Closing Arguments, November 25: Once, Twice, Three Times a Monopolist

Today we headed back to EDVA to hear closing arguments in #USvGoogle.

DOJ Closing: Aaron Teitelbaum

The DOJ’s Aaron Teitelbaum kicked off the day, reminding the Court of how Google rigged the rules, restricted customer choice, and undermined competition, with a veiled reference to The Commodores: “Google is once, twice, three times a monopolist.”

Before diving into the substance of the Government’s closing, he explains that all evidence is not created equal, reminding the Court that while DOJ brought a variety of fact witnesses and experts, Google only brought one live witness that wasn’t on Google’s payroll or in receipt of grant money from Google.

Judge Brinkema asks why DOJ didn’t call advertisers, and Teitelbaum explains that as we heard from the Census Bureau, advertisers themselves typically do not use Google’s tools - they have agencies that use these tools on their behalf, and DOJ called several agency witnesses (Lowcock, Lambert, Friedman).

He explains that Google conflates different types of competition in its arguments. There’s competition between firms in a sector, within a single auction, and between products in a particular antitrust market. Only the latter is relevant, despite Google’s sleights of hand. He references FTC v. Staples to underscore the point that just because a firm is deemed a competitor generally, it doesn’t necessarily mean it’s part of the relevant product market.

Market definition, he explains, is a pragmatic exercise, looking at an arena within which significant substitution occurs.  Within a broad market, well-defined submarkets may exist, he explains, citing US v. Bertelsmann. In this case, he reminds the Court of the ample evidence demonstrating that TikTok and Meta can’t be used to buy/sell open web display ads, and an ad server is not a substitute for an ad exchange, pointing out that this was acknowledged by Google’s own employees, Stewart and Mohan.  He explains that the DOJ’s argument is not at odds with Ohio v. Amex - that DOJ agrees the ad exchange is two-sided, but is in a different market than the ad network and ad server, which serve different customers.  He references how despite plenty of evidence from fact witnesses and Google’s own documents showing that open web display is seen as distinct, Google’s “zombie” argument that it is interchangeable with apps, social, etc. keeps “shuffling along.”  Similarly, he reminds the Court that direct-sold is not a substitute, contrary to Google expert, Dr. Israel’s suggestion that Publishers might just “work harder.” Julia Tarver Wood in DOJ’s rebuttal further underscores the arrogance of this argument, likening it to saying “let them eat cake.”   If it’s all one single two-sided market, Teitelbaum posits, why did Mohan refer to a 3-pillar strategy? Open web publishers  must use an ad server, and TikTok/Meta simply do not offer this service to publishers.  He calls back to the stoves/microwaves comparison from trial: just because many households own both, it doesn’t make them substitutes. Indeed, it supports an argument that these are complements, not substitutes.

He calls out inconsistencies in Google’s revised FOFs where the exhibits cited don’t line up to the assertions that Google suggests they support. For example, Google cites to a statement that the market was “hyper-competitive” with Dynamic Allocation, when in context, this statement was relative to the waterfall, not to real-time competition.

Judge Brinkema asks what happens to DOJ’s analysis if it is indeed one single market - i.e. where Google’s market share would be lower.  DOJ’s FOFs state that even in a market comprised of the entire ad stack, Google is dominant. Teitelbaum explains that DOJ presented direct evidence of Google’s monopoly power - their ability to control prices without consequence. Judge Brinkema did remark that publishers can’t go anywhere else because they want access to Google Ads advertisers, and the exchange ended with DOJ maintaining that while DOJ can prevail under either market definition construct, their 3-market definition makes more analytical sense.  He runs through this direct evidence, before moving on to market share, where DFP and Google Ads have 80-90%, and AdX, while lower (he says 56%) is still greater than its next largest competitors by a factor of 9.  He next discusses evidence of intent to monopolize in support of attempted monopolization claim for AdX - Google’s own documents say “the quiet part out loud” - and he cites to MSFT where district court also found that its pattern of exclusionary conduct was only rational if MSFT knew it had monopoly power.

We move on to why Google’s conduct was anti-competitive, recapping the AdMeld parking and the tying claims (Ads to AdX, AdX to DFP). The exclusivity allowed for the 20% take, according to Google’s own employees. This exclusivity is good for Google, not its customers - we’re reminded of a buy-side Google email about how advertisers wanted the ability to bid into all auctions unrestricted, but that’s bad for AdX.

He debunks the quality, safety, security babble Google cites as pro-competitive justification, summing it up as a “cybersecurity” argument. He says the evidence didn’t back this up ultimately. Obviously, as a tech company, cybersecurity is a concern of Google’s but this doesn’t suffice, but when AdX direct is referenced in internal Google employee chats as “a concept for antitrust,” that shows what really drove Google’s decisions.  Poorly-understood, impractical, or insufficient “workarounds” don’t nullify the anticompetitive nature of Google’s conduct. He runs through Poirot, First and Last Look, and UPR, stating that when Google removed UPR in the name of “simplicity” it effectively told its customers that Google knows best.

