DOJ v Google Closing Arguments

I’m back! Or rather, I’m still in DC at the Amtrak station eating a Jersey Mike’s sub and pounding out this summary of the final day of the ad tech trial. Read to the bottom for my predictions on the verdict.

The DOJ and Google gave their closing arguments this morning in the final day of the US et al v Google, LLC trial. We expect to get a final ruling from Judge Brinkema by the end of the year, so someone better look in their stockings for a lump of coal.

In the rebuttal to the closings, Mrs. Tarver Wood, the lead for the government, said “It’s like a Tale of Two Cities, is it the best of times or is it the worst of times?“ We’ll allow her the slight mangling of the quote because it was apt after listening to plaintiffs and defense give wildly different views of the evidence, the market, and the case.

There wasn’t any new evidence, since it was a closing, but we can see the arguments varnished and the evidence filtered to only the key parts each side wants the judge to consider.

Once, Twice, Three Times a Monopoly

The DOJ started with the Once, Twice… quote, which made me chuckle a bit. The “Three Times” refers to the allegation of three separate monopolies in publisher ad serving, ad exchange, and advertiser ad networks. They put their argument into 6 bullet points, which I’ll reproduce here, since I can’t do better:

1. The market definition offered is valid

2. Google has monopoly power

3. Google engaged in anti-competitive conduct

4. The conduct was not lawful under a “refusal to deal”

5. There were no valid product design decisions to back up the conduct

6. Google’s arguments to the contrary are invalid

I won’t summarize their whole closing, but will bring up some interesting points.

The DOJ made the argument that while both sides had witnesses, there was an “evidentiary disparity” between the two: While they had called witnesses who were real practitioners, Google’s witnesses were almost all on the payroll.

They pointed out that Google is sometimes conflating the definitions of “competition,” for example using Andrew Casale’s comments that post header bidding the market was “hyper-competitive” out of context. He was referring to competition in the auction, not the market.

Anticipating much of Google’s argument about market definition they pointed out that having a broad rivalry with another company, like Amazon, does not imply competition in every sub-market of relevance to antitrust. Underminding some of the expert witness testimony from Dr. Israel (that I covered extensively here), the DOJ made a series of clear points, for example that TikTok and others cannot be used as substitutes for DFP, nor can other sub-market tools like DSPs, nor can a publisher magically switch to direct sales from programmatic, etc.

Two separate Supreme Court precedents

Google’s closing focused on Amex and Trinko, boldly saying that to find for the DOJ would require the judge to ignore two different Supreme Court precedents! Those are fighting words.

For those not following too closely at home, “Amex” is the case where the Supreme Court decided that a two-sided market should be considered as a single entity for antitrust purposes. Google argues that when an ad impression is bought on the exchange, the “market” to evaluate for antitrust is effectively every atom in the ad tech universe. At that very moment there is no such thing as an ad server market, or an exchange market, or an ad network market, they are all bound together. Needless to say, it is pretty hard to claim someone has a monopoly on everything. More specifically the case allows for “indirect network effects” which means you can benefit from a monopoly on the other side of the market, such as the way AdX benefits from the DFP monopoly.

“Trinko” is the “duty to deal” case where a company with that very name tried to claim that Verizon should be forced to build a bunch of stuff to allow them to interoperate, and Verizon declined. The court found that Verizon did not have to “deal” just because they were a monopoly (more or less). The DOJ’s argument is that if you cannot force Google Ads to bid into other exchanges, and you can’t force AdX to interop with other ad servers.

Some other Google closing highlights (all roughly paraphrased):

• Repeat of allegations that Google’s opponents want the ad server to be a community good (note, they left out the part where one of their engineers said this in testimony and Jonathan Bellack’s email that said it about AdX!)

• Slide of Microsoft ad tech stack calling them a major competitor with some advantages even Google does not have (LOL)

• Slide of Criteo strategy saying they are on the buy- and sell-sides so competitive. Also the phrase “none of us in this courtroom ever heard of Criteo before now” — catching strays!

• Slide of the Google buy/sell side diagram but with Amazon products like “TAM” kind of rudely shoved in, to make the point that the diagram should not have omitted them in the first place.

• Memo from Brian O’Kelley to Microsoft CEO Satya Nadella from 2019 saying “Google is on the ropes and its time to get them”

• (intentionally) confusing “direct” with 70% of the revenue of some publishers with “programmatic direct” growing quickly to show competition

• Making the point that Google reacted properly to header bidding since it was competition, and that’s what antitrust law wants.

• The definition of advertiser ad network should include DSPs since most companies that spend a lot on Google Ads also have seats that spend a lot on DV360.

• The fact that DFP is free for most publishers (OK, fine), and that most publishers don’t use AdX (I think this is either wrong, or just the fact that the free version of DFP uses AdSense, not AdX).

While the Google close had a lot of good points, there were many, many times where they made arguments that just made zero sense out of the context of ad tech practitioners. It was definitely a “two cities” moment.

Judge’s choice

Previously Judge Brinkema indicated that she would likely have the draft of an opinion by today’s hearing. So the key to the hearing was to watch what she said and asked, more than what the lawyers did. Below are the questions she asked along with my reading-the-tea-leaves interpretation.

Judge: Why did the DOJ not have any advertiser witnesses, only publishers and agencies?

Tea Leaves: The weakest allegation against Google is that it has a monopoly in the advertiser ad network market (AdWords). Since AdWords is a non-transparent media buy with undisclosed margins, almost by definition anyone that spends money there is happy — or at least satisfied enough to keep spending. Leaving advertisers out of the case has a smell about it.

Judge: Does it make a difference whether the court finds there is one or three markets?

Tea Leaves: Same point as above. The Judge is probably convinced DFP is a monopoly, and possibly AdX, but not AdWords. She wants to know if DOJ thinks the legal basis will still be there.

Judge: What about latency?

Tea Leaves: The Judge was actually a little confused here about the subject, because latency was mostly an issue with header bidding and that was not the topic being discussed, but she still had a point. Part of Google’s argument against interoperability with other parties is technical feasibility and other non-monopolistic concerns, and she’s taking those to heart.

Judge: You said there were better ad tech products out there, which are they?

Tea Leaves: The judge is concerned that all the evidence against Google is coming from a concentrated group of folks with chips on their shoulders. But she should remember the testimony from News Corp and Daily Mail that they evaluated alternatives and could not switch.

Judge: There’s a much heavier emphasis in the case on the sell side. There are three tools being evaluated, but only two have been made part of the case.

Tea Leaves: Hoo boy. If it was not clear the DOJ has not made a case about AdWords being a monopoly, this the smoking gun query from the bench.

Judge (to Google): You’re on dangerous territory when you make claims about what people at Google were thinking.

Tea Leaves: Since there was an Adverse Inference ruling because of the company’s terrible document retention history the Judge assumes the worst for everything from Google that is ambiguous.

What’s likely to happen?

I’m only going to prognosticate about the ruling in this case, not the remedies.

There’s actually a lot of confusion among court observers as to the timing of remedies, and the process by which they will be evaluated. Everyone seems to think there will be some hearings or otherwise feedback from parties about the remedies, but not a full trial like in the search case.

Here’s my betting card as of right now: