This is How You Impeach an Expert Witness.
Google expert Dr. Mark Israel is thorough on direct examination, but his credibility is eviscerated on cross. And the "rocket docket" stays true to its reputation as the trial comes to a close.
Well, the anticipated six week trial ended in just three. We’ll hear closing arguments on November 25 and could have a final ruling on liability and remedies by the end of the year, or early next year. The expedited nature of this trial is a story unto itself.
But before we get to that, let’s summarize Google’s case.
Google’s case is that there is no defined or definable market for “open web display ads” and has done everything in its power to muddy the waters around this alleged market and its role in the ad tech ecosystem. Google has brought in a king’s ransom of economic experts to pound data favorable to the company’s arguments, which almost entirely exclude rival ad exchanges and publishers from the conversation. In Google’s telling—the company is the best place for advertisers to do business. By building the best ad exchange and harnessing immense scale, publishers will always choose Google.
Without Google, spam and fraud would savage the industry, and the government, Google contends, is on a fishing expedition to assert antitrust claims over a form of advertising that is both defunct and fungible. Advertisers today use many similar tools from other companies—“multihoming” or disbursing their ad spend widely. As ads bought in “walled gardens” (e.g., apps like Facebook or TikTok), where everyone now spends their digital time, often look identical to open web display ads on desktop, where hardly anyone goes anymore, the government is simply missing the mark. Google also would tell you that this is not a “tying” case, because there are technically ways to avoid Google’s ties—difficult as those methods may be—and the government is just trying to hide from unfavorable “refusal to deal” analysis.
Google’s lawyers deserve credit for doing a lot in a short period of time. While last Friday and on Monday the light touch approach to Google’s case seemed perplexing, it now makes some sense. Google’s strategy was to hammer market definition, and muddy the water so severely that you couldn’t see the river of government evidence right in front of you from inside the boat.
In some ways, Google succeeded. Talking to many in the gallery, the mechanics of bidding, the waterfall architecture, some of the economics, and the more technical elements seem like a wash, with each side presenting conflicting expert testimony with a limited ability for lay fact finders (or at least people who cannot see or understand the code all of this runs on) to know who is right.
Given that open web display ads are receding to some degree while people increasingly spend their time on apps, where “in app” ads look and function nearly identically but are sold through exchanges owned by the apps themselves— e.g., an advertiser using Meta’s ad exchange to place ads on Instagram—does Google have a point here?
It’s a debate at the crux of this case. And we’ve heard from advertisers who move their ad spend across these exchanges, even if they are unwilling to completely abandon open web display ads. I believe the government is still favored here because there are some things Google chose not to fight. For instance, unlike advertisers who move their spend between exchanges, publishers—like the newspapers who offer display ad inventory on their websites—rely almost exclusively on a single dominant firm that has a durable share of the market for publisher ad servers of over 90%, which it leverages to control exclusive access to advertisers. Its employees openly discuss anticompetitive conduct (if they didn’t delete it), and it takes affirmative steps to stymie potential rivals and force publishers to deal with lower quality ads (and advertisers to deal with high prices).
Now it’s up to Judge Brinkema to decide. Let’s turn to what happened in the last two days of the trial.
Day 14 began with Google calling Jessica Mok, a finance director at Google who used to work on the advertising side of the house.
Ms. Mok’s time on the stand lasted only about half an hour in its entirety. The purpose of this testimony was for Google to establish a few things on the record: that for a few years around 2017, Google lost money in the digital advertising business, made more of its profit from in-app ads than open web display ads, and operated at slim margins—essentially putting cold water on notions that Google had always enjoyed unchallenged monopoly profits. A number of pieces of this testimony involved sealed exhibits we couldn’t see, mostly related to Google’s internal finances in more recent years.
On cross the government went to work on the fact that Google broke down categories of ads with specificity, and indeed tracked revenue separately for display ads. Not bad.
We knew Thursday would be dominated by the testimony of Google expert witness Dr. Mark Israel of consulting firm Compass Lexicon. But the fireworks were less expected.
As an aside, I noticed star government witness Robin Lee in the elevator on his way to the courtroom. It is certainly a curious business where experts come to hear other experts trash each other’s reports.
Bill Isaacson lead the direct examination for Google. After going over Israel’s CV, noting his PhD from Stanford, his past experience as a professor, and current work, Isaacson got down to business.
