Opinion
20-cv-04011-VC
08-17-2023
KIMBERLY CARLESTE NEWMAN, et al., Plaintiffs, v. GOOGLE LLC, et al., Defendants.
ORDER GRANTING MOTION TO DISMISS THE FIFTH AMENDED COMPLAINT RE: DKT. NO. 146
VINCE CHHABRIA UNITED STATES DISTRICT JUDGE
The plaintiffs in this proposed class action are African American and Hispanic content creators who allege that YouTube's content-moderating algorithm discriminates against them based on their race. Specifically, they allege that their YouTube videos are restricted when similar videos posted by white users are not. This differential treatment, they believe, violates a promise by YouTube to apply its Community Guidelines (which govern what type of content is allowed on YouTube) “to everyone equally-regardless of the subject or the creator's background, political viewpoint, position, or affiliation.” The plaintiffs thus bring a breach of contract claim against YouTube (and its parent company, Google). They also bring claims for breach of the implied covenant of good faith and fair dealing, unfair competition, accounting, conversion, and replevin.
YouTube's motion to dismiss is granted. Although the plaintiffs have adequately alleged the existence of a contractual promise, they have not adequately alleged a breach of that promise. The general idea that YouTube's algorithm could discriminate based on race is certainly plausible. But the allegations in this particular lawsuit do not come close to suggesting that the plaintiffs have experienced such discrimination. The breach of contract claim must therefore be dismissed. As for the remaining claims, the current version of the complaint adds nothing to remedy the deficiencies identified in prior rulings, and so those claims must also be dismissed. See Newman v. Google LLC, No. 20-cv-4011-VC, 2022 WL 2556862, at * 1-*3 (N.D. Cal. July 8, 2022); Newman v. Google LLC, No. 20-cv-4011-VC, 2022 WL 20438233, at *2-*3 (N.D. Cal. Nov. 28, 2022).
The plaintiffs have had six opportunities to adequately plead their claims, and counsel made clear at the hearing for this motion that this is “the best I got.” Dkt. No. 154 at 31; see also Dkt. No. 154 at 17, 34. Dismissal is therefore with prejudice.
I
A. Procedural History
The plaintiffs brought this case against YouTube and Google in June 2020, and it was assigned to Judge Koh. Early on, the case had a constitutional orientation-the plaintiffs alleged that YouTube's racially discriminatory treatment violated the First Amendment of the United States Constitution as well as Article I, Section 2 of the California Constitution. See Dkt. No. 27. They also pressed claims for discrimination under 42 U.S.C. § 1981 and the Unruh Civil Rights Act, for false advertising, unfair business practices, breach of the implied covenant of good faith and fair dealing, promissory estoppel, breach of contract, conversion, replevin, accounting, and for a declaratory judgment as to the scope and constitutionality of section 230 of the Communications Decency Act. See Dkt. No. 27.
That version of the complaint was dismissed, largely because YouTube is not a state actor, and because the plaintiffs failed to allege that any discrimination by YouTube was intentional, as required for the section 1981 claim. See Newman v. Google LLC, No. 20-cv-4011-LHK, 2021 WL 2633423, at *5-*9 (N.D. Cal. June 25, 2021); see also id. at *12-*13 (declining to exercise supplemental jurisdiction over the state law claims after dismissing the federal claims). Three more amendments and two dismissals later (by the undersigned judge after the case was reassigned), this case has shed its intentional discrimination and constitutional claims, becoming-first and foremost-a breach of contract dispute.
The plaintiffs have adequately pleaded jurisdiction under the Class Action Fairness Act, and so their state law claims are properly before the Court. See Newman v. Google LLC, No. 20-cv-4011-VC, 2022 WL 2556862, at *1 (N.D. Cal. July 8, 2022).
B. The Allegations
To upload a video to YouTube or create a YouTube channel, users must agree to the platform's Terms of Service, which incorporate the Community Guidelines. The Community Guidelines govern what type of content is allowed on YouTube. They address topics such as graphic content, vulgar language, hate speech, and misinformation, to name only a few. While YouTube reserves broad discretion with respect to content moderation on its platform, the Overview section of the Community Guidelines contains the following language: “We enforce these Community Guidelines using a combination of human reviewers and machine learning, and apply them to everyone equally-regardless of the subject or the creator's background, political viewpoint, position, or affiliation.” Dkt. No. 144-8 at 3.
