In August, Judge Mehta of the Federal District Court in Washington, D.C., concluded in a careful and detailed opinion that Google had a monopoly in both the internet search market and the associated text advertising market. Google was found to have abused its market power by engaging in exclusionary conduct, including paying large sums of money to equipment makers, browser operators, and cell phone systems to retain this dominance. The opinion declared that while Google got its monopoly because of its “skill, industry, and foresight,” it then used unlawful tactics to entrench and reinforce that position. The decision also recognized the enormous cost of creating and maintaining an effective search engine, as well as a suggestion that the text advertising system involved substantial costs. Given the apparent durability of both these monopolies, the question that the Court now faces is finding an effective remedy.
This week, the Department of Justice (DOJ) is expected to file proposed remedies for this abuse of monopoly power. Several voices have weighed in on remedy design, including The Economist, in a leader titled “Dismantling Google is a terrible idea.” Divestiture of Chrome or Android should be avoided at all costs, argued the magazine, even if that means embracing behavioral remedies such as “limiting its ability to use its search engine to distribute its AI products,” or “mak[ing] public some of the technology that enables its search engine to work, such as its index of web pages and search-query logs.”
In two prior posts, we spelled out two alternative ways to remedy bottleneck monopolies. These monopolies are ones that connect otherwise competitive markets but for a variety of reasons are durable and unavoidable. Obvious examples include electric transmission systems, cell phone service, and natural gas pipelines. The internet world is also subject to a number of bottlenecks.
The Search and Text-Advertising Engines Are Bottleneck Monopolies
Google’s search engine stands between the great mass of users with questions and the entire internet’s resources. Its search engine functions to identify and classify potential responses to the question. The cost of creating the Google search engine was over $20 billion and it requires many billions annually to maintain and expand it. Only two other search engines exist, and one recent effort failed after massive investment. Of the survivors, Microsoft’s Bing has a 10 percent market share overall and the other, Yahoo, has less than 3 percent. Hence, neither is a significant competitor. Browser providers need to have one or more search engines easily accessible for users, and they can’t charge searchers for their searches.
Because the search engine is costly to create and maintain, the question is how to pay for this service. The text-advertising engine is the means for paying for all searches. Text advertisements are the textual lines appearing at the top of any search that implicates a good or service. The line links a searcher to the website of the advertiser that hopes to make a sale.
Google sells such access to advertisers as does Microsoft and Yahoo. Judge Metha found that the creation and maintenance of the text-advertising engine is also very substantial. But at the same time, the other search engines appear to have their own text-advertising engines. This at least suggests that such engines are more readily producible, and that Google’s dominance comes primarily from its control over the search engine. Both browser operators and advertisers agree that they have to use Google’s search engine. They accept the text search engine because that is the means by which Google is compensated. Google also shares that revenue with the browser operators.
The DOJ’s Theory of Harm and Implicit Remedy
The litigation seems to have focused primarily on the anticompetitive effects of the various exclusive dealing contracts that Google obtained to ensure the dominance of its search engine. These contracts involved multi-billion-dollar payments to cell phone makers (like Apple) and browsers (like Mozilla) to ensure Google search was the default option. Nominally, other options could be provided and were included in some browsers, but the effects of Google’s brand recognition and its placement in cell phone and computer browsers resulted in effective retention of a monopoly market share.
While the opinion focuses on the harms resulting from the exclusionary practices of Google, the underlying factual findings suggest that regardless of the specifics of the contracts at this point and for the foreseeable future, the Google search engine will retain its monopoly position. Removing the exclusionary terms from the contracts is unlikely to result in any significant change in the structure of the search engine market.
Perhaps the government belatedly recognized this situation and so tried to shift the focus of its case from an attack on the specific exclusionary effect of the contracts to a broader claim of monopolization. Judge Metha rejected that move because it came late in the litigation. This is somewhat similar to the government’s failure to think through its case against Microsoft in the late 1990s, which started as a challenge to the tying of the operating system to its browser but ultimately morphed into a broader challenge to Microsoft’s monopoly. The failure of the government in that case to have a remedy that would effectively address the monopoly bottleneck of the operating system explains why 25 years later, Window’s still has a monopoly share of computer operating systems and their applications.
The fundamental challenge is to find an effective remedy that will eliminate or significantly reduce the incentive to exploit the bottleneck monopoly and use it to exclude competition in the upstream or downstream markets that the monopoly serves. Eliminating at this point exclusionary contracts given the extraordinary costs of building and operating a search engine that would, at its best, basically duplicate the Google engine, is unlikely to affect the search market in the foreseeable future.
