Google’s Lawyers Pitch Competing Explanation of Ad Bidding, Reframe Case in Strong Day for Defense

While the government kept its focus on exhibits like these discussing competition, Google successfully reframed parts of the case on the second day of its case-in-chief, as we head into the final days of the trial.

In yesterday’s article summarizing the first full day of Google’s case-in-chief, I wrote that I had been hoping Google would finally bring it, and like a number of watchers of this trial, was disappointed that the company’s lawyers put on a case that felt stunningly punchless and without direction.

After all, these are some of the best lawyers there are. Much of the discussion after Monday was this sort of Wendy’s “Where’s the beef?” confusion as to why Google had turned in such a muted performance.

It was like turning up to see 1990s Michael Jordan or 2000s Tom Brady, and instead of seeing what you came for, you saw a middle of the roster bench player going through the motions—indifferent as to whether they won or lost, with their technique and desire for winning deeply unclear.

Well at long last, Google brought it on Tuesday.

I don’t want to get too hyperbolic, but this was a pretty good day for Google.

Tuesday was dominated by the testimony of Google expert Dr. Paul Milgrom, Professor at Stanford University and the Chairman of “Auctionomics.”

Dr. Milgrom is Google’s counterweight to the government’s Dr. Ravi.

As soon as his testimony began, I realized he would be a formidable witness, not just because of his outstanding intellect and subject matter expertise, but because he is the kind of witness that is simply challenging for a counterparty to deal with. An older gentleman, he was equal parts charming, intelligent, poised, unflagging while enduring hours on the stand, and spoke with the quiet confidence and humility of the wisdom of his years.

His testimony began with a timeline of Google’s ad products. I have found this testimony and the related demonstratives from Google to be a strong tool in its case. (Unfortunately, Google - unlike the Justice Department - is posting its exhibits and demonstratives days after the fact, so I can’t share it with you here.) It offers an organic and chronological framing of how Google’s ad empire came to be, not simply this giant incumbent Frankenstein of mergers and acquisitions that you might imagine from only having heard the government’s case.

From there, Dr. Milgrom presented his view of the “waterfall” ad auction process—and used graphics depicting this labeled as his “corrected” version, side-by-side with the government’s purportedly incorrect exhibits from its case.

I’m not sure if Google’s legal team heard all the criticism following Friday’s “spaghetti football” exhibit and decided to hire someone new to make their exhibits—but today’s exhibits were fantastic, and I found them to be quite potent in presenting a competing theory to what the government has alleged.

However, there’s still something strange about the “waterfall” itself—in that it is essentially a he said/she said as to how this actually works. The government has told us bidding works one way, mainly that Google can win auctions via Last Look (where Google can see the bids of competing advertisers and beat them by $0.01) or can win by merely bidding at their historical average. Google is telling us a different story, where advertisers first decide if they want to bid on the impression, and that the bid amounts and the instrumentality of deciding on the winning bid is not as preferential to Google as the government alleges. Unless somehow the Court looked at (and understood) the ocean of software code this process runs on (which is likely not possible)—there’s really no way to know who is right. After two weeks of the government dunking on Google over this—merely muddying the waters in this way might be all Google needs to do to take this line of attack out of commission.

From here, we then got a considerably long colloquy between Judge Brinkema and Dr. Milgrom. As I said yesterday, whenever a judge asks questions, particularly substantive questions of a witness, it is very important to pay attention to because the judge is showing you what they are thinking about.

This is only a hunch, but I got the sense Judge Brinkema looks quite favorably upon this witness and finds him credible. A day after telling Google’s lawyers a witness wasn’t adding anything, leading them to take him down from the witness stand—basically changing the channel to something she found more interesting—Google lawyer Jeannie Rhee, who was conducting the direct examination, stood by for an extended period while Judge Brinkema and Dr. Milgrom chopped it up about the finer points of the bidding system.

One thing Judge Brinkema was particularly focused on was how an advertiser bidding $7 could win at $3—Dr. Milgrom explained that if $3 is all that was necessary to beat the nearest rival, advertisers would only be on the hook for the amount necessary to win. Judge Brinkema also asked what happens in the event of a tie between two advertisers placing the same bid. Dr. Milgrom said in that case the winner is randomly decided.

Keep in mind all of this is happening in fractions of a second using opaque algorithms and - increasingly - machine learning, or “AI.”

One thing conspicuously absent, and growing ever more conspicuously absent in Google’s case are the role and function of rival ad exchanges and ad agencies. We hear very little about them. In Google’s version of events, rivals really do not seem to factor in much, nor does its beleaguered 20% take rate, publisher dissatisfaction or much else—just advertisers.