The plaintiffs allege that this language constitutes an enforceable promise to engage in race-neutral content moderation. They further allege that YouTube's machine-learning algorithm violates this contractual promise by subjecting their videos to adverse moderation decisions while treating similar videos posted by white users more favorably. This differential treatment occurs, according to the complaint, because “YouTube uses [its] algorithm[] to get the information that advertisers want in order to gather and analyze information about creators and viewers based on Plaintiffs' Identities, including Race.” Dkt. No. 144 at 22. “[T]his aggregated personal data is then embedded in [the] algorithm[],” resulting in an algorithm “tainted with and by identity based information and bias.” Dkt. No. 144 at 8, 34.
The plaintiffs allege that three different types of content-moderation decisions are infected with race discrimination: removal of videos from the platform; restrictions on who can view videos on the platform; and demonetization of videos, which is when YouTube prevents users from making money off their content. Although the plaintiffs purport to base their breach of contract claim on all three types of decisions, the complaint focuses almost entirely on restrictions-that is, decisions by YouTube's algorithm to tag the plaintiffs' videos for exclusion in “Restricted Mode.” Restricted Mode is a setting that shields users who enable it-for example, a school administration wanting to protect its underaged students-from content that has been tagged by YouTube's algorithm as inappropriate for certain audiences. As to removals and demonetizations, the complaint offers conclusory allegations lacking specific factual content. See, e.g., Dkt. No. 144 at 49, 51, 56. Indeed, only two specific removals are alleged, but one appears to have taken place before the relevant promise was made, and no basis is offered to support the conclusion that the other took place because of the plaintiff's race. See Dkt. No. 144 at 38, 40. This ruling therefore focuses on the allegations regarding restrictions.
II
A. The Promise
To state a claim for breach of contract under California law, the plaintiffs must adequately allege: (1) the existence of a contract; (2) performance by the plaintiffs or excuse for nonperformance; (3) breach by the defendant; and (4) damages resulting from the defendant's breach. Oasis W. Realty, LLC v. Goldman, 51 Cal.4th 811, 821 (2011); First Commercial Mortgage Company v. Reece, 89 Cal.App.4th 731, 745 (2001).
As to the first element, YouTube concedes that its Terms of Service, which incorporate the Community Guidelines, represent a valid agreement supported by adequate consideration and mutually assented to by the parties. See Dkt. No. 148 at 7. But it argues that the specific language relied on by the plaintiffs-the language pledging to treat everyone equally regardless of race- is not enforceable. For support, YouTube turns to a recent decision by the California Court of Appeal: Prager University v. Google LLC, 85 Cal.App. 5th 1022 (2022) (“Prager II”). YouTube argues that Prager II resolves the matter because it held that YouTube reserves “unfettered and unilateral discretion” as to content moderation and rejected the argument that YouTube's Terms of Service contemplate “identity neutral content moderation and access” or that “defendants are bound by a contractual promise to filter content neutrally.” Id. at 1039.
But elsewhere in its briefs, YouTube asserts that the language relevant to this case was not added to the Community Guidelines until late 2021. See Dkt. Nos. 146 at 11-12; 156 at 22. Taking judicial notice of earlier versions of the Community Guidelines filed in this case and in a similar case in this district, YouTube appears to be correct that the relevant language was not added until sometime between August 12, 2021, and November 16, 2021. Compare Dkt. No. 29-5 (November 1, 2020, version without the relevant language), and Divino Group LLC v. Google LLC, No. 19-cv-4749-VKD, Dkt. No. 77-3 (August 12, 2021, version without the relevant language), with Dkt. No. 144-8 (November 16, 2021, version with the relevant language). This explains why Prager II contains no discussion of this language-the complaint under review in that case was filed in May 2019, years before the relevant language was added to the Community Guidelines. See Prager University v. Google LLC, No. 19-cv-340667, 2019 WL 8645795 (Cal. Super. May 20, 2019) (First Amended Complaint). Prager II is therefore largely irrelevant.
