to the absent class members—that is, the value they indirectly receive through the cy pres provisions plus the value of the injunctive relief—is enough to justify the district court’s approval of the settlement agreement. Finally, we turn to

IN RE GOOGLE INC. STREET VIEW LITIG. 21

Lowery’s argument that class counsel and the class representatives breached their fiduciary duties, his First Amendment challenge to the cy pres provisions, and his argument against the district court’s award of attorneys’ fees.

A

As a threshold issue, we reject the suggestion that a district court may not approve a class-action settlement that provides monetary relief only in the form of cy pres payments to third parties. 4 We have repeatedly approved such settlements, see Google Referrer, 869 F.3d at 741–42; Lane, 696 F.3d at 822, and therefore adopting a blanket rule against these arrangements, as Lowery advocates, would be incompatible with our precedents in which we have recognized that cy pres awards are an acceptable solution when settlement funds are not distributable. Our reasoning has not turned on what portion of the settlement funds— some or all—is not distributable. Instead, we ask whether the cy pres disbursements “account for the nature of the plaintiffs’ lawsuit, the objectives of the underlying statutes, and the interests of the silent class members.” Lane, 696 F.3d at 821 (quoting Nachshin, 663 F.3d at 1036). In declining to “impose[] a categorical ban on a settlement that does not include direct payments to class members,” Google Referrer, 869 F.3d at 742, we note that other circuits have generally taken a similar approach to ours, approving cy pres settlements when they satisfy the appropriate standards for 4 Lowery does not directly assert that all such settlements are inappropriate. However, the dilemma he poses—either the funds were distributable, and thus cy pres relief was inappropriate, or the funds were not distributable, and thus class certification was inappropriate—is logically equivalent to arguing such settlements are never appropriate and requires us to consider whether Rule 23(e)(2) ever allows them. 22 IN RE GOOGLE INC. STREET VIEW LITIG.

fairness. See In re Google Inc. Cookie Placement Consumer Priv. Litig., 934 F.3d 316, 326 (3d Cir. 2019) (rejecting the argument that “cy pres-only settlements are unfair per se under Rule 23(e)(2)” and recognizing that “[i]n some cases a cy pres-only settlement may be proper”); see also, e.g., In re Lupron Mktg. & Sales Pracs. Litig., 677 F.3d 21, 31–34 (1st Cir. 2012); Powell v. Ga.-Pac. Corp., 119 F.3d 703, 706–07 (8th Cir. 1997).

B

Lowery argues that, even if permissible in some circumstances, cy pres relief was inappropriate here because it was feasible to distribute settlement funds directly to class members. The district court found otherwise “[g]iven the 60 million person class size and the $13 million Settlement Fund,” and because “it is unusually difficult and expensive to identify class members in this case.” Lowery argues that the district court applied the wrong standard for determining feasibility by asking “whether it is feasible to hand-deliver checks to every single class member” instead of focusing on “the ability of some class members to make a claim.” We disagree. Lowery cites no authority indicating that a district court must consider only whether settlement funds are distributable to “some” of a class, nor does he explain what proportion of a class would satisfy his proposed “some class members” test.

More fundamentally, even assuming that the subset of class members who claim payments would be small enough that the settlement fund could provide meaningful value to every claimant, Lowery does not identify a viable way for a claims administrator to verify any claimant’s entitlement to

IN RE GOOGLE INC. STREET VIEW LITIG. 23

settlement funds. 5 Google asserts that verifying that a person has a valid claim would require making three determinations: “(1) the [claimant] had maintained an unencrypted Wi-Fi network in the relevant period; (2) a Street View vehicle passed within range of that network; and (3) substantive communications (and not just technical network data) were transmitted within the precise fraction of a second when the Street View vehicle passed by and acquired payload data from the network.” Lowery does not dispute that a claims administrator would have to verify these three facts to determine whether a claim is valid, nor does he suggest any means of third-party claims verification besides the method the special master used—a process that took three years of intensive investigation and analysis to verify the claims of eighteen named plaintiffs. Instead,

5 Lowery argues that district courts have insisted on direct payments to class members in analogous cases involving very large classes. As an initial matter, presenting conflicting decisions from other district courts, without more, does not establish that the district court here abused its discretion. See Grant v. City of Long Beach, 315 F.3d 1081, 1091 (9th Cir. 2002) (“The abuse of discretion standard requires us to uphold a district court determination that falls within a broad range of permissible conclusions in the absence of an erroneous application of law.”). In any event, none of the examples Lowery cites involved the sort of technical challenges to identifying class members present here. See Fraley v. Facebook, Inc., 966 F. Supp. 2d 939, 940–49 (N.D. Cal. 2013) (involving no dispute that claims were readily verifiable); In re Carrier IQ, Inc. Consumer Priv. Litig., No. 12-md-02330-EMC, 2016 WL 4474366, at *3–4 (N.D. Cal. Aug. 25, 2016) (involving claims that were verifiable by reference to telephone numbers); In re Google Plus Profile Litig., No. 5:18-cv-06164-EJD, 2021 WL 242887 (N.D. Cal. Jan. 25, 2021) (involving claims that the defendant could easily verify by compiling a “class list”), appeal docketed, No. 21-15365 (9th Cir. Mar. 2, 2021). 24 IN RE GOOGLE INC. STREET VIEW LITIG.

