The U.S. Justice Department (DOJ) won its antitrust case against Google last year, establishing that the company illegally maintained its monopoly in “general search” and “general search text advertising” markets through exclusive default contracts. Now comes the hard part: crafting effective remedies.
I’m on record as saying the question of remedies would be difficult in this case, but new research from the National Bureau of Economic Research (NBER) suggests it may be even harder than I originally thought. In fact, the findings from researchers Hunt Allcott, Juan Camilo Castillo, Matthew Gentzkow, Leon Musolff, and Tobias Salze challenge many of our assumptions about competition in search markets and raise serious questions about whether any of the proposed antitrust remedies would actually help consumers.
The researchers conducted a randomized controlled trial (RCT) to examine directly how users actually interact with search engines. To do this, they recruited 2,354 desktop internet users and had them install a special browser extension that tracked their search behavior, thereby moving beyond surveys to observe real-world choices.
The study created four groups:
- a control group who continued normal search behavior;
- an “active choice” group, who were forced to actively pick their default search engine;
- a “default change” group, who were offered $10 to switch to Bing for two days;
- and a “switch bonus” group, who were offered varying payments ($1, $10, or $25) to use Bing for two weeks.
This design allowed the researchers to measure three key factors that might explain Google’s dominance: switching costs (how hard it is to change search engines); user inattention (not bothering to consider alternatives); and quality perceptions (whether users accurately assess Bing’s capabilities).
Their findings challenge several core arguments for aggressive antitrust remedies. High switching costs aren’t the barrier that many have assumed. When researchers implemented active choice between search engines (similar to the choice screens mandated in the EU), Bing’s market share increased by just 1.1 percentage points. The real barrier appears to be simple inattention: 34% of users are “inattentively loyal,” never even considering alternatives.
The findings on scale and data advantages are even more important, since choice screens aren’t on the table for iPhones. A central claim in the DOJ’s case was that Google’s exclusive contracts prevent competitors from reaching efficient scale. Yet when researchers simulated giving Bing access to ALL of Google’s search data, quality improved by less than 1.3 percentage points in clickthrough rates. This minimal quality improvement had virtually no effect on market share.
This creates a genuine dilemma for Judge Amit Mehta of the U.S. District Court for the District of Columbia, who is charged with crafting remedies in the case. The most aggressive option—forcing Bing as the default search engine on Android devices—would reduce Google’s share of the search market by 40 percentage points. But it would also slash consumer surplus by $70 per-user annually, an enormous welfare loss that’s difficult to justify.
The paper did find one intriguing bright spot for possible remedies: when users actually try Bing, 33% stick with it. This suggests the quality gap between search engines may be smaller than is commonly assumed. But this insight doesn’t easily translate into a workable remedy, since courts can’t realistically force users to try alternative search engines.
An even deeper challenge was revealed through an experiment in which the researchers deliberately degraded the quality of Bing’s search rankings by reversing the order of top results. Surprisingly, this had no effect on market share. Users didn’t seem to notice or care about these quality differences.
This finding has two troubling implications. First, even improving competitors’ quality through interventions like data sharing (a proposed remedy) may not meaningfully increase competition. The link between search quality and market share appears weaker than previously thought. That really throws off a lot of analysis on both sides.
The findings also raise fundamental questions about the competitive process itself. If users don’t actively choose based on quality differences, what exactly is the market failure we’re trying to fix? The DOJ argued that Google’s contracts prevented a “but-for” world where competition would drive quality improvements. But this research suggests competitive forces in that alternative world might be quite weak.
The study points to a stark reality about feasible remedies. The most legally and practically feasible option would be an injunction prohibiting Google from paying for default status on browsers and devices. While this wouldn’t dramatically shift market share, given users’ demonstrated inertia, Mozilla (and Apple) would lose out on the payments from Google.
This leaves Judge Mehta in an uncomfortable position. Liability has been established and the court must do something. But the research suggests we may need to accept more modest goals focused on removing specific anticompetitive practices, rather than dramatically reshaping market shares.
Perhaps this case will force us to grapple with an uncomfortable question: Should conduct be ruled illegal if we can’t identify a realistic way to improve the situation?