Next, DOJ counters Google’s refusal to deal argument, explaining that Trinko was about Verizon’s refusal to build infrastructure to connect to a rival, whereas this was about Google’s actions taken against its own customers. To accept that any conduct with downstream effect on rivals is lawful refusal to deal would mean that effectively everything can be a lawful refusal to deal.

Judge Brinkema references rival ad exchanges, and DOJ says this is a conditional refusal to deal, or, as the government has asserted, tying - imposing restrictions on its customers. He says Google is playing semantics, pointing the Court to a SCOTUS footnote. Predatory conduct, even if a refusal to deal, still violates the Sherman Act, he explains. He counters Google’s product design defense, again highlighting how Google’s procompetitive justifications are pretextual, and the burden is on Google to proffer, which they’ve failed to sufficiently do.

Judge asks if there were sufficient integration benefits, how this would impact the analysis. Teitelbaum says that those integration benefits would need to be sufficiently beneficial to competition to outweigh anticompetitive harms. Judge then asks about what if it is simply the best product, and he says that this would be competition on the merits - not what we have here. She asks for examples of a better adtech product. He reminds the judge that many of them have gone out of business, but points to Kevel which failed and Equative which has remained small not because of product quality but because of Google’s tying - calling back to publisher testimony where they considered switching but couldn’t lose the revenue.

Judge asks again why emphasize publishers over advertisers and Google Ads. DOJ explains that Google Ads is a critical part of why the tie matters - access to unique demand comprised largely of Search advertisers, and that Google internal discussions highlighted that advertisers wanted to be able to bid into other exchanges.  He closes by saying that Google isn’t here in court because of its size. DOJ is here asking the Court to hold Google to account for anticompetitive conduct that led to less innovation, less choice, and higher prices in markets that are central to supporting a vibrant internet.

Google Closing: Karen Dunn

Dunn begins by suggesting that DOJ is asking the Court to overrule Ohio v. Amex (2-sided transaction platform) and Trinko (refusal to deal), all while the digital ad market grew and Google fees decreased.  She tells the Court that Google spent billions making products interoperable, and this case is about  Google not providing equal access to Google’s own customers and tech comparable to its own.

She runs through the timeline of “conducts” at issue in this case, explaining why all of them tell an alleged story of innovation, before moving on to countering the DOJ’s arguments.  She preempts this by suggesting that the DOJ selectively highlighted statements from Google employees without full context in their case and closing.

Judge Brinkema brings up the spoliation issues, letting Dunn know that she is approaching "dangerous territory" because we don't know all that Google employees were thinking and saying (since Google failed to preserve this evidence).

We get into Ohio v. Amex.  Likely the most notable comment in terms of understanding Judge  Brinkema's analysis was a question/statement from the judge during Google closing arguments where she speaks to how after many readings of Amex, while it initially had a "certain attraction," it ultimately became less attractive. Judge  explains that Amex involved 1:1 transactions, not a dynamic auction  environment, as is the case here.

Ms. Dunn explains that in both cases there are tools to facilitate  transactions between buyer and seller, and thus indirect network effects. She insists that there is a single simultaneous transaction: the ‘match’ between publishers and advertisers. Judge Brinkema has heard this before, so if it didn’t land the first time, it’s unclear how much this explanation would change her perspective on the problems with Amex’s applicability to this set of facts as Google argues.

We hear what we did during trial about how this is all competition for advertiser dollars, and so Microsoft, Meta, and even Criteo, “which no one has ever heard of”(lol) are competitors, with the former allegedly having an advantage over Google because of LinkedIn (social), gaming (Xbox), and the Netflix deal.  Thankfully for the final time, we hear about how Supply Path Optimization (”SPO”) allegedly challenges DOJ’s market definition, failing to address the fact that publishers still need an ad server, and none of the aforementioned companies actually help publishers monetize their open web display inventory.

She defends Google’s waffling between market definitions - asserting separate product markets in Northern District of California, before arguing a single two-sided market here. She proceeds to argue that DOJ’s markets are gerrymandered, and social and display were both identified as upper funnel and thus substitutes.

Quick break for commentary: again, this line of argument ignores publishers - and if anything at this point in my opinion the blatant refusal to acknowledge publisher needs to sell their open web display inventory seems to me to prove the DOJ’s points about why Google shouldn’t own both sides of the stack.

