Day Three: Did Judge Brinkema Give Google a Ray of Hope?
Not exactly. But Judge Brinkema reminds the parties that a judge cannot just assume a party will break the law again, without evidence of recidivism.
True to Julia Tarver Wood’s word yesterday, the DOJ started right in with their technical experts on day three of the Google ad tech remedy trial. We heard from an economist, an investment banker, and a computer scientist, all three of whom testified at the earlier liability hearings. I guess you don’t break up a winning team.
Judge Brinkema took a more active role throughout the day, and one of her interventions raised something that DOJ will have to treat with particular care. It came a couple of hours into the testimony of Robin Lee, Professor of Economics at Harvard university and industrial organization (or “IO”) specialist.
Lee’s argument was rooted in three objectives:
Stopping Google’s anti-competitive conduct;
Resolve competitive harms from Google’s past conduct; and
Preventing the re-emergence of Google’s monopoly.
Although Lee testified that “behavioral” remedies (that is, telling Google not to do bad things) to be useful, especially for their relatively quick effects, he concluded that the long term achievement of those three objectives would require Google to divest both AdX and DFP, or “structural” relief.
Lee had done a number of studies of the two relevant markets—(1) publisher ad servers and (2) ad exchanges—to see what measures would improve overall customer welfare. And Google AdWords loomed large in his analyses. AdWords is Google’s huge buy-side network of advertisers looking to purchase display ads on the open web, which several witnesses have identified as the original and continuing source of Google’s power in ad tech. Since Google was not found liable for monopolization of the buy-side, Google is arguing in the remedy phase that AdWords should be off limits.
The DOJ, however, includes among its behavioral remedies a prohibition on Google bidding directly from AdWords into its publisher ad server DFP, and Lee’s testimony supported this measure with economic analysis. In the event that AdX is divested but DFP is not fully divested, a direct connection from AdWords to the remainder of DFP could would essentially re-create AdX and preserve Google’s illegal monopoly. Basically, Google holds such high market power in both the buy-side network of advertisers and the sell-side network of publishers that even without control of the exchange, all it has to do is glue the two sides back together. That would thwart Lee’s objective three.
But remedial objective three - preventing future monopolization - is what brought Lee a polite but firm rebuke from Judge Brinkema.
Here’s what happened. As an economic matter, Lee said, it is most effective to prevent the ability to commit a harm. Past behavior predicts future behavior, absent changes. Matthew Huppert, questioning Lee, turned to Google’s remedy proposals, and it was when Lee was explaining the inadequacy of those proposals to prevent future monopolization that Judge Brinkema stopped him. As Capitol Forum recounts it (I didn’t get exact wording), “‘A lot of your opinion is based on the assumption that Google would attempt to re-monopolize,’ she told him. ‘There is a presumption or an assumption that Google would try to re-engage in violating antitrust laws.’”
Of course, this is following two days during which many of the six witnesses expressed a similar attitude toward Google. The idea that Google cheats whenever it can and would do so again was pervasive, and unsurprising coming from business people who feel victimized by Google. But on day three, Judge Brinkema brought the trial up short during Dr. Lee’s testimony.
The next statement she made is the nuanced one that will require some care from DOJ: As Capitol Forum paraphrases it, “Brinkema asked if there were any cases where Google was enjoined as a part of an antitrust violation and broke that injunction.” I recall her asking what evidence there is of such a history. Of course, Judge Brinkema knows very well that Google has been held liable in three separate antitrust trials in the United States alone (including once by her). She is pointing out that three convictions in separate markets at roughly the same time holds a different legal significance from being convicted (or enjoined) and then re-offending. Only the latter is called “recidivism.”
In the DOJ’s remedy proposal and in Julia Wood’s opening statement on Monday, Google is called “a recidivist monopolist.” So Judge Brinkema’s statements today could indicate she has doubts that description is accurate, or that evidence of it has been produced. And that could mean that the measures the court can take to achieve Lee’s objective three—which is basically DOJ’s objective three: “ensure that there remain no practices likely to result in monopolization in the future,” quoted from United States v. United Shoe—those measures must be targeted at and responsive to the illegal practices Google has actually been shown to have committed. Google cannot be simply assumed to be a bad actor.
Judge Brinkema said more than that today, though. She said why it matters, why the law cares that we target remedies at untenable situations rather than untrustworthy parties. What holds up agreements is lack of trust, she said (my paraphrase). Conflicts do not resolve when people feel unfairly abused. The DOJ still has substantial Supreme Court precedent for courts taking bold measures, when necessary. In my view, Judge Brinkema told us today that courts should not easily go past what’s necessary. That said, there are differing views here; Check My Ads and DCN’s Jason Kint suggested that Brinkema is simply exploring various options, and the Google-backed Chamber of Commerce (unsurprisingly) sees it as a break for Google.
Lee’s testimony was a particularly apt time for Judge Brinkema to make her point. Not because he was particularly strident or more caustic than others, far from it. (She said that mistrust of Google has “permeated” the witness testimony so far.) But Lee’s economic analysis is the DOJ’s preferred methodology to distinguish what is necessary from what is excessive or punitive. As he said, firms maximize profits, that’s just what they do. He has good arguments about both the incentives that face Google and the incentives that face other ad servers and exchanges - all in service of supporting structural remedies, aka break-ups.
Following Lee was testimony by Paul Crisci, an investment banking expert. He was there to give assurance that divestiture of AdX and DFP would attract a pool of qualified buyers both credible and motivated. And to sort of educate the court on how exactly it would be managed. Since it seems intuitively clear that the problems, if there are any, with selling huge and valuable parts of Google will be technical problems, not business ones, Crisci was able to make his case.
That brought us to the last witness of the day, John Waissman, professor of Computer Science at the University of Minnesota, and a specialist in distributed systems and cloud computing. Waissman’s job is to demonstrate that divestiture of AdX and DFP is technically feasible, and we’ve explained that technical feasibility is a critical variable in the divestiture. In the hour or so he had at the end of the day, he certainly made a good start. But most of his testimony will be tomorrow morning, so I will take this up then.