Dr. Israel said that “ad tech providers are in the business of facilitating matches”—the notorious two-sided-platform Amex framing that Dr. Israel would hammer throughout his direct examination.
Dr. Israel criticized the government for excluding apps, in-stream, video, and retail media from their analysis.
He said the government failed to observe the “vibrant competition” that is going on in ad tech, leading them to exclude Meta, Microsoft, TikTok, The Trade Desk, and other alleged competitors.
He called Google’s ad business “a huge success story,” hearkening back to the sunny portrait painted in Karen Dunn’s opening statement (“This is a case about transformative innovation…”)
We were back in Google’s world.
With respect to pounding the table on the platform/2-sided market/Amex framing, Dr. Israel said that “ad tech sits in the middle”, that the government defined the market too narrowly, and that the relevant market in this case should be a “2-sided market focused on match[ing advertisers and publishers].”
Isaacson then displayed a chart showing that between 2010-2022, the time people spent with “non-social, non-video content on desktop, laptop, or mobile browser” declined from 73% to 23%.
In other words, this is evidence that people spend much less time in the places where open web display ads are, and more time in walled gardens, and therefore the government’s market definition is missing the mark, if not obsolete.
At this point, I wrote in my notes the question of whether the government brought this case too late. Of course, politics and administrations are what they are—this is arguably the soonest the antitrust regime in place in the United States was of the mindset to bring a case like this (although the investigation began during the Trump administration and was brought by State Attorneys General years before the Justice Department’s case), so it is what it is. I’ve noted frequently in these articles the fact that much of the evidence in this case is from 2008 to around 2018. But there is something unresolved about the fact the government is suing over a technology that seems to have more of its days behind it than in front of it. Put differently, assuming Google does have an illegal monopoly—is it a monopoly over a specific type of advertising that is declining, while other mediums are on the rise?
It might not matter to Judge Brinkema. The harms are real, and the facts speak for themselves. But if Google loses, this would no doubt be part of a fight over remedies—with Google likely to take the position that the government shouldn’t break up the firm at large over a monopoly that is narrow and waning. (Even if Google is playing its own part in the demise of the open web.)
Dr. Israel highlighted the notion of different ad channels being fungible—that advertisers can and do shift their ad spend between Google and Meta, for instance, and the ads even look the same. However, as the government will say each time this comes up—open web display ads can’t be bought from Meta to go on a publisher’s site. For that, you can only put a display ad onto a Meta property, not vice versa. When it comes to ad exchanges for the open web, Google stands alone—and on top.
Dr. Israel kept building on this foundation, emphasizing that roughly half of ad spending is now on social media platforms.
Bill Isaacson then pulled up Figure 114 from Dr. Israel’s report, showing astronomical ad spending on TikTok—another alleged(-by Google) rival.
The defense was hammering the idea that even if you don’t find display ads fungible, ad spending is fungible, with advertisers constantly shifting their money (and ads) into different channels.
Dr. Israel then referred to TikTok as “the next step” in the ads business, apparently unaware of a January deadline requiring TikTok to divest from Chinese ownership or shut down over national security, privacy, and user well-being concerns amid a deep distrust of the app and its developers. TikTok is currently appealing.
Israel then said that government expert witness Robin Lee failed to conduct a valid hypothetical monopolist test (HMT). Israel pointed to the fact 86.3% of DFP publishers aren’t big enough to even be charged fees (Google only charged fees to very large publishers) and pointed to competitors exiting the ad server market (meaning that prices should be expected to go up, but they did not.) This is an inversion of the 20% take rate attack on Google—instead of having its unchanging fee used against it as evidence Google never had to lower prices, turning it around to say they in fact never raised them either.
Dr. Israel then discussed the technical and economic reasons why publishers want users to use their apps (walled gardens) and not view their content on the open web, as the monetization capabilities are much stronger in apps.
This is true. We’ve all had the experience of trying to browse the web on our phones only to have many sites bug you and frustrate you into using their app. Israel alluded to this common experience.
This is also an attack on the value and relevance of open web display ads, and we haven’t really heard the government find a way to push back on this attack on its framing of the market.