At this point, I jotted in my notes it was beginning to make some sense (but not dispositive sense) why Google is going with this light touch approach in its case-in-chief.

There are so many pitfalls and landmines in the fields of bad facts put forth by the government—trying to army crawl through there and defuse all of them may just be too messy and not worth Google’s time. As we are starting to see, in Google’s world, customers find the world’s largest supply of demand, it keeps out spam and bad stuff (we’ll get to this), innovates, and the thing generally works—rivals ad exchanges (who it believes it is not required to deal with) and aggrieved publishers be damned.

At this point, the waterfall and a lot of the “auctionnomics,” to borrow the name of Dr. Milgrom’s organization, seem to be a wash between the government and Google. By midday, Google, in my view, had successfully muddied the waters. It is not that the government did not put forth a compelling case on this—it’s just hard to tell whose version of events, more specifically, whose technical explanation of how this works under the hood, is more likely than not the correct version.

As Dr. Milgrom’s direct examination was gaining steam, the government tried to short circuit it by objecting on the grounds the evidence was becoming cumulative—the third rail in the Eastern District of Virginia—but Judge Brinkema overruled it. Instead, anticipation continued to build for the cross examination of Milgrom.

Dr. Milgrom then testified on Prof. Weintraub’s report, and first price auctions.

While the government did lay out a lot of convincing evidence on these more technical topics, particularly Last Look and Google being able to crush rivals by bidding just a penny more for an impression than someone else, the ensuing network effect, along with other excesses, while first impressions (no pun intended) matter, final impressions matter too.

In other words, while the government had a strong two weeks to start this case, what one hears and remembers last also matters, and here, Google could be seen softening the iron of some of the government’s case.

Judge Brinkema then chimed in with a very incisive and noteworthy question, asking Prof. Milgrom about the learning processes (machine learning) that require access to data. This was a key gripe among advertisers and publishers, that Google was not transparent and limited access to its data. It was interesting Judge Brinkema was thinking back to this, and reinforces the extent to which she has demonstrated a deep understanding of the issues in this case.

Prof. Milgrom then discussed what Google did during the transition from First Price to help advertisers. He mentioned that Project Poirot (which involved “bid shading” to lower bids on rival ad exchanges) attempted to detect auction rules and optimize bidding accordingly to the benefit of advertisers. Once again, in Google’s world, there’s no room for rival ad exchanges or agencies—just advertisers and their dollars.

But more than this, Google has also steadily been building up a classic tech industry framing—that Google is not actually making bids, setting prices, or controlling the process—it is in fact the advertisers and publishers doing so. Google has layered into its direct examinations a number of questions intended to get this into the record, to lay the foundation of the argument that some of the conducts held out as problematic on Google’s part are not actually its fault—but in fact the doing of advertisers, and that Google simply puts parties in a position to transact.

This is sort of the entire basis of Section 230, the DMCA, and the philosophy of how not only a good amount of the law applies to big tech companies, but the business strategy of the companies themselves—that they are trying to “be like water” in the words of Bruce Lee—merely platforms that act as membranes through which buyers and sellers deal, but that Google itself has little no hand in, and especially no liability in, whatever happens from there.

I think this argument from Google is stretching things too far, and I doubt Judge Brinkema will buy an attempt to try to dilute Google’s malfeasance as documented time and again in internal emails and witness testimony by trying put things on the advertisers by painting them as being in the driver’s seat of this auction process.

Cross examination was led by Jeffrey Vernon of the Department of Justice. I don’t believe I’ve mentioned Mr. Vernon thus far by name in these articles, so I wanted to take a moment to recognize the superb job he has done. Mr. Vernon is a phenomenal litigator, who behind Julia Wood, has been a standout second chair attorney on this case. He is well spoken, poised, calm under fire, and has been adept at leading the balance of the direct examinations for the government in this case. He has a commanding, warm, but firm courtroom presence, and rarely has fumbled.

With that said, Mr. Vernon had his work cut out for him with Dr. Milgrom. As has been seen in this trial, and as is the case with any sizeable litigation that is headed to one—lawyers must divide up the work, and decide who will be responsible for the direct and cross examinations of each witness. It is not an exact science, but the idea is to match lawyers with witnesses with whom they will be an effective and balanced fit. This stage managing of sorts means being able to match the energy (and the speed and wits) of the witness without coming across as a bully, or alternatively being pushed around one’s self.