Lowery asserts that the district court erred by refusing to allow claimants to “self-identify” as class members. 6

But his observation that “proof beyond a reasonable doubt is not required to ascertain a class member in a claims process” is misplaced. As the district court found, “[t]he only evidence” of class membership “is the intercepted data, and that evidence is not in the class member’s possession” or readily accessible to the claims administrator. Lowery offers no alternative way for claimants to determine with any degree of probability whether they are class members.

Because self-identification would be pure speculation, and any meaningful forensic verification of claims would be prohibitively costly and time-consuming, we affirm the district court’s finding that it was not feasible to verify class members’ claims as would be necessary to distribute funds directly to class members. Further, as “proof of individual claims would be burdensome [and] distribution of damages costly,” Lowery has not shown that the district court abused

6 Lowery observes that the district court permitted the named plaintiffs to proceed based on self-identification, and that it recognized Lowery’s own standing based on self-identification. He argues that by allowing some class members to self-identify but not others, the district court violated Rule 23’s requirement that settlements “treat[] class members equitably relative to each other.” Fed. R. Civ. P. 23(e)(2)(D). However, the district court permitted self-identification only at the pleading stage and when evaluating standing. It approved the settlement’s provision for service awards to the named plaintiffs, but service awards are compensation “for work done on behalf of the class” throughout litigation, not damages awarded for substantive claims. See Rodriguez v. W. Pub. Corp., 563 F.3d 948, 958 (9th Cir. 2009). Moreover, by the time the district court approved the service awards, the named plaintiffs’ claims were supported not just by their self- identification, but also by the special master’s extensive forensic analysis.

IN RE GOOGLE INC. STREET VIEW LITIG. 25

its discretion by approving the use of cy pres payments in the settlement. Lane, 696 F.3d at 819.

C

Alternatively, Lowery argues that if it was impossible to distribute settlement funds to class members, then class certification was an error of law because the class device was not superior to other available methods for fairly and efficiently adjudicating the controversy, as Rule 23(b)(3) requires. But cy pres provisions are tools for “distribut[ing] unclaimed or non-distributable portions of a class action settlement fund to the ‘next best’ class of beneficiaries.” Nachshin, 663 F.3d at 1036 (citation omitted). If it were feasible to distribute the settlement fund to the class members, a cy pres settlement would not be employed. Thus, in the guise of a Rule 23(b)(3) “superiority” argument, Lowery essentially repackages his argument that cy pres provisions, which by definition are used when settlement funds cannot be distributed to class members, are always improper. We have already rejected this argument, explaining that a blanket prohibition on so-called “cy pres- only” settlements, as Lowery advocates, would conflict with our precedent.

We addressed a similar argument in Briseno v. ConAgra Foods, Inc., 844 F.3d 1121 (9th Cir. 2017), a class action lawsuit against a cooking oil manufacturer for false labelling, in which the defendant opposed class certification, arguing that plaintiffs “did not propose any way to identify class members and cannot prove that an administratively feasible method exists because consumers do not generally save grocery receipts and are unlikely to remember details about individual purchases of a low-cost product like cooking oil,” so they could not verify their status as claimants. Id. at 1125. We rejected that argument, reasoning 26 IN RE GOOGLE INC. STREET VIEW LITIG.

that Rule 23 never “mention[s] ‘administrative feasibility’” and that recognizing a standalone “feasibility” requirement for class certification could render other Rule 23 provisions, such as “the likely difficulties in managing a class action,” Fed. R. Civ. P. 23(b)(3)(D), superfluous. Briseno, 844 F.3d at 1125–26.

Lowery maintains that he is not making “a stand-alone ascertainability argument of the sort repudiated by Briseno.” Instead, his argument, he says, is that “the superiority requirement of Rule 23(b)(3) demands the possibility of class benefit at the time of certification,” and that if it is practically impossible to identify absent class members at the time of certification, then a class action “cannot be a superior method of adjudicating th[e] controversy” because there is no possibility of providing meaningful relief. To be sure, if there were no possibility of providing meaningful relief via a class action settlement, Lowery’s point might be persuasive. But in making his argument, Lowery assumes a critical premise: that it is impossible to provide meaningful relief to a class when there is no feasible way of identifying class members.

This premise is not supported by our case law. In upholding the validity of cy pres arrangements, we have repeatedly recognized that class members do benefit—albeit indirectly—from a defendant’s payment of funds to an appropriate third party. See Lane, 696 F.3d at 819 (describing cy pres remedy as “a settlement structure wherein class members receive an indirect benefit (usually through defendant donations to a third party) rather than a direct monetary payment”); Nachshin, 663 F.3d at 1038 (“In the context of class action settlements, a court may employ the cy pres doctrine to put the unclaimed fund to its next best compensation use, e.g., for the aggregate, indirect,

IN RE GOOGLE INC. STREET VIEW LITIG. 27

prospective benefit of the class.” (internal quotation marks and citation omitted)).

