The Sky Is Red: Meta's Flailing Defense Takes Serious Credibility Hit

Meta's in-house consumer survey guru Curtiss Cobb defends the indefensible on Day 11 of FTC breakup trial after executives from Tumblr and TikTok testify.

A few years ago, Judge Yvonne Gonzalez Rogers, a federal judge of the U.S. District Court of the Northern District of California (who now happens to be overseeing the case brought by State Attorneys General against Meta), sat for a panel discussion to give trial tips for antitrust practitioners. The talk, which I attended, took place at the 2022 “Spring Meeting” of the American Bar Association’s antitrust section, which brings together thousands of antitrust lawyers from all over the world to Washington, DC each year. Judge Gonzalez Rogers drew a large crowd because she had, the year before, presided over the antitrust trial brought by Epic Games challenging Apple’s App Store business practices.¹

At the talk, Judge Gonzalez Rogers explained how important it was for expert witnesses to keep their credibility. Nothing hurts credibility more, she said, than trying to claim with a straight face that the “sky is red.” Rather than dispute every little issue and risk a credibility hit, experts would be better advised to make concessions and focus their attention on disputed points that really matter. While she might not be fluent in all the complexities of an economic model, when an expert says the sky is red, she notices.

An AI rendering of what we experienced in court today. Not a quote from testimony.

I couldn’t help but think of that sage advice—which applies as much to fact witnesses as it does to experts—in court today listening to Meta’s head of its internal Demography and Survey Science team, Curtiss Cobb. In our curtain raiser, we were alone among the press in previewing that Curtiss Cobb was one of a handful of key witnesses to watch for at trial. Today, our prediction proved true: sparks flew as the FTC introduced survey after survey prepared under Cobb’s purview that supports the FTC’s personal social networking services (“PSN”) market and helps show Meta’s monopoly power. But Cobb tried at every turn to add new caveats to his prior testimony—even denying that one survey result was based on user “experience” even though that word was in the sentence summarizing the study’s findings. Meta’s strategy at this late hour is becoming clear: to have its executives say that the sky is red and hope for the best.

If you can believe it, while I was writing today’s summary and thinking about the advice Judge Gonzalez Rogers gave, she issued her decision on Epic’s motion to enforce the injunction it won against Apple in that aforementioned suit. It’s a doozy: she found that Apple willfully violated the court’s order to let Apple device users make purchases outside the App Store using links provided by developers without paying Apple’s 30% commission, and then referred Apple and one of its executives to the United States Attorney’s office for potential criminal prosecution. Her opinion, covered by Matt Stoller earlier, doesn’t mince words about the pitfalls of relying on testimony that contradicts the contemporaneous record (highlights added):

“In stark contrast to Apple’s initial in-court testimony, contemporaneous business documents reveal that Apple knew exactly what it was doing and at every turn chose the most anticompetitive option. To hide the truth, Vice-President of Finance, Alex Roman, outright lied under oath. Internally, Phillip Schiller had advocated that Apple comply with the Injunction, but Tim Cook ignored Schiller and instead allowed Chief Financial Officer Luca Maestri and his finance team to convince him otherwise. Cook chose poorly. The real evidence, detailed herein, more than meets the clear and convincing standard to find a violation. The Court refers the matter to the United States Attorney for the Northern District of California to investigate whether criminal contempt proceedings are appropriate.

This is an injunction, not a negotiation.”

With all that in mind, how does Meta think Chief Judge Boasberg will react to similar that’s not what I meant testimony in his court? Read on to learn how he ended Cobb’s day.

Surveys, Smoke, and Mirrors

On Day 11, the FTC saved the best witness for last. Cobb is in charge of a team at Meta that builds consumer surveys and then reports on those findings to senior executives. Here he is in 2019 discussing how “powerful” and “impressive” network effects are for tech platforms like Facebook and Uber:

The FTC’s Albert Teng questioned Cobb, and it was one of the Commission’s better examinations of this trial. Teng methodically laid what we attorneys call foundation: that’s the personal knowledge that the witness has about a subject that allows you to ask the witness questions on that topic so that their answers can be admitted as evidence. Heeding this blog’s free advice to avoid asking the ultimate (or “burrito”) question, with the first few exhibits, Teng simply asked foundational questions so that the court could consider the contents of those studies for their truth. And this is a smart move with a slippery witness. The documents are admissible as “party-opponent” admissions: generally, if certain facts are shown, the court can consider this evidence from the defendant, even if it would not be able to consider the same kinds of statements made outside of court from a third party. Once the document is in evidence, it “speaks for itself,” as Chief Judge Boasberg has said throughout trial. You don’t really need to ask the witness about it, especially if you think the witness will try to add all kinds of caveats in an attempt to change what the document says.