Dr. Israel also testified that Google had to worry about the cost of it “not behaving well,” otherwise publishers would leave open web display ads and make their own apps and ad servers. This is problematic for a few reasons, mainly that in the government’s case we heard from multiple publishers how impractical (if not impossible) that is; it requires a massive amount of capital and expertise related to things that the publishers are not in the business of, and would likely not even get off the ground given the existing market conditions.
On balance, though, Dr. Israel came across as a decently credible witness on direct examination (this would change very quickly on cross examination). Part of this was the way he stole the government’s thunder by hedging certain answers or acknowledging the limitations of his testimony, and jumping in front of potential objections, with qualifiers like, “I’m not saying they’re all moving to apps…”
At long last, more than four hours (!) after it began, Dr. Israel’s direct examination concluded.
Cross examination was conducted by Aaron Teitelbaum for the United States.
What happened next was some of the most striking and remarkable impeachment of a witness’ credibility money can buy.
Charging out of the gate, Teitelbaum asked Dr. Israel about the sheer quantity of trials he has been paid to testify in, including just two weeks ago in the Kroger-Albertson’s merger showdown—where Israel testified on behalf of the defendants, of course.
Teitelbaum addressed that Google is a repeat client for Dr. Israel, who testified in the Google search antitrust trial last fall. He then asked Israel if it is true that 80% of his income actually comes from being a serial “expert,” with only around 20% coming from his actual job. Israel confirmed.
Teitelbaum also asked Israel to confirm that he’s never held a tenured faculty position (which seems like the ultimate diss for these sorts of people), and rattled off other cases such as FTC v. IQVIA-Propel, the Google Search trial, the JetBlue/American Airlines (Northeast Alliance) trial in Massachusetts, the JetBlue-Spirit merger, among others in which Dr. Israel served as an expert witness for the defense.
Then Mr. Teitelbaum dropped the mic—reading quotes from the decisions in cases in which Dr. Israel was paid to be an expert, where judges had taken him to task, calling him “not credible”, and finding that he “misunderstood antitrust law.” Mr. Teitelbaum took note that the company lost in each of these cases.
As it happens, just a few weeks ago Assistant Attorney General for Antitrust Jonathan Kanter delivered remarks at the Fordham Competition Law Institute’s 51st Annual Conference on International Antitrust Law and Policy, speaking to the issue of pervasive conflicts in expert economic testimony. You can read it here, but the gist is that companies like Google having rolodexes of supposed “experts” to be called upon to “opine” as to whatever conclusion the company needs, is contributing to a “crisis of expertise.”
The government then put forth a series of hypos aimed at having Dr. Israel agree that just because the percent of a budget spent on one thing goes up over a period of time while going down on another, does not mean the two items are necessarily substitutes. This was aimed at showing that this sort of data alone is insufficient to prove that various types of ads or ad channels (recall the desktop/browser/non-video exhibit described above) are substitutes for each other.
Dr. Israel also acknowledged that evidence like a May 2008 email—wherein Larry Page (at the time in between stints as CEO) refused to agree to an open ad network that wouldn’t restrict competitors—was not pro-competitive evidence (with apologies for the triple negative.)
The government continued its impeachment of Dr. Israel’s otheriwse thorough research, arguing that since he reviewed so many Google documents, it would seem they were heavily at odds with his findings. Of course, Dr. Israel almost always finds for the company paying him as an expert.
The government then displayed a terrific graphic, which I’ve included at the top of this post, showing a graph of Google’s dominance (which Israel concluded was not a monopoly) put alongside the same data for Standard Oil, AT&T, and Microsoft—three firms famously found to be monopolies, to highlight the folly of his claims.
Another exhibit showed a Google email stating that “if service is the main competitive lever on the sell side, we lose.” A December 2018 Chris LaSala email, which was also used in the government’s case-in-chief, reflected LaSala’s view that Google “can only retain a 20% rev share because of AdX’s unique GDN demand.”
Then there was this: Dr. Israel agreed that at the time of Google’s Admeld acquisition, Admeld was a competitor to Google. That goes directly to the government’s allegation that Google’s acquisitions were themselves anticompetitive monopoly maintenance in violation of Section 2 of the Sherman Act. The government then highlighted a Facebook document stating that header bidding threatened Google’s monopoly.
When asked about Google’s habit of deleting messages concerning the discussion of competition issues, Dr. Israel said, stunningly, that this did not affect his analysis, because those records (missing evidence that tends toward incrimination, and for which the judge has said she’ll draw adverse inferences) are not relevant. If Dr. Israel had much in the way of credibility left, it really unraveled with this remark.