Indeed, the factors that guide judicial oversight of cy pres settlement provisions—whether the distributions “account for the nature of the plaintiffs’ lawsuit, the objectives of the underlying statutes, and the interests of the silent class members”—are designed to ensure that cy pres payments particularly “benefit the plaintiff class.” Id. at 1036, 1040. If a cy pres award has a “direct and substantial nexus to the interests of absent class members,” Lane, 696 F.3d at 821, as it must under our precedents, then it necessarily prioritizes class members’ interests, even if it also provides a diffuse benefit to society at large. 7 Thus, the infeasibility of distributing settlement funds directly to class members does not preclude class certification.

D

Accordingly, we next consider whether the settlement agreement provides sufficient value to the class, in the form of both cy pres relief and injunctive relief, to be “fair, reasonable, and adequate.” Fed. R. Civ. P. 23(e)(2). We hold that the district court did not err by concluding that it does.

7 Lowery cites In re Hotel Telephone Charges, 500 F.2d 86 (9th Cir. 1974), to support his argument that “[w]hen a ‘great variety’ of individualized determinations preclude class benefit, class certification should be denied.” But In re Hotel Telephone Charges is inapposite: it simply held that a class action involving “over six hundred defendants,” “millions of plaintiffs,” and “a great variety of individual questions” did not satisfy the requirements of predominance and manageability, not that an inability to identify class members precludes certification. Id. at 90– 92. 28 IN RE GOOGLE INC. STREET VIEW LITIG.

The injunctive relief in the settlement agreement, which required Google to “destroy all Acquired Payload Data,” refrain from collecting or storing additional payload data through Street View without notice and consent, and comply with other AVC provisions specifically referenced in the settlement agreement, largely duplicated Google’s obligations under the AVC. However, the injunctive relief extends beyond Google’s AVC obligations. It requires Google to maintain its compliance until five years from final settlement approval—that is, at least two years longer than the AVC required. Moreover, the injunctive relief in the settlement requires Google to post additional educational material online that the AVC did not require. The district court found that this injunctive relief offered “adequate, if not the main benefit to the class.” Considering the unique challenges plaintiffs would have faced in proving their claims, we hold that the district court did not err by concluding this injunctive relief, together with the indirect benefits conferred by the cy pres provisions, was fair, reasonable, and adequate compensation to the class members.

In Campbell v. Facebook, Inc., we considered a settlement agreement that included injunctive relief requiring “Facebook [to] make a plain English disclosure on its Help Center page” for one year, informing users about its “message monitoring practices.” 951 F.3d 1106, 1123 (9th Cir. 2020). We affirmed the district court’s finding that this relief “had value to absent class members,” reasoning that it “ma[de] it less likely that users will unwittingly divulge private information to Facebook or third parties in the course of using Facebook’s messaging platform.” Id. We explained that “the relief provided to the class cannot be assessed in a vacuum” and that “the class did not need to receive much for the settlement to be fair because the class

IN RE GOOGLE INC. STREET VIEW LITIG. 29

gave up very little.” Id. We emphasized that the “class members’ claims were weak enough that the class was fairly likely to end up receiving nothing at all had this litigation proceeded further,” and that the injunctive relief provided a benefit that, while very small, was more than “nothing.” Id. We also affirmed the district court’s finding that this relief was not “duplicative” of a “change Facebook had already made,” because it required the disclosure “to stay on display for a year” and required an explanation written “in plain English.” Id. at 1123 n.12.

Similarly, although the injunctive relief here requires relatively little of Google, it does extend Google’s obligations beyond those in the AVC. Moreover, it does so in exchange for class members’ relinquishment of legal claims that might have been quite difficult to prove and would likely have yielded very little per class member in damages. As the district court observed, the context of this settlement was “a case in which a vast but nonetheless difficult-to-identify class of people suffered intangible injury, and minimal damages.”

The Arizona Attorney General argues that “the privacy landscape for technology companies has fundamentally changed” since 2013 and that companies like Google have “been forced to focus on user-privacy questions” for reasons independent of the Street View litigation. Given these changes, he asserts that “there can be no doubt that Google will be independently maintaining privacy training, privacy- related advertising, and management-level attention to questions of user privacy and unauthorized collection or disclosure of user information.” To that point, we have recognized that injunctive relief in a class action settlement is illusory if it “does not obligate [a defendant] to do anything it was not already doing,” or if it merely requires a 30 IN RE GOOGLE INC. STREET VIEW LITIG.

defendant to “continue” practices “it voluntarily adopted” before the settlement. Koby v. ARS Nat’l Servs., Inc., 846 F.3d 1071, 1080 (9th Cir. 2017). Here, however, the district court specifically noted that the injunctive relief required Google to make “changes . . . it would not have made without the settlement,” which would provide “some value to the class.” On clear error review, we will not second-guess the district court’s factual findings based on speculation about what Google might hypothetically have done absent the settlement agreement. Campbell, 951 F.3d at 1123.