Consider the Alternatives
The central thesis of our prior posts was that where there was an unavoidable bottleneck monopoly, an effective remedy is to change the ownership of that monopoly in a way that would eliminate or greatly attenuate the incentives to exclude and exploit. We also recognized that the first best option would be to break up the monopoly. But in many cases, this is not a feasible option. We suggested that there were two other ways to reallocate ownership and control of the bottleneck. One way is to create a “condominium” that collectively owned the bottleneck, but each user had its own piece to use. The alternative is to move ownership to a “cooperative,” which would both own and operate the bottleneck. While a divestiture remedy is possible, we think that the more likely option is to have either a condo or cooperative own and operate the search engine. As suggested earlier, our assumption here is that the text engine is one that has little exclusionary power on its own and will be further weakened if the current case, in trial, concerning Google’s monopolization of the “ad stack” (e.g., ad server, ad network, and ad exchange) results in dissolution of that monopoly.
Option A: Divestiture
The historic response to monopoly expressly declared in the Standard Oil and American Tobacco cases is to break up the monopoly into separate competing firms. It is hard to imagine how a search engine could be subdivided, but it is possible to imagine that multiple entities could receive the right to use the existing engine, “hiving off” the search engine. Each might then have the right to undertake further development of the search engine. There probably would have to be some significant compensation to Google given its massive investment to date in the search engine. This would limit the number of browser or cell phone operators that could even consider a license.
A second concern would be brand loyalty. Would searchers, assuming that Google was allowed to retain its own version of the search engine, be willing to use alternatives in sufficient quantity that the result would be economically attractive? To recover the costs and make money, text advertising needs to be attractive to advertisers.
Finally, the engine itself needs continued work. This means that those entities that took the engine would need to develop additional capacity to perform those tasks, which is unlikely to be easy or inexpensive. This suggests that divested versions of the Google search engine would struggle to compete in the market.
It would also appear that if the text-advertising engine is currently a bottleneck monopoly in its own right, a licensing system for its use with the further right of each user to amend and improve its version would probably resolve this part of the monopoly. There is no brand loyalty for such engines. Moreover, as noted above, the pending Google ad tech monopoly case is likely to result in a further increase in competition in that area of technology.
Option B: A Condominium Solution
If the search engine is sufficiently distinct from the operation of browser and cell phone operating systems, then one remedy would be to transfer ownership of search engine to an entity owned by the various users, but with the on-going maintenance performed by a separate entity that contracts to provide this service to the owners. This is analogous to a condominium association contracting with a management company. Each owner of a condominium would pay for the managerial services and would be able to use the search engine.
The manager would have significant capacity, however, to exploit this system. If compensation were on a cost-plus basis, that might reduce the risk. An even more open system in which the manager’s task is only to review and implement proposed improvements developed by third parties might reduce further the risk of exploitation. The puzzle then would be how to compensate third parties for developments.
Overall, we are skeptical that a condominium-type structure would be a very effective solution to the monopoly bottleneck that the Google search engine presents.
Option C: Cooperative Solution
A cooperative type of organization would own the search engine itself and share the ongoing costs of its operation, based on usage by the participating enterprise. The cooperative would in turn either have its own staff to maintain the engine or it would contract with various third parties to supply necessary inputs. Each participant would be able to use the search engine as it saw fit and match it with whatever text-advertising system was most attractive given the customer base and technology of that entity.
A cooperative solution to the search engine monopoly is a much more promising solution than the options of injunction, divestiture, or condominium ownership. But we see two real risks and problems. First, there is a question of the incentives to innovate especially where some users would be advantaged over others. The risk is that if most distributors of the search engine are using the same vehicle, they may have a hard time supporting innovations that might favor some types of users, e.g., cell phones, over others, e.g., computers.
Second, as Judge Metha observed, and as The Economist points out, there is a possibility that AI may eventually make search engines obsolete or offer a very credible and open alternative. The judge concluded, however, that this potential was only that. There is no current or immediately foreseeable AI search system. The concern would be that if most search providers are participating in a cooperative that provides a search engine, they may have a collective disincentive to support or sponsor the potentially costly and time-consuming effort to develop an AI search system.
Conclusion
Finding an effective remedy for the monopoly created by Google search and text-advertising engines is a major challenge. Our concern is that the government has too narrowly focused on Google’s exclusionary contracts. Removing those contracts at this late date is unlikely to produce any significant change in the monopolization of these markets and potential for ongoing exploitation and exclusion. It is regrettable that the government did not initiate its case with an explicit focus on a remedy or remedies that could actually affect the future structure and conduct in this market. We have here examined three options that could dissipate the underlying monopoly power. Each has risks and problems, but each is a better alternative than a simplistic elimination of exclusionary contracts.