Searching for a Solution: Remedies for Google’s Search Monopoly

Joshua Levine is a Research Fellow at the Foundation for American Innovation.

The United States government’s case against Google is the first significant antitrust victory against a major technology company since the late 1990s, and could mark the start of a disruptive period for Big Tech. Some of the current cases against leading American technology firms began under the first Trump administration, with others launched by Biden administration enforcers, but all so far are being sustained by President Trump’s picks to head the Federal Trade Commission (FTC) and Department of Justice (DOJ).

The DOJ antitrust case against Google is in the end game: a remedies trial that will determine the consequences of DC District Court Judge Amit Mehta’s August 2024 ruling that Google’s conduct violated Section 2 of the Sherman Antitrust Act by monopolizing the market for general search. Judge Mehta singled out the firm’s use of exclusive agreements that granted Google Search default status on web browsers. While both parties have filed briefs on remedies that would address this practice, the DOJ’s proposal goes much further, calling for Google’s divestiture of Chrome, restriction from acquisitions related to “querying” or other search products, and a requirement to open its search index and data corpus to competitors. To stay within the intended goal of remedies, namely to return competition to the market for general search services, Judge Mehta should consider proposals for default agreements and data syndication.

How it started, how it’s going

Judge Mehta’s ruling outlines how Google’s monopoly on search was created through the combination of best-in-class technology and anticompetitive conduct. The ruling notes that Google’s general search product has been a market leader for nearly two decades. The story goes that Google beat out and maintained its lead against competitors such as Yahoo, Microsoft, DuckDuckGo, and others because of its superior method of indexing and ranking web pages to return the best response to an individual’s query. This, bolstered by $292.1 billion in capital expenditures since 2005, with large percentages directed toward infrastructure that serves its search and cloud businesses, turned Google into the world’s go-to answer machine.

But after reaching the technological mountaintop, Google began working to fortify its position. Specifically, Google established default agreements with various browsers and device manufacturers to put its engine in pole position throughout the web. Agreements were made with Safari (Apple), Firefox (Mozilla), Android manufacturers such as Samsung and LG, as well as on Apple devices such as iPhones and iPads. Such contracts reached tens of millions of dollars per year, with Google spending $26.3 billion in 2021 in total across all default contracts. These agreements provide a significant revenue stream to browser developers and handset makers while ensuring Google's search engine is the pre-loaded default. While Apple receives the lion’s share of these funds ($20 billion), 80 percent of Mozilla’s operating budget is covered by its default agreement with Google. On the device side, Samsung received between $1 and 1.5 billion from Google for pre-loading Chrome and the Google Play Store on its handsets.

While Google’s investment in its search product is undeniable, whether such investments translated into a better product and experience for consumers is less certain. Recent reporting and prior research have highlighted Google’s manipulation of search results to serve more ads or drive users to Google searches over organic results. Advertising was not initially part of Google’s vision for search, but has proven essential to monetize search. Intentionally degrading search results and compromising product quality would seem to be a total capitulation on one of the core principles established by founders Sergey Brin and Larry Page when they started the firm: Don’t be Evil.

Which way, Judge Mehta?

Considering the facts of the case and analysis of proposed remedies, there are three potential “paths” for remedies.

The first, let’s call it the “rubber mallet” option, is what Google itself proposes and is supported by some who disagree with Judge Mehta’s ruling. This path would impose some restrictions on Google’s contracts to be the default search engine or require the pre-loading of Google on certain browsers or devices, but have little effect on Google or competitors, outside of Mozilla.

This path would put a restriction on the exclusionary behavior, but is unlikely to restore competition in the market in the three years Google believes a remedy should last. Shifting to AI models with web search capabilities or vertical search tools, such as Amazon or Walmart, to search for consumer items, or Spotify to find information on music, are not near-term replacements for general search engines. But aside from significant shifts in consumer behavior, Google will continue as the undisputed leader for search. This would fail the main goals of remedies, such as punishing the firm for anti-competitive conduct and restoring competition in the market for general search.