YouTube also relies on Block v. eBay, Inc., 747 F.3d 1135 (9th Cir. 2014). There, the Ninth Circuit held that a provision in eBay's User Agreement explaining that “[eBay is] not involved in the actual transaction between buyers and sellers” did not constitute an enforceable promise by eBay. Id. at 1138. The court reasoned that the provision contained no future-tense, promissory language and could only be interpreted as “a general description of how eBay's auction system works,” not as a promise not to be involved in the transaction. Id.
Block does not stand for the proposition that only future-tense statements can constitute binding promises. See id. (citing cases where present-tense statements constituted binding promises). In context, the statement in the Community Guidelines does not serve merely an “explanatory function” like the statement in Block. Id. Notwithstanding its present tense, the statement reads like a guarantee that users can expect identity-neutral treatment from YouTube when they use its service. Moreover, the statement is definite enough for the Court to ascertain YouTube's obligation under the contract (it must avoid identity-based differential treatment in its content moderation) and to determine whether it has performed or breached that obligation. See Weddington Productions, Inc. v. Flick, 60 Cal.App.4th 793, 811 (1998). The plaintiffs have therefore adequately alleged an enforceable promise. See Divino Group LLC v. Google LLC, No. 19-cv-4749-VKD, 2023 WL 4372701, at *5 (N.D. Cal. July 5, 2023) (considering the same Community Guidelines language in a similar case and reaching a similar conclusion).
B. The Alleged Breach
The plaintiffs fail, however, to adequately allege a breach of that promise. YouTube's promise is to not treat users who post similar content differently based on individual characteristics like race. As discussed more fully at the hearing on this motion, the general idea that YouTube's algorithm could apply the Community Guidelines in a racially discriminatory manner is plausible. But at the motion to dismiss stage, the plaintiffs must do more than gesture at plausible ideas in the abstract. They must allege sufficient factual content to give rise to a reasonable inference that their content has been treated in a racially discriminatory manner by YouTube's algorithm. See Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009).
The plaintiffs rely primarily on a chart that purports to compare 32 of their restricted videos to 58 unrestricted videos posted by white users. Dkt. No. 144 at 26-31. To begin with, the plaintiffs have dug themselves into a bit of a hole by relying on such a small sample from the vast universe of videos on YouTube. The smaller the sample, the harder it is to infer anything other than random chance. But assuming a sample of this size could support a claim for race discrimination under the right circumstances, the chart provided by the plaintiffs is useless.
As a preliminary matter, 26 of the 58 comparator videos were posted by what the complaint describes as “Large Corporations.” The complaint alleges that “Large Corporation” is a proxy for whiteness. See Dkt. No. 144 at 26-31; Dkt. No. 144 at 4 (defining, without support or elaboration, “users who Defendants identify or classify as white” as “including large media, entertainment, or other internet information providers who are owned or controlled by white people, and for whom the majority of their viewership is historically identified as white”). The plaintiffs have offered no principled basis for their proposition that corporations can be treated as white for present purposes, nor have they plausibly alleged that YouTube actually identifies or classifies corporations as white.
In terms of content, many of the comparisons between the plaintiffs' restricted videos and other users' unrestricted videos are downright baffling. For example, in one restricted video, a plaintiff attributes his recent technical difficulties in posting videos on YouTube to conscious sabotage by the company, driven by animus against him and his ideas. The chart in the complaint compares this restricted video with a tutorial on how to contact YouTube Support. In another example, the chart compares a video where a plaintiff discusses the controversy surrounding Halle Bailey's casting as the Little Mermaid with a video of a man playing-and playfully commenting on-a goofy, holiday-themed video game.
Compare Dr. SYN-Q, Is Someone at YouTube Subverting the Constitution?, YOUTUBE (Mar. 8, 2020), https://www.youtube.com/watch?v=47SZWsNMmwQ, with David Walsh Online, YouTube Support - How to Contact YouTube, YouTube (Apr. 15, 2014), https://www.youtube.com/watch?v=h8-6QltZQlk.
Compare Nicole's View, Halle Bailey & the New Little Mermaid Casting “Controversy”, YouTube (July 5, 2019), https://www.youtube.com/watch?v=iTsgfpOC7JM, with TheGameSalmon, I'M TRAPPED IN A HOLIDAY NIGHTMARE!! | Helliday Limbo (MicroHorrorArcade), YOuTuBE (Jan. 10, 2021), https://www.youtube.com/watch?v=YcScpg9kDMk.