The challenge here for Mr. Vernon is one I’ve encountered myself before and can be challenging to overcome—a younger guy as the lawyer going up against a witty older gentleman who is the witness. A younger lawyer cannot be seen as being overly aggressive in this case, but a witty and vexing witness can make this dynamic challenging by limiting how much you can push back. By presenting with charisma, charm, wit, and as one person described to me during a break, “grandpa vibes,” an outside observer who hails from most of the world’s cultures intrinsically wants to like Prof. Milgrom, and might see this as a young guy who is trying to pick on the older guy, waiting for the older guy to “take him to school.” Moreover, Prof. Milgrom has not given the Court any reason to dislike him, and he does not come across as an intemperate, dishonest, egomaniacal, or ornery individual of the sort one roots for the counterparty to hold to account on cross examination.

Mr. Vernon however did a pretty good job all things considered, although he did not land as many blows as we are used to seeing—they just weren’t there. Mr. Vernon had Prof. Milgrom admit to the advantages of First Look (where Google can see and bid on ad inventory before it was made available to other advertisers or exchanges) and highlighted a Google email discussing competition and strategy with respect to Google rival Rubicon. Cross examination then covered most favored nations rules, and how Last Look advantages AdX.

Moving on to Per Bjorke, Director of Product Management on the AdSpam Team at Google. I’ve noted routinely that concerns about “spam” or “fraud” or “bad actors” make a sort of daily cameo appearance in this case, with Google making a fleeting reference to these issues as a sort of pretext for anticompetitive behavior that never is developed beyond that—something that is common among big tech companies that often use security concerns as a smokescreen for antitrust violations.

That all changed today.

Mr. Bjorke took the Court through an exhaustive summary of the efforts Google undertakes to detect and limit spam in its ads business. I’m not sure it’s worth your time to reproduce all of this testimony, in large part because it is not connected to the legal issues in this case, apart from what I will describe below. His testimony was legitimate—Google really does work hard at stopping waves of criminality, malware, and spam that emerge from dark corners of the internet and the Gotham City of cybercrime that is Eastern Europe, as Mr. Bjorke would explain with respect to the “3ve” bot network that the FBI and the DOJ were able to catch working spam into Google’s ad network in a massive $30-40m fraud. Mr. Bjorke was genuine, and was also very clearly removed from the crooked machinations of the ads business inside Google, working in the security and spam prevention end of the house.

The testimony was replete with all the familiar tropes of this issue—at times I felt like we were about to start hearing about Russian interference in Google Ads.

I think this testimony was relevant because it made the spam and safety topic more than a daily checkbox, and underscored not only value Google creates by virtue of its strong position in the market structure, but the potential perils of the disintermediation of Google with respect to potential remedies should Google lose.

However, Google would have been well served by spending more time explicitly connecting this testimony with a pro-competitive basis for the conducts the government has taken issue with, in other words arguing that Google making all of its demand open to other ad exchanges would prevent Google from being able to ensure its ad inventory was not appearing on objectionable, spammy, or fraudulent content. After all, Mr. Bjorke discussed how important these concerns are to advertisers.

On cross, the DOJ impeached Mr. Bjorke by bringing in evidence a lot of other industry participants, including many rivals or aspiring rivals we’ve heard about in this case, contributed to the anti-spam measures Google has touted and taken credit for.

To end the day, we got deposition video from, Kenneth Blom, Chief Business Officer at BuzzFeed, a rare Google publisher witness. There was nothing here to write home about.

Lastly I want to end this article by finishing a discussion I started in yesterday’s article, and close the loop on the key antitrust analyses in this case. Yesterday I explained the difference between a refusal to deal case (the analysis Google would like) and a tying case (the analysis the government would like).

The second prong of this discussion—is how might Judge Brinkema actually decide which is appropriate. Without getting into which I think is or right or believe she will decide—I want to look at how this decision turns.

It starts with market definition. This is a question of fact rather than a question of law. In a bench trial, the judge, in this case Judge Brinkema, decides what the relevant market it is. If she finds the relevant market is the market for open web display ads, which the government has zeroed in on, that’s a bad outcome for Google. If instead she finds that the relevant market is the market for digital advertising at large, that’s a bad outcome for the government. In the second case, it would be harder to show how the Google conducts the government has challenged have generated anticompetitive harm in a market that broad. In the first case however, as we have seen during this trial, finding the harm done in the market for open web display ads due to Google’s conduct is much clearer.

In Google’s Proposed Findings of Fact and Conclusions of Law in this case, the defendants essentially argue that the government is trying to avoid unfavorable refusal to deal case law:  “Plaintiffs cannot avoid Supreme Court law on a duty to deal by alleging that the denial of access to rivals is a tying claim.”  (Google’s Proposed Conclusions of Law, 328)

The government for its part basically alleges the massive breadth of Google’s anticompetitive conducts is so heavy as to go beyond mere refusal to deal analysis.

That’s all for today! I’ll be back tomorrow with what will either be the final, or second to final day of this trial before Google rests its case.