Israel conflated market definition time and again, suggesting the New York Times could replace Google with Facebook for open web display ads, even though we’ve heard time and time again how this is not possible.
One more time for those in the back: Facebook does not operate a publisher ad server. Therefore the NY Times could monetize its content on other platforms, but could not put Facebook display ads on the NY Times.
On redirect, in one of the final questions of Google’s case-in-chief, and one of the last questions of the trial, Bill Isaacson asked Dr. Israel, “how many small independent publishers are even left?” when discussing how small publishers spend on ads.
What more fitting, more ironic way could there possibly be to signify what this trial is all about?
Picking up on Friday, September 27th—the trial, which was originally thought to take 4-6 weeks, ended after three—with the Rocket Docket of the Eastern District of Virginia being true to its name.
Going into Friday, we expected a government rebuttal that could possibly go into Monday. When the day began, the courtroom was packed, with each side’s full legal team in attendance. Karen Dunn and Julia Wood each thanked and acknowledged their teams who have worked tirelessly on this, leading Judge Brinkema to interject with what we all realized: “It sounds like the trial will be ending today?”
When Thursday ended, there was a dispute between the parties over a list of government rebuttal witnesses. While we have little to no insight into what happens outside of court in negotiations, apparently the issues were resolved, and the plans cut down.
Judge Brinkema thanked each legal team, telling a crowded courtroom: “This is how all civil cases should be.”
Google finished two deposition read-ins: Ryan Pauley, Executive Director of Revenue Operations at Vox Media, and Bryan Bumpers, Marketing Analytics Manager at Zulily. Mr. Polly discussed how Vox used price floors and its Concert ad network. Bumpers touched on shifting ad spending away from Google, and the many different channels through which they do their advertising. These read-ins were brief, and primarily focused on painting a market where people can forgo Google and find success, that competitors can and do emerge, as well as the more practical purpose of admitting evidence using the depos.
Google then rested its case.
Given how the morning started, it was possible the government had decided to do no rebuttal. Instead, they called back one witness, Matthew Wheatland, Chief Digital Officer at Daily Mail. As a publisher witness, the government wanted to pose some of the scenarios Google has offered, like the practicality of using Google alternatives (specifically alternatives to open web display ads), and more details about how Google’s pricing and policies play out in real life. There were a flurry of objections on both direct and cross between Google lawyer Jeannie Rhee and lead government lawyer Julia Wood—which continued on cross.
The government also had Mr. Wheatland pour cold water on Google’s notion that publishers can build their own ad server, or switch to in-app monetization.
On cross-examination, Ms. Rhee was ferocious. She went in on the fact Daily Mail has been suspended multiple times by Google for content violating its policies, including sexual content. Recall the role brand/ad safety has played during the trial. Google seems to have saved this impeachment in reserve in the event Mr. Wheatland was a rebuttal witness, as I do not recall hearing it on cross during the government’s case-in-chief. Mr. Wheatland was an honest witness, trying to give fulsome explanations, but Ms. Rhee pressed him for yes/no answers that weren’t helpful to the government, leading some to wonder whether the rebuttal was even worth it if it meant ending the case on a sour note. Ms. Wood objected feverishly, but Judge Brinkema allowed a lot of the testimony on the grounds the government opened the door to it on direct.
Finally, Ms. Rhee once again brought out the framing that publishers and advertisers are in control of whether and how to use Google’s ads platform—something the government attacked by arguing Google and only Google decided what ads to put on an impression, and at what economics.
With no further questions, Ms. Dunn and Ms. Wood came to the podium and rested, and just like that, the trial was over.
Judge Brinkema again commended each side on superb lawyering, and scheduled closing statements for November 25th.
Thanks to all of you for following along for the last few weeks of this trial. It’s been a pleasure to attend and cover it, and I thank you all for your readership and kind comments. I or someone else at the American Economic Liberties Project team will have more to say in the weeks and months ahead, so stay tuned.
Thanks Tom for covering and reporting on this case for us. It was a lot to read and digest, so imagine your time commitment was extensive.
Brilliant reporting, amazing to be able to get such detail about such a significant (and significantly under-reported elsewhere) case, thank you