Last week, President Biden signed a landmark executive order that requires the administrative agencies of the federal government to promote fair competition in the marketplace. The order arrives against the backdrop of the greatest resurgence in antitrust enforcement in almost a generation, most of it focused on Big Tech. Leveraging the House antitrust subcommittee’s landmark 450-page report released in October, federal and state antitrust enforcers initiated a blockbuster case against Google and Congress introduced five bipartisan bills aimed squarely at Google, Facebook, Apple, and Amazon.
While federal administrative agencies carry out Biden’s order and Congress puts its antitrust legislation through the lengthy review process, federal antitrust enforcers like the Department of Justice and the FTC should consider a speedier remedy that hasn’t been used much lately: compulsory licensing.
Compulsory licensing requires a firm to share certain intellectual property with all interested parties. History shows that compulsory licensing helps other remedies, such as constraints on which markets a corporation may enter, creating a vibrant marketplace with significant new competitors. Google would be an especially strong candidate for compulsory licensing.
Do you even have to ask?
Google leverages its near-total dominance of the search-engine market to extract highly favorable terms from companies dependent on its services. For example, Google requires phone manufacturers that want to use its Android smartphone operating system to make its search engine the default provider for mobile searches. Android thus allows Google to extend its search dominance from desktops and laptops to mobile phones all over the world.
Google also uses its search engine to favor its own services. According to the House antitrust subcommittee report, Google services cover nearly 100 percent of the screen for mobile searches and approximately 25 percent for desktop searches. In 2007, then-Google executive Marissa Mayer admitted that Google intentionally ranked its own services ahead of other sites.
Google’s digital advertising fortifies its search monopoly because the data it collects about users refines search results. This process strongly incentivizes users to stay on its search platform, allowing Google to continue collecting data and refine its service further. The same feedback loop exists with Google’s YouTube video algorithm.
At the center of Google’s search engine is its index.
An index is what allows a search engine to answer a user’s search query. Search indexes are compiled by web crawlers that scan websites and record a webpage’s contents to a single master ledger. The search engine takes a search query and pulls the relevant content from the ledger. Precisely how Google determines how to rank the sites shown depends on, by some estimates, hundreds of factors, such as the number of words related to a user’s search query or the number of hyperlinks on a webpage.
In 2000, Google had indexed only 500 million pages. By December 2020, Google had indexed somewhere between 500 and 600 billion. By comparison, Microsoft’s Bing, Google’s only real competitor in the search market, has indexed between 100 and 200 billion pages. Because of internet bandwidth limitations and website owner preferences, websites limit the number of crawlers allowed to index their sites, raising the barrier for competitors to enter the search engine market.
Google’s search engines and index are bundled in a web of intellectual property protections, including patents, copyright, and trade secrets. It isn’t clear to what extent Google relies on each type of intellectual property protection for its search engine index, or to create the algorithm that determines which results to display. Google will say only that it relies on an amalgam of “trade secrets and other intellectual property.” But Google’s original search ranking system (known as “PageRank”) was called “one of the most famous and valuable of all modern software patents” by a former patent lawyer for the company.
Google’s exclusionary and predatory practices are so numerous and vast that really, the company should be broken up. But a useful interim step would be to impose compulsory licensing.
The heyday of compulsory licensing was the 1940s to the 1970s. During that period, compulsory licensing was used in more than 100 antitrust cases against companies like IBM, DuPont, and General Electric. More than 40,000 patents came to be covered by compulsory licensing requirements.
The most notable instance was a 1956 Justice Department settlement with AT&T, then the monopoly provider of U.S. telephone services and equipment. The government sought to block AT&T from using its control over telephone wires to foreclose competitors or enter the then-nascent computer industry. To that end, DOJ prohibited the corporation from providing any service other than common carriage telephone services. It also required AT&T to share its massive patent portfolio through a compulsory licensing agreement.
The result of that agreement was a wave of innovation. AT&T had known as early as the 1920s that the public would want a device to record telephone calls, and in 1934 it invented one. (The technology wasn’t all that complicated.) But it didn’t release it until 1951; in the meantime, AT&T prevented other companies from doing so by barring the attachment of ancillary devices to the telephone network.
Years later, the public would find out that AT&T also withheld the telephone answering machine (which didn’t catch on until the 1970s) for fear that it would deter customers from using their telephones. Similarly, AT&T for years refused to use its wireless patents because these threatened monopoly profits from wired telephones.
A common argument against compulsory licensing is that it reduces innovation. In fact, the opposite is true. Once smaller companies had access to AT&T’s patents, they generated revenue equivalent to $5.7 billion in today’s dollars. Gordon Moore, the founder of Intel, said the AT&T antitrust suit and its resultant compulsory licensing agreement was what allowed the U.S. semiconductor industry “to really get started.”
One notable study covering 69 companies concluded that businesses judged patent protection far less important than investing in their sales channels, creating efficient production systems, and diversifying corporate operations. The same study found that companies that signed compulsory licensing agreements experienced “little to no unfavorable impact” on research and development decisions. Other evidence has shown that when enforcers compel firms to share patents, those firms spend even more on research and development and generate upwards of 30 percent more patents.
Compulsory licensing was killed off in the 1970s by the rise of the Chicago School of economics, which reoriented the goals of antitrust to maintaining low prices. The Chicago School preached deference to corporate consolidation so long as the result created economic efficiency.
Four decades later, the Chicago school has lost much of its luster; even conservative Republicans seldom bother anymore to enunciate libertarian principles to justify policies that favor politically powerful companies. That creates an opening for policymakers to revive compulsory licensing as an antitrust enforcement tool. Legal obstacles to doing so are few, because judges already enjoy broad authority to impose antitrust remedies.
Imposing compulsory licensing on Google could bring benefits comparable to those realized from DOJ’s 1956 settlement with AT&T. As the federal government litigates against Google’s exclusionary conduct, it should consider an expansive approach that includes this largely forgotten tool. Antitrust enforcement is an arduous, complex business, but compulsory licensing is simplicity itself. All it takes is for the government to make Google open up the vaults and share a few secrets.