Other comparisons, while perhaps not as ridiculous as the previous examples, nonetheless hurt the plaintiffs. For instance, the chart compares plaintiff Osiris Ley's “Donald Trump Makeup Tutorial” with tutorials posted by two white users likewise teaching viewers how to create Trump's distinctive look. But there is at least one glaring difference between Ley's video and the comparator videos, which dramatically undermines the inference that the differential treatment was based on the plaintiff's race. About a minute and a half into her tutorial, Ley begins making references to the Ku Klux Klan and describing lighter makeup colors as white supremacy colors. Ley certainly appears to be joking around, likely in an effort to mock white supremacists, but this would readily explain the differential treatment by the algorithm. See Dkt. No. 144-8 (indicating the relevant Community Guidelines prohibition on hate speech, violent criminal organizations, harassment, and other harmful or dangerous content).
Compare Osyley, Donald Trump Makeup Tutorial, YOuTuBE (Mar. 15, 2016), https://www.youtube.com/watch?v=b96JEAyqdPc, with Kat Sketch, Donald Trump Makeup Tutorial!, YOuTuBE (Apr. 5, 2016), https://www.youtube.com/watch?v=ucn7KWJcilA, and Trump the Internet, Donald Trump Makeup Tutorial, YouTube (June 30, 2020), https://www.youtube.com/watch?v=X01mNubQJx8.
Only a scarce few of the plaintiffs' comparisons are even arguably viable. For example, there is no obvious, race-neutral difference between Andrew Hepkins's boxing videos and the comparator boxing videos. Both sets of videos depict various boxing matches with seemingly neutral voiceover commentary. The same goes for the comparisons based on Ley's Halloween makeup tutorial. It is no mystery why Ley's video is restricted-it depicts graphic and realistic makeup wounds. But it is not obvious why the equally graphic comparator videos are not also restricted. YouTube suggests the difference lies in the fact that one of the comparator videos contains a disclaimer that the images are fake, and the other features a model whose playful expressions reassure viewers that the gruesome eyeball dangling from her eye socket is fake. But the content is sufficiently graphic to justify restricting impressionable children from viewing it. These videos are the closest the plaintiffs get to alleging differential treatment based on their race.
Compare Dru Story News, Anthony Joshua's Redemption | The Responsibility of Success, YouTube (Dec. 9, 2019), https://www.youtube.com/watch?v=68lsRzaRb4Y, and Dru Story News, Canelo Versus GGG | Who Wins and Why, YouTube (Sept. 15, 2018), https://www.youtube.com/watch?v=bb84LElBrQU, with Whistle, 10-Year-Old Female Boxing Prodigy, YOUTUBE (Oct. 22, 2019), https://www.youtube.com/watch?v=HtMdiNrsg00, and The World of Boxing!, Instant Karma in Boxing | Oleksandr Usyk, YouTube (Sept. 19, 2021), https://www.youtube.com/watch?v=7l-TaLMZzZA.
Compare Osyley, Como hacer cortadas o heridas falsas con maquillaje paso a paso [How to Make Fake Cuts or Wounds with Makeup Step by Step], YOUTUBE (Oct. 9, 2014), https://www.youtube.com/watch?v=6O-BXCg0p7I, with ellimacs sfx makeup, Halloween Torn Out Fake Eye Makeup Tutorial, YOUTUBE (Aug. 2, 2015), https://www.youtube.com/watch?v=pRhQix-8VeA, and ellimacs sfx makeup, Easy Realistic Wound Halloween Makeup Tutorial, YOUTUBE (Mar. 10, 2019), https://www.youtube.com/watch?v=44nhQ48au5Q.
But the complaint provides no context as to how the rest of these users' videos are treated, and it would be a stretch to draw an inference of racial discrimination without such context. It may be that other similarly graphic makeup videos by Ley have not been restricted, while other such videos by the white comparator have been restricted. If so, this would suggest only that the algorithm does not always get it right. But YouTube's promise is not that its algorithm is infallible. The promise is that it abstains from identity-based differential treatment.
