The U.S. Justice Department (DOJ) late last month filed its much-anticipated initial proposed final judgment in the Google Search antitrust case. The proposal—to use a bit of baseball parlance—swung for the fences. Maybe they’ll get a hit, or maybe even a home run. Or not. Dodgers superstar Shohei Ohtani hit a whopping 54 home runs in 2024, but he struck out three times as often.
Back to the case. Readers of this space are no doubt aware that Judge Amit P. Mehta bifurcated the Google Search proceedings into separate liability and remedy phases, as permitted by the Federal Rules of Civil Procedure—sometimes for the better, if often for the worse. No doubt you’re also aware that, in an Aug. 5 opinion, he found Google liable for violating Section 2 of the Sherman Act in two markets.
As I noted at the time, his memorandum opinion was, in many ways, a careful and thoughtful one—which is not to say it was either correct or uncontroversial. I happen to think there were key errors of law and fact in his decision, and I’m not alone in that regard. For more concerns about the case, see, for example, my International Center for Law & Economics (ICLE) colleague Geoff Manne here; my ICLE colleagues Brian Albrecht and Gus Hurwitz here; Greg Werden, former chief counsel for economics at the DOJ’s Antitrust Division, here; former FTC General Counsel Alden Abbott here; and me here.
Understandably, the DOJ’s proposed remedies have now attracted a good deal of attention and controversy, as well. DOJ proposes a combination of behavioral and structural remedies that can fairly be described as “broad” (perhaps in the way that the number line can fairly be described as “long”). At issue in the liability phase were certain contracts that Judge Mehta found to be unlawful: Google entered into agreements with certain device manufacturers (notably, but not exclusively, Apple) as well as Mozilla, which developed and markets the Firefox browser. Under those contracts, Google is the default general search engine (GSE) pre-loaded in, e.g., Apple’s Safari (and hence on iPhones) and in Firefox. These were deemed to be de facto exclusive-dealing contracts and, although they did not bar the use of other search engines, were found to be unlawful monopolization in violation of Section 2 of the Sherman Antitrust Act.
Given that the DOJ and state plaintiffs prevailed (on two counts, not on all), one might expect them to ask the court to void such agreements and restrict their use going forward. And one understands why the DOJ—in the role of proposing, not stipulating, remedies—might ask for something more than it thinks necessary to remedy the harm it thinks has been done. Ask for x; settle for x-y.
But the proposed remedy goes beyond the beyond, and then perhaps ‘round the bend, prompting economists Tom Lenard and Scott Wallsten to opine in the Wall Street Journal that “Antitrust Officials Want to Sell Google for Parts.” I’m not going to cover everything in the DOJ’s proposal, as the specific proposed prohibitions, obligations, and divestitures run for about 26 pages.
Data-Sharing Obligations
Under the DOJ’s proposal, Google would have to share their search data, ad data, and user data—some at marginal cost and much of it for free, never mind the cost of gathering it, processing it, or sharing it. And they’d have to do so on a nondiscriminatory basis, all while providing for privacy and security and imposing no restrictions on the downstream use of that data.
Google would also be required to divest certain properties and intellectual property (the Chrome browser, a good amount relating to AI, and perhaps the Android operating system, among other things) and would be subject in their internal operations to quite a few restrictions on the types of agreements they might enter into going forward—including, but not nearly limited to, the default placement agreements that were at-issue in the trial. As just one example, Google would be obligated to share its search index: a database of web sites that Google might include among search results. With respect to the search index, DOJ proposes that:
- Google must provide, at marginal cost, ongoing access to its Search Index to Qualified Competitors such that it is equally available to Qualified Competitors and Google.
- Google must make available, through the Search Index, all content from any Google-owned website, property, or other operated platform (e.g., all Google owned or operated properties such as YouTube) which Google uses in its own Search Index.
- Google must provide the Search Index with latency and reliability functionally equivalent to how Google is able to access its Search Index.
That alone is a great deal of data to be made available to present and future (actual and would-be) competitors on an ongoing basis. Moreover, Google must engineer the means to make this data available with the same latency and reliability it can access the data itself internally. And it must do so at marginal cost, never mind the fixed costs (sunk and otherwise) and never mind the effect on incentives to develop such resources going forward.
