Google's monopolistic practices to be explained in detail for the third time in US federal court
A piece I submitted to the American Economic Liberties Project as they were looking for someone to provide daily coverage of the upcoming trial.
Like I said in the subtitle, this is an article I wrote for the American Economic Liberties Project when they were considering whether to hire me for daily reporting on the Google antitrust trial starting tomorrow, September 9th, 2024, in Virginia.1 They ultimately went with a lawyer by the name of Tom Blakely, but I’m pleased enough with what I wrote that I thought I’d post it here. If you’re interested in the trial, you should follow Big Tech On Trial to get that daily coverage. For more general monopoly-related news and analysis, check out Matt Stoller’s BIG; he’s got some good coverage on everything from Boeing to WeWork to Varsity Brands. Please excuse the somewhat abbreviated introduction; the article is written to fit as part of Big Tech On Trial’s ongoing coverage of this saga.
Since this trial is centered on a part of the internet that few members of the public have much visibility into and the allegations against Google are somewhat more involved than in the prior case, I’ll be providing an overview of the ecosystem and what the government and Google have to say about it; this should give you a reasonably clear picture of what the trial is actually about. I’ll also talk about a couple of major pretrial highlights, and how they relate to what’s to come.
What is a “Market”, Anyway?
The advertisements at issue in this case are the most basic sort you find on the internet, often small banner ads supporting a blog, webcomic, or the like. This sort of advertising plays a major role in supporting small to medium-sized communities outside of major platforms like Facebook, Youtube, reddit, or Substack. For many years now, it’s been a key revenue stream for newspapers as well. Because these sorts of sites, “publishers” in the parlance of the suit, have an extremely fluid audience, they must sell advertising in an extremely fluid way as well. The result that advertisers and publishers have converged on is a set of automated systems that take a summary of what various advertisers would pay a publisher to serve an ad to particular types of customers, together with a description of the publisher’s incoming audience and advertising space, and then dynamically auction off slices of the publisher’s audience to the best-suited advertiser(s). All of this happens as you, the user, load a page in your web browser, and the result is (ideally) that an auto dealership in the next town over can afford to advertise in the national newspaper you subscribe to, without having to itself be part of a national franchise.
The antitrust question in this case, as framed by both the government and Google, is more complex than the one in the trial Google lost at the beginning of August. In the internet search arena, Google - like Microsoft before it - sought to make its product the one most users would go with by making it the first choice that any user might see. It was a simple scheme to dominate the market through sheer size, starve potential competitors, and collect effort-free rents on the product forever after with no need to improve it.
In the digital advertising arena, however, Google has applied a more sophisticated model to build a more subtle engine for market power. The Department of Justice and its state-level co-complainants describe a scheme to capture both ends of the small-scale web advertising market, using a series of business acquisitions to collect a product at each layer of a middleware software stack, and then subsidizing the value of its tools with captive market share from elsewhere on the internet while locking alternative tools out of its stack, until the vast majority of these small scale, high speed advertising transactions passed through Google’s proprietary pipeline. The resulting monopoly is much closer to the kind of system a pharmacy benefit manager constructs, as opposed to the classic monopoly structure that Standard Oil or AT&T relied on.
To the government, the market that Google sought monopoly power over consists, on the one hand, of internet publishers too small to support an internal advertising ecosystem, and on the other hand internet advertisers too small or too regional to advertise across the internet as a whole. These actors need some sort of intermediate service to reorganize publisher ad-space inventory and advertiser purchasing power so that many different advertisers can divide up the audiences of many different publishers in an efficient way. With a popular tool at every level of the advertising stack, plus an independent source of advertising demand through search advertising, some part of Google’s ad stack ends up involved in almost every ad purchase that uses any sort of middleware. This means that, unless a publisher has an independent audience so large as to demand that advertisers buy their ad space directly, not to mention the technical capacity to enable advertisers to target relevant slices of the audience in question, Google is able to impose a tax on every advertiser/publisher transaction.
According to Google, this is all fine because advertisers can maybe switch to advertising on Facebook, or through mobile apps, so if Google’s tools are involved in so many middleware transactions, it must be because Google’s middleware is really good, and anyway no consumers were harmed. As to the factual allegations, say Google, well, those quotes may come from our documents, and we may have built certain mechanisms into our advertising tools at various times, but who can say what the effects were? And in any case, we definitely didn’t have any anticompetitive intent when we tried to “build a moat around our demand.”
Judge Brinkema was skeptical of Google’s arguments at the motion to dismiss stage. We shall see how she responds to the various experts and other witnesses in the trial to come.
The Spectre of Sanctions
As far back as the private antitrust suit brought by Epic Games (concerning Google’s app store policies), Google has faced scrutiny over its document retention practices. Readers will no doubt remember the relevance of those practices in the recently concluded search trial. The government has pointed to a laundry list of dubious choices by Google, from automatically deleting chat messages to running employee seminars on how to avoid creating discoverable documents.