And when Teng started asking more substantive questions, that’s how Cobb tacked. Do Meta’s surveys use sound scientific principles? “We aspire” to do that, Cobb said. Was YouGov the highest rated external survey source? After an exchange, up came Cobb’s deposition testimony where he said that. “My testimony was imprecise at the time,” Cobb said, prompting Chief Judge Boasberg to smile. Asked about a Meta survey that found that privacy is “the single biggest detractor” in consumer sentiment about Meta, Cobb answered that he didn’t write that specific deck, so “it’s probably not the language I would have used at the time.” Teng pointed out that the metadata showed that Cobb had edited the deck, that he was the head of the research group, and that it was being sent to senior Meta executives—what about all that? “I don’t recall what review I did of the document itself,” Cobb testified. Already, the credibility doubts were creeping in. What kind of executive at a public company would randomly let a deck be sent to executives or the board of directors without fixing a mistake?

But Teng was just getting warmed up. After looking at a couple of decks (and the highlights of what we saw from them are further below), Teng put up a more important document, this one from 2021 and emailed to senior leaders like Sheryl Sandberg, then Meta’s COO. It was based on interviews, focus groups, and a survey of 3,500 consumers. The key finding (emphasis added):

“Their pre-existing beliefs and experiences about targeted ads make them think these ads are only slightly more beneficial than other types of advertising in terms of relevance and cost.”

Now the point here was to show that Meta users don’t care all that much about personalized ads because they don’t like ads. The FTC has alleged that Meta has monopoly power, and has caused anticompetitive harm, by increasing the ad load (the number of ads relative to other content) that users see on in its apps. Meta has responded by arguing that personalized ads are good, actually. As Sandberg put it in her testimony, Facebook worked on its ads until they were “as high quality or even better than the organic content,” and that she knew that users liked ads because they clicked on them. The FTC, though, showed that Sandberg lacked foundation for that spin because it wasn’t her job to measure consumer sentiment about ads. That’s Cobb’s job.

So what did Cobb have to say about this finding in this document that his team prepared for Sandberg? “No, I wouldn’t agree with that,” he testified. Teng asked, “You disagree with the statement in this document?” Cobb answered, “What it’s saying in this sentence . . . . [it’s] not that their experiences are causing necessarily this but that these are the beliefs that they have so this is why it’s important to understand their stated attitude.” Teng asked the obvious: “But you see the word experience there?” Cobb replied simply, “Yes.” How could that statement about the survey’s findings *not* be about users’ experiences when it says it is based on their experiences?

Teng was building to his high point: a survey about the fallout from the Cambridge Analytica scandal, a topic covered earlier in trial with Sandberg. Cobb’s team concluded that Cambridge Analytica was the “most extreme” shock to consumer sentiment to date and “the most likely significant event that would have had a negative impact on both revenue and engagement.” But Meta “fail[ed] to detect significant and consistent effects of sentiment . . . on these metrics.” In other words, consumer opinion about Facebook plummeted, but users didn’t use Facebook less, and Facebook didn’t sell fewer ads. The ability not to lose users or advertisers to competition in response to a perceived decrease in privacy of this type suggests that Meta is a monopolist. So Cobb, naturally, didn’t agree with Teng’s questions about that finding. The sentence doesn’t say significant “or” consistent effects, it says significant “and” consistent effects, Cobb said. Then we saw Cobb’s deposition testimony again, where he seemed to agree with the statement. “Let me clarify something,” Cobb said today in response. He went on:

“This study does not find a significant effect as a result of Cambridge Analytica. It also doesn’t isolate just Cambridge Analytica. So when we talk about consistency, that becomes an important part overall of what we’re describing in the study.”

Teng moved to admit the deposition testimony into evidence as a prior inconsistent statement. Chief Judge Boasberg, who at one point in this line of questioning leaned his head back and rolled his eyes, admitted the prior testimony, saying that it would be “considered for truth as well as credibility.” On that note, he dismissed court for the day.

And it was an ominous note for Meta. This is the theatre of the absurd that Meta executives seem to think will persuade some court, somewhere, to rule in its favor. It wants the court, the Trump administration, and the public at large to disbelieve their own eyes and ears. Its executives are so shameless about this that they will do it with words that everyone can see are in the sentence, or as Mark Zuckerberg tried, to dispute words we just heard him say in a video clip. And it’s the same kind of battle between contemporaneous statements and newfound testimony that Judge Gonzalez Rogers wasn’t buying from Apple, or Judge Mehta from Google in the antitrust case it lost over its monopolization of general online search, for that matter. Why does Meta think anyone will believe them when they say the sky is red?