Viewing the modest injunctive relief together with the indirect benefits the class members enjoy through the cy pres provision, we affirm the district court’s finding that the settlement was “fair, reasonable, and adequate.” Fed. R. Civ. P. 23(e)(2).

E

Lowery argues that the settlement violates the First Amendment’s prohibition on compelled speech by distributing class settlement funds to organizations “that take lobbying positions adverse to” his own interests and beliefs. The district court found no First Amendment violation, reasoning that “[t]he settlement agreement between the parties is not state action, . . . and class members ha[ve] the opportunity to exclude themselves from the settlement.”

As a threshold matter, the parties dispute whether a district court’s approval of a settlement agreement constitutes state action such that it implicates First Amendment protections. See IMDb.com Inc. v. Becerra, 962 F.3d 1111, 1120 (9th Cir. 2020) (“Private parties may freely bargain with each other to restrict their own speech, and those agreements may be enforced, without implicating

IN RE GOOGLE INC. STREET VIEW LITIG. 31

the First Amendment.”). We do not decide today whether, or under what circumstances, a district court’s approval of a class action settlement agreement is “state action” for purposes of the First Amendment. Instead, we hold that the settlement agreement does not compel class members to subsidize third-party speech because any class member who does not wish to “subsidize speech by a third party that he or she does not wish to support,” Harris v. Quinn, 573 U.S. 616, 656 (2014), can simply opt out of the class. 8

Lowery cites Janus v. American Federation of State, County, and Municipal Employees, Council 31, 138 S. Ct. 2448, 2459–60 (2018), and Knox v. Service Employees International Union, Local 1000, 567 U.S. 298, 321–22 (2012), to argue that “silence is not consent and a waiver of First Amendment rights cannot be presumed.” It is not entirely clear what connection Lowery intends to draw between these decisions and his First Amendment arguments, but Janus and Knox are inapposite. The Supreme Court held in Janus that states cannot require paycheck deductions for public employees to subsidize unions that engage in advocacy those employees find objectionable. It explained, “[n]either an agency fee nor any other payment to the union may be deducted from a nonmember’s wages, nor may any other attempt be made to collect such a payment, unless the employee affirmatively consents to pay.” 138 S. Ct. at 2486. But Janus involved mandatory deductions from an employee’s paycheck, while the settlement here

8 The district court found that the parties’ notice to the class members, as approved and directed by the court, complied with Rule 23(c), (e), and (h) and the Due Process Clause, and provided notice of the lawsuit, the settlement, and the class members’ rights, including their right to object to, or opt out of, the settlement. Lowery does not challenge this finding. 32 IN RE GOOGLE INC. STREET VIEW LITIG.

involves funds that, regardless of the cy pres provisions, could not feasibly be paid to class members. See id. (“Unless employees clearly and affirmatively consent before any money is taken from them, this standard cannot be met.”). Knox is similarly inapposite because it dealt with whether a union must provide fresh notice and seek affirmative consent before exacting funds from nonmembers through paycheck deductions. 567 U.S. at 321–22.

Lowery observes that class members’ decisions to opt out “wouldn’t reduce the contribution in the class members’ name[s].” But opting out does not entitle a class member to his pro rata portion of a settlement. On the contrary, it entitles him to retain his legal claim by not participating in the settlement. See Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 176 (1974). If Lowery opts out, he will have disassociated himself from the subsidization of the cy pres recipients’ speech. He will also have disclaimed any interest he might have had in the settlement funds as a class member. Thus, he would have no further interest in the terms of the settlement agreement.

F

Lowery also argues that the district court abused its discretion by approving cy pres recipients who had a “significant prior affiliation” with defense counsel and class counsel. In particular, he argues that one of the recipients, EPIC, “supported plaintiffs in an earlier appeal in this case,” that four other cy pres recipients “previously received Google cy pres money” in unrelated cases, that “[m]any of the recipients had received cy pres funds from other class actions involving big tech firms,” and that the ACLU “had a pre-existing relationship with class counsel.” These arguments are unconvincing. We have never held that merely having previously received cy pres funds from a

IN RE GOOGLE INC. STREET VIEW LITIG. 33

defendant, let alone other defendants in unrelated cases, disqualifies a proposed recipient for all future cases. Moreover, we have affirmed cy pres provisions involving much closer relationships between recipients and parties than anything Lowery alleges here.