We see a diagram of how Amazon’s ecosystem offers an end-to-end solution comparable to Google - with in-house ad serving being the Amazon answer to DFP…which is next-level conflation. Shade is thrown at DOJ for calling several witnesses from the Prebid board, which Dunn characterizes as formed by competitors.  As she continues running through the reasons why DOJ’s markets all allegedly fail, she once again references how NYT, IMDB, Amazon, Snap, and Reddit built in-house ad servers, despite Amazon, Snap, and Reddit being excluded from DOJ’s market definition. Of course,  with the latter, we’re talking about large tech companies bringing ad servers in house versus publishers of various sizes.

Dunn then gets into why Google didn’t engage in anticompetitive conduct, but engaged in lawful refusals to deal under Trinko. Judge Brinkema at one point asks how publishers could access Google’s unique demand - who are Google’s competitors? Dunn brings up AdSense (which wouldn’t help with direct sold) and DV360 (which wouldn’t help small advertisers). We get another reference to the disingenuous argument about how Google’s competitors and a few whiny publishers wanted Google’s stack to be community property. She mentions AWbid’s bidding on 3rd party exchanges (again, limited to remarketing) - and how initial AWbid tests showed high IVT. Dunn says “Google is very concerned with vetting publishers and advertisers” in its ad network, and I roll my eyes.

As we near the end of Google’s 90 minutes for closing, Dunn begins to appear flustered.  She stated that a  majority of Google ads advertisers are multi-homing; however, Dr. Lee’s report shows 4M advertisers (mostly SMEs) that use Google Ads alone. She also states that "the vast majority" of publishers using DFP don't use AdX. While I assume the former meant to reference spending volume vs. # of advertisers,  the second statement seems a product of being frazzled by the 5 minute warning.

She brings up the Standard Oil, AT&T, MSFT, Google chart that DOJ used to demonstrate that just because a market is growing, it doesn’t mean that there’s not a monopolist engaging in anticompetitive conduct. She superimposes a line that shows a much smaller market share for Google. Judge Brinkema calls her on it, asking what’s included, and Dunn explains that it’s based on emarketer data that includes Apps, CTV, and Social. Judge asks if there’s a version of the chart showing only open web display, and she says no, because the data isn’t broken out that way. Skipping to the end, Dunn brings up a slide of all of the alleged principles of antitrust the Court would need to disregard to find for DOJ, saying that the list is under-inclusive much like the plaintiff’s markets. It was a funny one-liner - but fell flat against the weakness of Google’s closing, particularly during the back third.

Rebuttal: Julia Tarver Wood

Ms. Wood begins by addressing a question from the Judge on evidence of Google’s dominance in the market for open web display, where Dunn had suggested that Lee looked at transaction volume vs. the product market. She explained that log level data used by Dr. Lee to form his analysis is the evidence.

She then brings us home with a fast and fiery closing. She says as she watched this trial play out, she’s reminded of Dickens’ Tale of Two Cities - it’s up to the Court to decide which is consistent with what they heard on the witness stand - particularly from market participants (again nodding to Google calling only one not on payroll). Opinions of their experts were, she says, at war with the facts.

She says it’s rich to be accused of gerrymandering market definitions by a company that instructed employees to use preferred terms and speak freely only in automatically deleting chats, asking Judge again for adverse inferences. Wood claps back at the accusation by Google of DOJ making up points for litigation when it’s Google that switched positions between the California case and this one.

We go back to the 4 monopolists chart and how output expansion is not indicia of a lack of monopoly behavior. Markets can thrive while still having monopolists in them.  She reminds the Court that while defense referenced TikTok, Amazon, and Facebook repeatedly, we never heard how they compete for open web display ads. Nine out of ten open web display transactions pass through Google’s hands, she explains.  Google’s dominance was so pervasive that the industry came together to develop a “hack” to get around an ad server not serving its own customers.

All roads lead back to Google. After a HB auction, the price goes right back into the ad server for Google’s last look, she reminds. And so, despite HB being “alive and well today,” these figures are included in the market shares that DOJ presented to the court.

She speaks to the irrationally high rents Google is able to charge while degrading product quality, and the view by the market of Google as an “authoritarian intermediary,” calling back to Xoogler Chris LaSala’s statements. Wood calls Google on their games, saying that when you open up the aperture and zoom out you lose sight of the details and the realities - maybe all that matters is buyers and sellers, for example - but that this obscures the purpose of antitrust. As Judge Brinkema makes a Dickens/guillotine joke, Ms. Wood brings her rebuttal to a close.  DOJ isn’t asking the Court to overturn Amex or Trinko - simply to apply facts as the Judge has heard them to the law as it has been written.

‍

What comes next?

Judge Brinkema closes us out by reiterating that this is the best tried case of her career, and she’s incredibly happy with how both sides argued their cases and treated one another.

With that, we wait for a decision. No further update from what we heard at end of trial, which was that EOY may not be possible, seeming to suggest a Q1 ‘25 opinion.