Some other interesting tidbits in the survey exhibits:

• From 2020: “In the U.S. this year, Facebook has slid to the 21st place and falls behind all other tech companies” in consumer sentiment.

• From a “Privacy Zuck Review” in 2016: “We have consistently seen from data and research that the complexity of privacy controls on Facebook as well as data use issues are key drivers of distrust and decreased sentiment.”

• From the aforementioned 2021 document sent to Sandberg: “Both consumers and policy elites believe that the primary beneficiaries of personal advertising are companies like Facebook . . . . Both groups express that their concerns about privacy and data collection outweigh any potential benefits.”

• From an August 2018 survey: In response to the question, “In the past month, which activity did you do the most with the feed feature on the app?”, 20-25% of Facebook, Instagram, and Snapchat users answered that they used the feed to “see daily casual moments,” vs. percentages in the single digits for Twitter and YouTube users.

• From a companion survey based on U.S. data only: “Users who perceived most or some of their family/friends using the apps . . . tend to have better sentiment of the platforms compared to users who think only few or none of their family/friends . . . use the apps.” This is gold for the FTC’s theory that users of Meta’s apps go there for friends and family sharing, in contrast to others like TikTok.

• From a “Facebook User Goals and Intent Survey”: 60% of Facebook users who did not use Instagram said the main reason they use Facebook is to keep up to date with friends and family, vs. 54% of users on both Facebook and Instagram who said the same. Cobb’s team concluded that “Facebook’s core value proposition is to keep up with friends and family.”

With evidence like that from Meta casting significant doubt on Meta’s litigation strategy to broaden the PSN market away from friends and family sharing, it’s no wonder Cobb wanted to tap dance his way around his team’s own findings.

Tumblr, Tucker, and Taylor

Cobb ended the day, but Day 11 started with Eli Tucker, previously the CTO of Tumblr, which was acquired by a company called Automattic about five years ago. Peter Taylor of the FTC started the examination by showing Tumblr’s description on the Apple App Store:

The point here and throughout the FTC’s direct was that, unlike Facebook and Instagram’s core use case of friends and family sharing, Tumblr is an “interest-based platform” that’s all about creativity: think art, fan-fic, and in a past life, more adult content. Tumblr’s home page doesn’t emphasize family or friends; Tumblr doesn’t even have a friends tab or feed, and it doesn’t have a “social graph.” When users create accounts, they choose interests, not connections to import. And, like Reddit, Discord, and Twitter, Tumblr handles tend to be pseudonymous; Tumblr doesn’t ask for a user’s name at registration. The FTC also asked questions and got answers that Tumblr can do things like reduce latency and monitor for harmful activity on its own; this cuts against Meta’s claims that all roads to safety for Instagram and WhatsApp ran through Meta.

It wasn’t all roses for the FTC, though. Tucker testified (on direct) that while Tumblr began as a blogging platform, it became “more of a social network with mobile features” over time. And Chief Judge Boasberg at one point moved the questioning along, saying he’s heard the friends and family questions a few times already. On cross, Meta showed that Tumblr has a content ranking algorithm based on a user’s past interactions, like Facebook and Instagram do. And users can like, comment, reblog, and direct message within the platform any posts they enjoy. What’s more, Tumblr introduced a short-form, Reels-like feature of its own called Tumblr TV. For good measure, Meta pointed out on cross that Tumblr originally sold to Yahoo for $1 billion (like Instagram did to Meta), before selling for less than $10 million for Automattic—making the point that success is no guarantee. And Tumblr had to hire outside help to moderate content when it banned pornography.

Asked on cross if Tumblr competes with Instagram and Facebook, Tucker said, “In some ways, yes.” That would be “for user time and attention,” Meta’s view of how competition works. The bottom line, though, was that Tumblr isn’t an alternative to Instagram and Facebook for getting updates about real-life connections.

TikTok on the Clock

The second witness of the day was Adam Presser, Head of Operations and Trust & Safety at TikTok. Presser is a well-credentialed fellow, going to Yale as an undergrad and grad student, before earning a law degree and MBA from Harvard. We’ve already seen deposition videos from two former TikTok executives, V. Pappas and Blake Chandlee, but Presser was the first TikTok witness to testify in person. (Pappas is expected to be called live later in the trial.) Before trial kicked off, TikTok fought subpoenas for Presser to come testify in person, but Chief Judge Boasberg ordered that he had to appear. TikTok is probably the most important third party in the case; Meta is hanging its hat on the idea that it competes with TikTok so that it can broaden the FTC’s proposed personal social networking services (“PSN”) market and thereby lower Meta’s market share, making it look less like a monopolist. Given the importance of TikTok to Meta’s defense, it makes sense that we’re seeing multiple TikTok witnesses.