The question of how to determine marginal cost disappears when it comes to DOJ’s proposal that Google be obligated to share “user-side data,” because DOJ proposes that access to such data be provided “at no cost” and “on a non-discriminatory basis” (in other words, for free) and that it must do so “while safeguarding personal privacy and security.” That last bit shouldn’t be a problem, because the DOJ assured the court that:
Any User-side Data that Google collects and uses as part of any of its products consistent with this Final Judgement can presumptively be shared with Qualified Competitors consistent with personal privacy and security, as Google is prohibited from using and retaining data to which access cannot be provided to Competitors on the basis of privacy or security concerns.
Or it might be a problem, in that Google does not sell personally identifiable information (pii) to third parties, and only transfers it under certain circumstances, and subject to limitations. That may be imperfect as a privacy and security assurance—in part, because such assurances are inevitably imperfect. See, for example, the 2017 Equifax data breach “that exposed the personal information of 147 million people” or, for a personal unfavorite of mine, the 2015 breach of approximately 22 million records at the U.S. government’s Office of Personnel Management.
But that’s not at all to say that consumers shouldn’t care about what goes where, and how it may be used (Liad Wagman and I reviewed many of the tradeoffs here). This part of the proposal includes both obligations to share data (for free) and new restrictions on Google’s collection and internal use of data. But not forever, mind you—just for 10 years. What could possibly go wrong?
If you’re wondering what they mean by “user-side data,” it’s rather a great deal, including things that don’t look much like user data. As the proposal defines it:
“User-side Data” means all data that can be obtained from users in the United
States, directly through a search engine’s interaction with the user’s Device, including software running on that Device, by automated means. User-side Data includes information Google collects when answering commercial, tail, and local queries. User-side Data may also include data sets used to train or fine-tune Google’s ranking and retrieval components, as well as artificial intelligence models used for Google’s AI Product.
Google would also be required to provide access to its ads data, also “at no cost, with access to all Ads Data on a non-discriminatory basis while safeguarding personal privacy and security.”
That’s not the whole of it, but it should give readers some sense of the scale and scope of the requested data-sharing obligations.
Bye-Bye, AI
Google would be obligated to promptly notify the plaintiffs (the United States and various plaintiff states—roughly, all the states plus several U.S. territories) of:
…any investment, holding, or interest in any Competitor, any company that controls a Search Access Point or an AI Product, or in any technologies, such as AI Products, that are potential entrants into the GSE or Search Text Ads markets or reasonably anticipated competitive threats to GSEs. Within six (6) months, Google must divest any such interest and immediately refrain from taking any action that could discourage or disincentivize that company from developing products or services that compete with, disrupt, or disintermediate Google’s GSE or Search Text Ads.
Moreover, within six months, Google would have to divest any such AI interests and investments “and immediately refrain from taking any action that could discourage or disincentivize that company from developing products or services that compete with, disrupt, or disintermediate Google’s GSE or Search Text Ads.”
But wait, there’s more. Without the DOJ’s prior written consent, Google would be unable to:
…acquire any interest in, or part of, any company; enter into a new joint venture, partnership, or collaboration, including any marketing or sales agreement; or expand the scope of an existing joint venture, partnership, or collaboration, with any company that competes with Google in the GSE or Search Text Ads markets or any company that controls a Search Access Point or query-based AI Product.
Breaking Up Is Hard to Do
Google would be required to divest its Chrome browser, and it might have the option of divesting its Android operating system in lieu of certain other proposed requirements regarding Android. To be precise, that would initially be an option, perhaps to later be revisited as a mandate.
Google would also be prohibited from a range of conduct deemed to be self-preferencing, and from entering into various types of contracts well beyond the default agreements at-issue in the matter.
Various restrictions on contracting are interesting, not least because they raise the question of how browsers and search engines might be monetized. Yale University economist Fiona Scott Morton suggests that Chrome, once out of Google’s hands:
provide strong evidence that resourcing is unlikely to be a concern. Mozilla spends about $450M annually to operate the Firefox browser. If we assume that an independent Chrome for desktop would cost approximately that much to operate, this amounts to about 2% of Bing’s current search ad revenue. Given how few resources are needed to operate a competitive browser relative to the total search advertising market, an independent Chrome interested in monetizing its browser through search revenue sharing seems like it would have options available to do so.