The second potential path I’ll term the “sledgehammer” option is offered by the DOJ and supported by Google competitors and advocates who believe Google should be smashed. This path creates maximum pain for Google by requiring the firm to divest the Chrome browser, and potentially its Android operating system, requiring syndication of its search index, prohibiting Google from making acquisitions related to querying and search, and imposing monitoring of Google’s AI plans that could impact the competitive environment.

While this path has some positives, requiring the structural separation of Chrome and potentially Android would create significant uncertainty in markets other than the general search services market, potentially depressing consumer welfare and increasing concentration in adjacent markets. Google’s management of its Blink browser engine has created a sustained and more widely used alternative to Apple’s WebKit, outperforming Apple, as well as Gecko (Mozilla), in terms of security, speed, and customization, by some metrics.

When it comes to Android, the smartphone market is an oligopoly, and forcing divestment may promote more competition, but again, Google has ensured that developers and consumers have a choice outside of the closed ecosystem offered by Apple. While the market share remains small, new operating systems and devices are emerging, all of them forked from Android. Some will point out that this is to Google’s economic gain, that by “commoditizing the complement,” specifically by investing capital and labor to improve Chrome and Android and offering them for free, Google benefits through the firm’s control of advertising and store purchase fees. While true, this has also created more open alternatives with more diverse product offerings and, in some cases, superior features than their rivals in Cupertino. Rather than remedying anticompetitive conduct and optimizing conditions for greater market competition, this option imposes a maximum punishment on Google and overlooks the secondary effects that such a breakup would create, including destabilizing valuable projects and assuming that a new benefactor will provide an equal or better alternative. More measured behavioral or structural remedies, such as preventing Google from imposing exclusive licenses on device manufacturers or third-party stores to box out competitors, could be an alternative.

Why the claw hammer should be adopted

The third option, which I dub the “claw hammer” option, would punish Google for its anti-competitive conduct while avoiding more structural measures in order to minimize potential externalities on consumers and American technological strength. Claw hammers are a versatile tool for carpenters and are often key in restoration and building projects, whether by professionals or hobbyists. Like a claw hammer, which can both drive and remove nails, this option would deliver targeted remedies that dismantle harmful practices without tearing down the entire structure. This option would impose requirements related to data sharing and openness, specifically some form of syndication, as well as reform practices of paying to be the default browser and limiting competition for such a spot. This is the best option because it creates conditions for greater competition in the market over a longer remedy duration, is tailored to the anticompetitive conduct found by Judge Mehta, and avoids significant unintended consequences that the “sledgehammer” approach could create.

This path, which involves removing defaults and opening up Google’s data reservoir and search index, built on Google’s anticompetitive distribution, would achieve the stated goal of antitrust remedies: stopping anticompetitive conduct and creating conditions that return competition to the market. These remedies are forward-looking and would impose costs that will largely be borne by Google (rather than individual consumers), ideally fostering competition and innovation in the market.

Preventing default payments creates more opportunities for firms to compete, but enforcers should consider that ending such payments alone is unlikely to spur greater competition. The greatest cost of this new equilibrium would be to Mozilla, which relies significantly on such payments to sustain its for-profit business, including maintenance of the Gecko browser engine. Losing out on these payments would put the future development and maintenance of Gecko, the only open-source alternative to Blink (Google’s engine) and WebKit (Apple’s engine), at risk. Device manufacturers and browsers may continue to use Google as the default anyway because of its popularity among consumers, or they may require users to choose their preferred search engine when opening the browser for the first time. Initial choice screens for mobile device search in the European Union failed to induce greater competition, but that does not mean they cannot be improved iteratively. Several Google competitors have offered some principles to enhance choice screens, some of which align with some of the behavioral remedies discussed to limit anticompetitive instances of self-preferencing.