Daniel Matheson, the FTC’s lead lawyer, examined Presser first. We’ve seen the parties try to use statements in submissions by third parties to antitrust and other regulators in Europe and Australia to help their case, and that was the approach the FTC took with Presser. What makes these filings so interesting? Presser agreed on cross that TikTok’s filings were probably drafted by attorneys, and so these white papers tend to use the language of antitrust. They are challenging documents to use with a fact witness who may not have had anything to do with their creation—and it wasn’t clear that Presser did. (We’ll spare you the details on the divide in legal authority out there about whether corporate representative testimony, known as 30(b)(6) testimony after the Federal Rule of Civil Procedure that permits it, can happen at trial. The court here already ruled that such testimony was admissible at trial, giving the parties an opening to ask Presser about these submissions.)

Presser also submitted a declaration in 2024 on behalf of TikTok in its challenge to the Protecting Americans from Foreign Adversary Controlled Applications Act (aka the TikTok Ban).² The declaration had this good quote that supports the FTC’s market definition (emphasis mine):

14. Although there are other platforms that allow users to post and share content, TikTok differs from these platforms in important respects. For example, unlike other platforms, TikTok does not host written posts (except insofar as a user posts a video or picture showing written text), and it is not as focused on users' interactions with existing friends, family, or co-workers, like some other platforms are.

Presser testified that the “other platforms” language referred to Facebook and Instagram.

Some other highlights, which were largely consistent with statements from the other TikTok witnesses

• “We believe that Facebook and Instagram are used as a complement in addition to our services. This is because our services are not intended for networking purposes.”

• Q. “Does TikTok currently, today, offer within the United States personal social networking services as TikTok used that term in its response to the European commission? A. No, I don't think so.” Same answer for YouTube.

• “In our view, Facebook does not offer services/apps/websites . . . which directly compete with TikTok or TikTok.com.”

• Only 1% of TikTok accounts use the friends tab launched in 2022.

• Users on Instagram have 4x as many, and users on Facebook have 15x as many, friends than users on TikTok have.

• < 10% of TikTok users import contacts from Facebook or Instagram.

Aaron Panner of Kellogg Hansen took Meta’s portion of the examination. Panner mentioned that he and Presser hadn’t met before, so presumably Panner didn’t depose Presser. So his cross was more laborious. Panner tried to ask questions using the exact wording from foreign regulatory submissions, but because those submissions were written by lawyers, the questions adopting the same language were long and easy to disagree with.

Panner tried for points we’ve heard a lot already about the similarity in functionality between TikTok and Reels. But Presser held up well: “The parts that aren’t similar are the experience(s) of content . . . If you were to click out of this view you’d see a very different product.” There was a lot of back and forth over language from lawyers in an Australian regulatory submission that it’s impossible to determine a “primary purpose” for why people use YouTube as opposed to TikTok or Facebook. “Primary purpose” is legalese that comes up in some areas of the law, so Panner didn’t get anywhere with this.

One new thing Panner drew out on cross was that TikTok launched a group chat feature in 2024. But only 4 or 5% of users are in group chats on TikTok, Presser said. On re-direct, Matheson asked Presser if any of the new features discussed had changed TikTok’s primary purpose, and Presser answered that they did not. Presser also agreed that TikTok competes with Instagram, Facebook, and YouTube, but Matheson had already preempted that by having Presser distinguish between the ways in which they compete (short-form video, time and attention) and don’t (friends and family sharing).

We’ll be back for a short Day 12 tomorrow.

Stray Thoughts

• Things are getting pretty repetitive and cumulative now, and Chief Judge Boasberg has voiced that concern a few times trying to move things along. We didn’t really need to hear from Presser, since the other TikTok witnesses in sum and substance said the same things. There are still more fact witnesses due up, but we need to hear from the economists.

• Correction: Due to a misspelling in the trial transcript, an earlier version of this post misspelled the first name of the witness from Tumblr. It is Dusten, not Dustin, and the witness goes by Eli Tucker.

1. Full disclosure: as a Cravath associate in 2020, I represented Epic Games in that action before leaving to start a law firm. And as disclosed previously, I worked as a Special Assistant Attorney General with the other attorneys of my firm for the State of Utah raising similar claims against Google. The contempt proceedings against Apple discussed here began years after I stopped working for Cravath and Epic Games.

2. Another disclosure: I served as counsel of record for Asian Americans Advancing Justice | AAJC and the Fred T. Korematsu Center for Law and Equality as amici curiae supporting none of the parties when this case reached the Supreme Court.