Maybe, but how competitive is Firefox? It trails well behind Chrome, Safari, and Edge, enjoying about a 5% market share of U.S. browser usage. Mozilla gets most of its revenue from agreements with search engines, with Fortune reporting that more than 85% of that comes from (wait for it) Google. But Google would be prohibited from entering into such agreements with a standalone Chrome and, apparently, Mozilla, among others.
So maybe Chrome would not be viable, after all. And maybe Mozilla’s Firefox would circle the drain close behind it. That might be fine for Apple’s Safari and Microsoft’s Edge, but it’s not at all clear how it could be good for browser competition or consumers.
Are we getting bleary-eyed yet? That’s not the whole of it, but it’s a lot. Maybe I should have stuck with the Lenard and Wallsten gloss that antitrust officials want to sell Google for parts.
The U.S. antitrust agencies tend to prefer structural remedies to behavioral ones, not least because of the oversight required with many behavioral remedies. Here, we start to see the complexity of the DOJ’s proposed behavioral remedies, most of which are to run for 10 years, if they are not extended.
On the other hand, the structural remedies proposed are not so simple themselves, and the likelihood that they’ll benefit consumers and competition might charitably be described as “unknown.” Certainly, the DOJ has no idea.
What Next?
How much of the DOJ’s proposed remedy might fly with Judge Mehta, I cannot say. On the other hand, there is the question of how DOJ could show that all of this is necessary to remediate the alleged harms in the case, given the court’s finding that:
Google has not achieved market dominance by happenstance. It has hired thousands of highly skilled engineers, innovated consistently, and made shrewd business decisions. The result is the industry’s highest quality search engine, which has earned Google the trust of hundreds of millions of daily users.
And Google charges those “hundreds of millions of daily users” (at least) zero dollars to use that search engine, which it monetizes via various agreements and advertising revenue. That is, by revenue that would be undermined by the proposed remedy, and that’s before we consider the DOJ’s separate Google adtech case.
Long before Judge Mehta’s August decision on liability, Herbert Hovenkamp suggested that some of the proposed divestitures might be a “disaster.” He’s more recently been quoted saying “[t]argeting big tech as the Biden administration did was bad politics, but it was also bad antitrust.” And that “the association between ‘corp monopoly’ and ‘big tech,’ even though the vast majority of Americans reap benefits from it that exceed losses by many times,” is just so much “rhetorical clutter.”
Of course, the tremendous benefits conferred by Google’s zero-price GSE does not mean that no consumer tradeoffs are implicated. And it’s not to say that Google or any other productive firm gets a pass on the antitrust laws—or the tax code, for that matter.
I disagree with key parts of the decision, but it is, in fact, the decision, and it stands unless or until it is reversed on appeal. The DOJ was bound to propose something. And perhaps it was unlikely to propose that the court simply enjoin the default agreements at-issue. In part, they were probably unlikely to do so because the imposition of choice screens in Europe (if only on Android devices) did nothing to change Google’s share of search there.
If a narrowly tailored remedy won’t change anything (but perhaps Apple’s revenue), maybe it was wrong to bring the case in the first place. Plainly, the DOJ doesn’t think so. But why the obvious overreach on proposed remedies? I understand the rationale behind asking for x and settling for x-y. And I understand the role of the home-run hitter. But when the home-run hitter succeeds, it’s an unalloyed good for the batter, his team, and the team’s fans. It’s not at all clear who would win if the DOJ gets its way here.
If the strategy is to “throw everything against the wall and see what sticks,” one has to live with what sticks. And what sticks might well be a mess, rather than an unalloyed good.
My Own Modest Proposal
To pilfer from Jonathan Swift, I have a far more modest proposal to revive GSE competition. There’s a straightforward way to make Bing and other GSEs more competitive with Google Search.
On the one hand, I have little confidence in the ability of DOJ’s lawyers to identify, with any specificity, the data that would make competing GSEs true rivals to Google. And I have considerably less confidence in their ability to provide rivals with the code or models that would do it.
On the other hand, I become more and more convinced that they could increase the relative competitiveness of Google’s rivals by ruining Google–I think that this is what Cory Doctorow means by “enshittification.”
One easy way to do that would require the cooperation of Microsoft. If all attempts to use Google simply linked to searches actually carried out by Bing, then Bing would be every bit as good as Google—which would, in fact, be Bing.
It might take consumers a while to realize that, but I have faith that they’d get there. Whether they’d like it is, of course, a different question entirely.