In order to properly structure a choice screen, an independent technical committee that will oversee the court’s remedies should think about how to engender competition through default search, not simply punish Google. Users rarely switch out of the default offering, and when prompted or given a choice, the familiar or best-known option wins the day. Periodic auctions for default position or priority ranking within the choice screen have been ineffective when tried in the European Union. A recent study on convincing individuals to switch between search engines emphasizes that search engines are “experience goods,” meaning users fail to adequately evaluate the relative quality of one service versus another without interacting with the service. In the study, researchers paid people to use Bing instead of Google, some participants for 2 days and others for 2 weeks. Exposure to Bing led larger fractions of users to continue to use it after being paid, including more than half of the individuals who chose to switch remaining as Bing users after 2 months. Perceived quality is important in the market for general search engines, particularly when it comes to Google versus competitors.

This leads to the second, and perhaps most effective remedy: syndication of Google’s search index and accompanying data. By making this accessible through an application programming interface (API), competitors could remotely access this corpus and leverage it to produce competing products that could realistically compete with Google. Other search engines have built their own indexes or license similar indexes from Google or other sources, but often cannot reach the scale or performance to challenge the monster from Mountain View. Such an approach would require addressing several operational hurdles, which are worth investigating.

Once remedies are finalized, the court will appoint a panel of technology professionals and outside experts to manage third-party access to Google’s internal data and systems and oversee the implementation of remedies. Retrospectives from Microsoft illustrated that the interoperability requirements ended up taking longer to implement from the company side, and a certain amount of “malicious compliance” became evident. This can be attenuated in two ways. First, the court should consider including language that would impose monetary fines on Google for non-compliance. While not traditional, such a policy has been used in the DOJ’s 2010 consent decree with Ticketmaster to ensure compliance, and creates a hedge against “malicious compliance”-like behavior. Second, the technical committee should work with Google to create a pilot of search syndication within the first 6 months of the remedies period to glean information of overall cost structure, engineering time, and performance moving forward. A pilot will create an opportunity for evaluation of syndication efforts, ensure that resources are deployed effectively, and establish fields of communication between the committee, Google, and firms interested in accessing the index.

Privacy and security considerations will be important vectors of cost and operational efficiency that the technical committee and its partners must address when opening up Google’s data reservoir and index. One way to mitigate such concerns is through the use of filtering out sensitive information using natural language processing, which has been successful in protecting sensitive healthcare data, as well as broader applications. Another idea would be to explore utilizing differential privacy, which uses mathematical methods to add “noise” into a data set to mask sensitive data, but retaining statistical patterns that enable efficient indexing. Such techniques are used within Google Cloud to limit information leakage and to remove sensitive information from data sets used for AI training. Because compute per query and engineering costs to support a differential privacy framework are likely to be high, using a pilot program focused on a portion of Google’s search index to test out feasibility would be wise. To ensure security, the technical committee should explore integrating Google Cloud's identity and access management systems to restrict permission to its index to approved actors. Another option would be to consider integrating new authentication technology that leverages advanced cryptography, such as NuID, as well as restricting access to firms based in or connected to foreign adversary nations, as defined in 15 USC 9901.

Choosing the right tool for the job

The DOJ’s victory in Google search could mark a new period of turmoil for large technology companies, but that should not be the barometer for success. For enforcement actions to resonate beyond elite policy and legal circles, there must be competitive churn in the market. Both the claw hammer and sledgehammer options would create churn, but the more versatile claw hammer is preferable. It is focused on correcting Google’s illegal conduct and returning competition to the market by sharing the benefits Google has accrued through what the court deemed monopolistic and anticompetitive behavior. While proponents of more extreme remedies would argue that their solution follows this path, ensuring that Google’s anticompetitive conduct ends, this ignores the externalities that more aggressive actions would create, likely impacting competition in adjacent markets. Such moves, while politically enticing, would cheapen and potentially limit the long-term success of this case. As we debate the appropriate punishment for Google, remedies that will restore competition and help heal the scars left by anticompetitive conduct should prevail over those that create new harms for the sake of retribution.