This has led the DoJ to request evidentiary sanctions in this case, as they did in the search trial. The redactions in that filing make it hard to determine the details of what Google was interested in hiding, but the broader picture remains quite damning. Quoting heavily from Judge James Donato’s opinion in the Epic case, the government’s motion describes a long-term project by Google to build a culture of preemptive discovery evasion among its employees by encouraging them to treat ordinary communications records as inherently hazardous, and to create a haze of potential privilege around what records they did produce by rotely incorporating attorneys into every conversation. This culture paid dividends for Google during this investigation, resulting in substantial slowdowns in the release of many email chains and the outright destruction of several years’ chat messages.
“It is no wonder then that this case has lacked the kind of nakedly anticompetitive communications seen in Microsoft and other [Sherman Act] Section 2 cases…. Google clearly took to heart the lessons from these cases. It trained its employees, rather effectively, not to create ‘bad’ evidence.” - Judge Amit Mehta
Google was ultimately able to avoid sanctions in the search case - by losing. In that case and this one, the government’s requested remedies largely boil down to a presumption that the destroyed evidence would weigh against whatever arguments Google makes. They offer the possibility of a public reprimand, should the judge be so inclined, but ultimately don’t seek a concrete penalty beyond what they ask for the underlying monopolistic conduct. Judge Mehta felt that what evidence escaped Google’s habitual documentary purges was sufficient to find against them, and thus that there was no need to explicitly infer that the evidence destroyed would also have been unfavorable. He nevertheless went out of his way to note, in his decision, “It is no wonder then that this case has lacked the kind of nakedly anticompetitive communications seen in Microsoft and other [Sherman Act] Section 2 cases…. Google clearly took to heart the lessons from these cases. It trained its employees, rather effectively, not to create ‘bad’ evidence.”
On August 27th, Judge Brinkema heard argument on how she should respond to this conduct; she took the matter under advisement and did not issue a formal ruling. Given the way things went before the other two judges, it is hard to see how Google could have actually gained ground in that hearing; the question is more whether they were be able to stop the bleeding. I was not at the hearing and the transcript is not yet available, so I can’t personally comment what happened, but reports from those who were there seem, one might say, ominous for Google.
Peerless
One other interesting twist in this case came in late May, when Google, perhaps afraid of what a jury like the one it faced in San Francisco might endorse, sent a cashier’s check for about $2.3 million to the DoJ to cover the alleged monetary damages involved in this case. Since that was the hook on which the government’s jury demand rested, Judge Brinkema agreed in June that a jury would not be needed after all.
Based on Judge Brinkema’s analysis, Google was always going to have the power to decide whether the case would go before a jury, regardless of whether they dispute the core allegations, because the government was required to specify some maximal amount of monetary damages it was claiming, and Google would be able to moot the issue by paying that amount prior to trial. It’s ultimately a consequence of the fact that the money damages really are ancillary to the core question at issue in the trial: whether or not Google is a monopolist.
A monopolist certainly can cause harm in monetary terms, extracting excessive rents from the economic process(es) they control because nobody, not even the entire rest of the economy, has enough market power to tell them “no” and make it stick, but there are also other important ways that monopolists can and do create harm. Most relevant to the internet, monopolists are a key driver of enshittification, the gradual degradation of once-valued products and services that ensues when those products or services become absolutely required to meet some customers’ needs. The costs of enshittification are, at best, difficult to measure in dollars, and attempting to do so will often distract from efforts to prevent these harmful processes from continuing or recurring.2 The relevant tool here, as in many regulatory matters, is the injunction, and in the United States we have decided that juries do not have a role in crafting injunctions.3
Is that a good policy choice, as a general matter? I am honestly not sure. I certainly don’t like the idea of a monopolist being able to bamboozle a jury with complex arguments about whether and to what extent their customers are truly locked in to the monopolist’s products, but monopolistic interests have proven quite capable of bamboozling judges in exactly this way. At the very least, it seems foolish to let the monopolist decide who’s going to be evaluating their claims.
In this case, at any rate, Judge Brinkema will be the only one deciding whether Google truly has built itself a monopoly in web advertising, and she would always have had the final say as to what should be done about it.
Not to be confused with the antitrust trial that Google lost about a month ago in DC, or the one they lost last December in San Francisco, or the one in London I’m less interested in looking up for this bit…
What is the monetary cost when there’s an income floor to participate in a classic American sport? How about when a community-compiled history of a mascot character is replaced with an advertisement? Does it matter? Or is it more important that these things not happen in the first place?
In addition to that article on policing I’ve theoretically owed the world for over a year now, I’ll likely be putting out a piece on my experience as a grand juror this past May, in which I expect to once more examine this question of how the US legal system uses (and talks about) juries.