Another problem is that these restrictions occurred before YouTube added its promise to the Community Guidelines. The Fifth Amended Complaint alleges that Ley's Halloween makeup tutorial was restricted on December 28, 2022. Dkt. No. 144 at 31. But, as YouTube points out in its supplemental brief, the plaintiffs alleged the restriction of this video in an earlier complaint filed on September 21, 2020, long before the promise was added to the Guidelines. Dkt. No. 27 at 122. The same is true for other alleged restrictions. See Dkt. No. 21 at 82 (First Amended Complaint, filed on August 17, 2020, alleging that Denotra Nicole Lewis's “Halle Bailey & The New Little Mermaid” video was already restricted); see also Dkt. No. 21 at 94-97 (suggesting that Hepkins's boxing videos were also restricted as of August 17, 2020). These restrictions cannot support a breach of contract claim because YouTube cannot be liable for breaching a promise it had not yet made.
YouTube also argues that, even if these comparisons plausibly suggested differential treatment, they would not support the plaintiffs' claim for breach of contract, because the Community Guidelines (and therefore the promise the plaintiffs rely on) apply only to removals. The Court finds no support in the record for that proposition, and it finds support to the contrary. See Dkt. No. 144-8 at 6 (identifying the act of “age-restrict[ing] content” as an “action [that] YouTube take[s] for content that violates [the] Community Guidelines”); Dkt. No. 144-8 at 49 (similar); Dkt. No. 144-13 at 3 (“YouTube monetization policies . . . include YouTube's Community Guidelines.” (emphasis added)). But because the plaintiffs fail to state a claim in any event, there is no need to reach this issue.
As for the rest of the complaint, only a few allegations-namely, allegations regarding purported admissions made by YouTube in 2017-merit some discussion. The plaintiffs allege that, in response to criticism by members of the queer community, Google's Vice President of Product Management, Johanna Wright, “acknowledged that its video content review tools were ‘filter[ing] out content belonging to individuals or groups based on gender, race, religion, or sexual orientation.” Dkt. No. 144 at 22. They also allege that, a few months after Wright's alleged admission, YouTube “convened a meeting of approximately 15 aggrieved YouTube creators ... to discuss and explain YouTube's inherent bias and classification of video content filtering based on identity[.]” Dkt. No. 144 at 22. At this meeting, YouTube's representatives allegedly “stated” that the company's “algorithms and computerized filtering tools discriminate and ‘target' users like Plaintiffs based on their Identities, including Race, when making decisions regarding [monetization and restriction],” that they do not “consider only the content in the video,” and that this “results in the ‘targeting' of historically marginalized consumers, like Plaintiffs, at rates that are substantially, materially, and disparately higher than those of white users.” Dkt. No. 144 at 22-23.
These allegations are not meaningless. But they do not come close to making up for the glaring deficiencies in the plaintiffs' chart. First, the allegations are vague as to what exactly was said. For example, the complaint purports to quote Wright, but it is not clear where Wright's words end and the plaintiffs' recitation of legal buzzwords begins. See Dkt. No. 144 at 22 (opening quotation with the word “filter[ing]” but never closing the quotation). Similarly, the plaintiffs attribute a great many (buzzword-laden) statements to YouTube's representatives but barely quote them.
Second, and more importantly, these alleged admissions were made in 2017, four years before YouTube added its promise to the Community Guidelines. In machine-learning years, four years is an eternity. There is no basis for assuming that the algorithm in question today is materially similar to the algorithm in question in 2017. That's not to say it has necessarily improved-for all we know, perhaps it has worsened. The point is that these allegations are so dated that their relevance is, at best, attenuated. Finally, these allegations do not directly concern any of the plaintiffs or their videos. They are background allegations that could help bolster an inference of race-based differential treatment if it were otherwise raised by the complaint. But, in the absence of specific factual content giving rise to the inference that the plaintiffs themselves have been discriminated against, there is no inference for these background allegations to reinforce.
III
The motion to dismiss is GRANTED. Dismissal is with prejudice. Both sides are ordered not to remove or otherwise make unavailable the videos cited in the complaint until the appellate process has run its course.
IT IS SO ORDERED.