In Lane, the district court approved a settlement agreement that included a cy pres payment of approximately $6.5 million to “a new entity whose sole purpose was to designate fund recipients consistent with [the] mission to promote the interests of online privacy and security.” 696 F.3d at 817. This entity “would be run by a three- member board of directors,” one of whom was Facebook’s own Director of Public Policy, as well as a “Board of Legal Advisors,” which “consist[ed] of counsel for both the plaintiff class and Facebook.” Id. at 817–18. Several objectors challenged the settlement agreement, arguing that the presence of a high-level Facebook employee on the foundation’s board of directors “creates an unacceptable conflict of interest” and that “the settling parties’ decision to disburse settlement funds through an organization with such structural conflicts does not provide the ‘next best distribution’ of those funds and thus is categorically an improper use of the cy pres remedy.” Id. at 820. We disagreed, explaining:

We do not require as part of [the cy pres] doctrine that settling parties select a cy pres recipient that the court or class members would find ideal. On the contrary, such an intrusion into the private parties’ negotiations would be improper and disruptive to the settlement process. The statement . . . in our case law that a cy pres remedy must be the “next best distribution” of settlement funds 34 IN RE GOOGLE INC. STREET VIEW LITIG.

means only that a district court should not approve a cy pres distribution unless it bears a substantial nexus to the interests of the class members . . . .

Id. at 820–21.

Lowery argues that Lane only dealt with conflicts between defendants and cy pres recipients, and that it “has no bearing on a distribution that raises conflicts between class counsel and the recipient.” This assertion is incorrect, as the cy pres arrangement in Lane also provided for class counsel to sit on the recipient’s board of legal advisors. Id. at 817–18.

Citing the American Law Institute’s Principles of the Law of Aggregate Litigation and out-of-circuit authority, Lowery argues that “[t]he correct legal standard” for approving a proposed cy pres recipient is whether “any party has any significant prior affiliation with the intended recipient that would raise substantial questions about whether the award was made on the merits.” But we have never adopted Lowery’s expansive proposed test, and Lowery cites no binding authority that would have precluded the district court from approving the cy pres recipients here.

Lowery cites Radcliffe v. Experian Information Solutions Inc., 715 F.3d 1157 (9th Cir. 2013), to argue that there existed a “potential conflict of interest of class counsel in favoring a former client and co-counsel” (apparently EPIC and the ACLU) over class members. But Radcliffe is entirely inapposite. We held in that case, relying on California law governing attorney ethics, that “conditional incentive awards” to class representatives “caused the interests of the class representatives to diverge from the interests of the class because the settlement agreement told

IN RE GOOGLE INC. STREET VIEW LITIG. 35

class representatives that they would not receive incentive awards unless they supported the settlement.” Id. at 1161. Lowery points to no such improper incentives here.

He also cites Nachshin, 663 F.3d at 1039, but nothing in that decision suggests the sort of scrutiny that Lowery argues we should apply to the cy pres settlement here. In Nachshin, we explained that “[w]hen selection of cy pres beneficiaries is not tethered to the nature of the lawsuit and the interests of the silent class members, the selection process may answer to the whims and self interests of the parties, their counsel, or the court.” Id.; see id. (“To remedy some of these concerns, we held in Six Mexican Workers that cy pres distribution must be guided by (1) the objectives of the underlying statute(s) and (2) the interests of the silent class members.”). The district court’s approval of the cy pres recipients comported with those standards, and we find no abuse of discretion.

G

Lowery argues that the district court abused its discretion by “blindly apply[ing]” a 25% benchmark for attorneys’ fees without regard for the actual benefit the settlement conferred on the class. We disagree.

The district court devoted several pages of analysis to the issue of attorneys’ fees, correctly beginning with the premise that “in the Ninth Circuit, the ‘benchmark’ fee award is 25%, which can be adjusted upward or downward based on the circumstances of the case.” See Fischel v. Equitable Life Assurance Soc’y of U.S., 307 F.3d 997, 1006 (9th Cir. 2002) (“We have established a 25 percent ‘benchmark’ in percentage-of-the-fund cases that can be ‘adjusted upward or downward to account for any unusual circumstances involved in [the] case.’” (alteration in original) (citation 36 IN RE GOOGLE INC. STREET VIEW LITIG.

omitted)). It found that “the overall result and benefit to the class from the litigation supports the requested percentage” of 25% because the cy pres relief “benefits the class members by serving the goals of this litigation and the [Electronic Communications Privacy Act].”

The district court specifically considered Lowery’s argument that the benchmark should be reduced to reflect the lack of direct monetary payments to class members. It rejected this argument, reasoning that “where the settlement fund is non-distributable, counsel should not be penalized for fashioning a cy pres-only settlement that stands to accomplish some good.” The district court noted several other factors supporting a 25% benchmark: that the case “required skill and expertise,” “involved novel issues,” took “nearly ten years of work,” and was “risky” for counsel to take on. The court also conducted a lodestar analysis and determined that the benchmark-based award would be lower than a lodestar-based award, “strongly suggest[ing] the reasonableness of the requested fee.”

The district court’s reasoning makes clear that this was not a “blind” application of a benchmark to the circumstances of the case. And Lowery does not challenge any of the district court’s specific factual findings supporting its fee award. Instead, he urges us to hold as a general matter that “it [is] inappropriate to value cy pres on a dollar-for- dollar basis” equivalent to direct monetary relief to class members. See In re Heartland Payment Sys., Inc. Customer Data Sec. Breach Litig., 851 F. Supp. 2d 1040, 1077 (S.D. Tex. 2012). Certainly, a district court must consider a settlement’s benefit to the class in determining appropriate attorneys’ fees, and thus, attorneys’ fees are not solely a function of the size of a settlement fund. See e.g., In re HP Inkjet Printer Litig., 716 F.3d 1173, 1182, 1185–87 (9th Cir.

IN RE GOOGLE INC. STREET VIEW LITIG. 37

2013) (“Plaintiffs attorneys don’t get paid simply for working; they get paid for obtaining results.”); In re Baby Prods. Antitrust Litig., 708 F.3d 163, 170 (3d Cir. 2013) (“[W]e confirm that courts need to consider the level of direct benefit provided to the class in calculating attorneys’ fees.”).

But there is no uniform rule that district courts must discount the value of any cy pres relief, regardless of the feasibility of distribution to class members or other relevant circumstances. Indeed, we have repeatedly approved attorneys’ fees for cy pres settlements in proportions similar to the award here. See Google Referrer, 869 F.3d at 747–48 (affirming fee award of 25% of cy pres settlement); Lane, 696 F.3d at 818, 823–24 (affirming lodestar-based fee award of 24.89% of total cy pres settlement); see also Campbell, 951 F.3d at 1115, 1126–27 (rejecting argument that $3.89 million fee award was excessive when settlement provided only injunctive relief). Other circuits have similarly declined to adopt such a rule. See Baby Prods., 708 F.3d at 178 (“We think it unwise to impose . . . a rule requiring district courts to discount attorneys’ fees when a portion of an award will be distributed cy pres.”).

Lowery argues that by failing to decrease the benchmark given the lack of direct payments to class members, we would permit “perverse incentives [that] will result in a disproportionate number of cy pres-only settlements.” But our approach does not “make[] class counsel financially indifferent between a settlement that awards cash directly to class members and a cy pres-only settlement,” as Lowery warns, because it does take into account the benefit to class members. And, “[o]f course, the percentage may be adjusted to account for any unusual circumstances.” Williams v. MGM-Pathe Commc’ns Co., 129 F.3d 1026, 1027 (9th Cir. 38 IN RE GOOGLE INC. STREET VIEW LITIG.

1997). Thus, if class counsel fails “to seek an award that adequately prioritizes direct benefit to the class,” it might be “appropriate for the court to decrease the fee award.” Baby Prods., 708 F.3d at 178. Doing so might also be appropriate when “a cy pres . . . settlement . . . has a tenuous relationship to the class allegedly damaged by the conduct in question,” or when it appears that the settlement “serves only the ‘self- interests’ of the attorneys and the parties, and not the class.” Dennis v. Kellogg Co., 697 F.3d 858, 868 (9th Cir. 2012). But here, the district court properly considered all relevant circumstances, including the value to the class members, and concluded that a 25% benchmark was appropriate. We affirm the district court’s fee award.

H

Finally, Lowery argues that class certification was inappropriate because, by deciding to settle, class counsel and the class representatives breached their fiduciary duties. See Fed. R. Civ. P. 23(a)(4) (conditioning class certification on a finding that “the representative parties will fairly and adequately protect the interests of the class”); id. 23(g)(4) (“Class counsel must fairly and adequately represent the interests of the class.”). Lowery asserts that under these fiduciary duties, class counsel and representatives cannot “agree[] to accept excessive fees and costs to the detriment of [absent] class plaintiffs.” See Lobatz v. U.S. W. Cellular of Cal., Inc., 222 F.3d 1142, 1147 (9th Cir. 2000).

Lowery’s fiduciary duty arguments are simply a repackaging of his other arguments against the settlement: he asserts that “class counsel structure[d] a settlement to benefit third parties over any single absent class member,” that the settlement included excessive attorneys’ fees and lacked “any benefit for the class,” and that counsel should have advised “absent class members of the superiority of

IN RE GOOGLE INC. STREET VIEW LITIG. 39

opting out en masse.” Because we affirm the district court’s finding that the settlement does provide adequate value to the class, and because there is no indication that counsel accepted excessive attorneys’ fees or favored third parties over class members, we hold that class counsel and class representatives did not breach their fiduciary duties by entering the settlement.

V

We AFFIRM the district court’s order certifying the class, approving the settlement agreement, and awarding attorneys’ fees.

BADE, Circuit Judge, concurring:

The district court correctly applied our circuit’s law and did not err in certifying the class for settlement purposes or approving the proposed settlement agreement. Indeed, in varying contexts, we have upheld class action settlements that provided cy pres awards to third parties in lieu of damages for the class members. See In re Google Referrer Header Priv. Litig., 869 F.3d 737 (9th Cir. 2017), vacated and remanded on other grounds, Frank v. Gaos, 139 S. Ct. 1041 (2019); Lane v. Facebook, Inc., 696 F.3d 811 (9th Cir. 2012). And we have implicitly approved the use of cy pres awards even when rejecting settlements on other grounds. See Nachshin v. AOL, LLC, 663 F.3d 1034 (9th Cir. 2011); Six (6) Mexican Workers v. Ariz. Citrus Growers, 904 F.2d 1301 (9th Cir. 1990). Because I am constrained to follow these precedents, I authored and joined the majority opinion.

But as Chief Justice Roberts has noted, “fundamental” questions about “the use of [cy pres] remedies in class action 40 IN RE GOOGLE INC. STREET VIEW LITIG.

litigation” remain unanswered. See Marek v. Lane, 134 S. Ct. 8, 9 (2013) (Roberts, C.J., respecting the denial of certiorari) (explaining that, among other questions, the Court has not yet addressed “when, if ever, such relief should be considered” and “how to assess its fairness as a general matter”). Therefore, I write separately to express some general concerns about cy pres awards.

First, I recognize that “federal courts frequently use the cy pres doctrine ‘in the settlement of class actions where the proof of individual claims would be burdensome or distribution of damages costly.’” Nachshin, 663 F.3d at 1038 (quoting Six Mexican Workers, 904 F.2d at 1305); see also A.L.I., Principles of the Law of Aggregate Litigation § 3.07(c) (2010) (approving cy pres settlement provisions “[i]f the court finds that individual distributions are not viable”); William B. Rubenstein, 4 Newberg on Class Actions § 12:26 (5th ed. 2011) [hereinafter Newberg] (same). I also recognize that cy pres awards present a practical solution for settling cases “[w]hen a class action involves a large number of class members but only a small individual recovery, [and] the cost of separately proving and distributing each class member’s damages may so outweigh the potential recovery that the class action becomes unfeasible.” Six Mexican Workers, 904 F.2d at 1305. I question, however, whether we have allowed these practical advantages to inappropriately displace other concerns implicated by cy pres awards.

Such concerns, which have been ably identified by jurists and commentators, include: conflicts of interest between class counsel and absent class members, Keepseagle v. Perdue, 856 F.3d 1039, 1060 (D.C. Cir. 2017) (Brown, J., dissenting); In re Baby Prods. Antitrust Litig., 708 F.3d 163, 173 (3d Cir. 2013); Jay Tidmarsh, Cy Pres

IN RE GOOGLE INC. STREET VIEW LITIG. 41

and the Optimal Class Action, 82 Geo. Wash. L. Rev. 767, 772, 782 (2014); incentives for collusion between defendants and class counsel, Lane, 696 F.3d at 829–30 (Kleinfeld, J., dissenting); the role of the court and the parties in shaping a cy pres remedy and the potential appearance of impropriety, S.E.C. v. Bear, Stearns & Co., 626 F. Supp. 2d 402, 415 (S.D.N.Y. 2009); Goutam U. Jois, The Cy Pres Problem and the Role of Damages in Tort Law, 16 Va. J. Soc. Pol’y & L. 258, 265–66 (2008); the use of Rule 23 of the Federal Rules of Civil Procedure, “a wholly procedural device,” to shape substantive rights, arguably in violation of Article III, the Rules Enabling Act, 1 and the separation of powers doctrine, Klier v. Elf Atochem N. Am., Inc., 658 F.3d 468, 481 (5th Cir. 2011) (Jones, J., concurring) (citing Martin H. Redish et al., Cy Pres Relief and the Pathologies of the Modern Class Action: A Normative and Empirical Analysis, 62 Fla. L. Rev. 617, 623, 641 (2010)); “whether a cy pres award can ever be used as a substitute for actual damages,” Molski v. Gleich, 318 F.3d 937, 954 (9th Cir. 2003), overruled on other grounds by Dukes v. Wal-Mart Stores, Inc., 603 F.3d 571, 617 (9th Cir. 2010) (en banc); the propriety of importing a doctrine originating in trust law into the context of class action litigation, Klier, 658 F.3d at 480 (Jones, J., concurring); In re Pet Food Prods. Liab. Litig., 629 F.3d 333, 363 (3d Cir. 2010) (Weis, J., concurring in part and dissenting in part);

1 In Wal-Mart Stores, Inc. v. Dukes, the Court cautioned that “the Rules Enabling Act forbids interpreting Rule 23 to ‘abridge, enlarge or modify any substantive right.’” 564 U.S. 338, 367 (2011) (quoting 28 U.S.C. § 2072(b)); see also Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins. Co., 559 U.S. 393, 408 (2010) (“A class action, no less than traditional joinder (of which it is a species), merely enables a federal court to adjudicate claims of multiple parties at once, instead of in separate suits. And like traditional joinder, it leaves the parties’ legal rights and duties intact and the rules of decision unchanged.”). 42 IN RE GOOGLE INC. STREET VIEW LITIG.

Redish, supra, at 630; and whether class action litigation is superior to other methods of adjudication if parties must resort to cy pres relief, Frank, 139 S. Ct. at 1047 (Thomas, J., dissenting). I do not expand on those justified concerns here. Instead, I focus on the predicate of cy pres settlement provisions—the theory of indirect benefit to the class members.

Courts have upheld cy pres awards based on the premise that they provide an indirect benefit to the class when a direct monetary payment is not feasible. See Lane, 696 F.3d at 819; Nachshin, 663 F.3d at 1038; Six Mexican Workers, 904 F.2d at 1305; Klier, 658 F.3d at 475. Institutional commentators and treatises have also embraced this theory of indirect benefit. See A.L.I., supra, at § 3.07 cmt. b (“Cy pres is preferable to other options available to a court when direct distributions are not viable.”); Newberg, supra, at § 12:26 (“[C]y pres distributions provide indirect compensation to the plaintiff class by funding activities that are in the class’s interest.”).

But there is an increasing skepticism about whether cy pres provisions actually provide an indirect benefit to class members. See Frank, 139 S. Ct. at 1047 (Thomas, J., dissenting) (“[C]y pres payments are not a form of relief to the absent class members and should not be treated as such . . . .”); Lane, 696 F.3d at 830 (Kleinfeld, J., dissenting) (“It is hard to imagine a real client saying to his lawyer, ‘I have no objection to the defendant paying you a lot of money in exchange for agreement to seek nothing for me.’”); Molski, 318 F.3d at 954 (stating that “it seems somewhat distasteful to allow a corporation to fulfill its legal and equitable obligations through tax-deductible donations to third parties”); In re Baby Prods. Litig., 708 F.3d at 173 (concluding that cy pres settlements are permissible, but

IN RE GOOGLE INC. STREET VIEW LITIG. 43

noting that they substitute “an indirect benefit that is at best attenuated and at worse illusory” for compensatory damages); Klier, 658 F.3d at 482 (Jones, J., concurring) (“Our adversarial system should not effectuate transfers of funds from defendants beyond what they owe to the parties in judgments or settlements.”); Mirfasihi v. Fleet Mortg. Corp., 356 F.3d 781, 784 (7th Cir. 2004) (explaining that in cy pres settlements “[t]here is no indirect benefit to the class from the defendant’s giving the money to someone else”); Six Mexican Workers, 904 F.2d at 1312 (Fernandez, J., concurring) (“[Cy pres] is a very troublesome doctrine, which runs the risk of being a vehicle to punish defendants in the name of social policy, without conferring any particular benefit upon any particular wronged person.”); Redish, supra, at 623 (“Cy pres creates the illusion of class compensation. It is employed when—and only when— absent its use, the class proceeding would be little more than a mockery.”). And, despite the acceptance of the theory of indirect benefit, there is, in my view, a compelling argument that class members receive no benefit at all from a settlement that extinguishes their claims without awarding them any damages, and instead directs money to groups whose interests are purportedly aligned with the class members, but whom they have likely never heard of or may even oppose.

Moreover, even if we accept the premise that cy pres awards provide value to the public at large, there is practical appeal in the argument that such settlements provide no unique consideration to class members because they receive the same generalized benefits as non-class-members and opt-outs. Indeed, cy pres settlements arguably benefit opt- outs more than class members because opt-outs reap any positive externalities of the settlement provisions while 44 IN RE GOOGLE INC. STREET VIEW LITIG.

retaining the value of the claims that the settlement extinguished for class members. 2

I am therefore not convinced that cy pres awards to uninjured third parties should qualify as an indirect benefit to injured class members, and I am concerned that “the ‘cy pres’ remedy . . . is purely punitive,” Mirfasihi, 356 F.3d at 784, with defendants paying millions of dollars in what are essentially civil fines to class counsel and third parties while providing no compensation to injured class members. See Klier, 658 F.3d at 481 (Jones, J., concurring) (citing Redish, supra, at 623); see also Six Mexican Workers, 904 F.2d at 1312 (Fernandez, J., concurring) (“[Cy pres’] use may well amount to little more than an exercise in social engineering by a judge, who finds it offensive that defendants have profited by some wrongdoing, but who has no legitimate plaintiff to give the money to.”); Newberg, supra, at § 12:26 (stating that one purpose of cy pres distributions is to “ensure that the defendant is disgorged of a sum certain, even if that money does not compensate class members directly”).

I further question whether cy pres awards are inherently unfair when the class receives no meaningful relief in exchange for their claims, see Fed. R. Civ. P. 23(e)(2), and whether such awards can be justified given the serious

2 In cases where a class settlement provides injunctive and cy pres relief, but no damages for class members, the concern that non-class- members and opt-outs fare better than class members could be mitigated by certifying injunctive and declaratory relief classes under Rule 23(b)(2), without cy pres awards and without requiring class members to release damages claims, rather than damages classes under Rule 23(b)(3). Cf. Campbell v. Facebook, Inc., 951 F.3d 1106, 1113–15, 1124 (9th Cir. 2020) (affirming approval of injunctive-relief-only class settlement that did not release class members’ damages claims).

IN RE GOOGLE INC. STREET VIEW LITIG. 45

ethical, procedural, and constitutional problems that others have identified. Therefore, I respectfully submit that it is time we reconsider the practice of cy pres awards.