My previous posts have focused specifically on interoperability — what that means in the digital platform ecosystem, and why it’s relevant to the goal of promoting a competitive market and the economic benefits that derive from one for businesses and individuals. In this post, I’m going to shift focus a bit towards the backdrop behind all this: the state of American antitrust law. Changes in how competition approaches tech seem inevitable today, for a number of reasons. Whether that shift will recognize the importance of and promote interoperability remains an open question, and the future of the internet as we know and love it may depend on the answer.
I’m not going to pretend to summarize American antitrust law here. (You could try the Federal Trade Commission’s thorough “Guide to Antitrust Laws” for that.) Colloquially, antitrust authorities have two major paths for intervention in business practices that are applicable in the context of digital platforms offering third-party accessible APIs for interoperability: merger review and single-firm conduct enforcement. With both, the initial burden is heavily on the government, as the Department of Justice must bring an affirmative lawsuit before a federal court to make the claim that the law would be violated by the combination. This is in strong contrast to an agency with rulemaking authority, which can set out specific guidelines as rules in interpretation of the law and thus establish clearer prima facie cases.
Antitrust law — specifically, Section 1 of the Sherman Act — states that any merger “in restraint of trade or commerce … is declared to be illegal.” In the United States, this authority is most commonly used to prohibit horizontal mergers in highly concentrated markets — businesses that offer substitutable products, competing in the same market for the same customers. And where the result of the merger would be a market that is still effectively competitive, the merger is generally allowed, as efficiency gains are considered to outweigh the reduced competition. With vertical mergers, such as Facebook and WhatsApp (the subject of my first post), the efficiency gains are more compelling and the competitive losses can be less direct and harder to measure through traditional tools, making it harder to block the combination. With either type of merger, the government can settle the issue with the merging companies by agreeing on structural and/or behavioral conditions to be undertaken by the companies, such as divesting certain overlapping assets or agreeing to maintain or refrain from certain business practices for a time.
The second path for intervention is to determine that a single company’s practices constitute unlawful behavior, specifically monopolizing trade. To violate this part of antitrust law, Section 2 of the Sherman Act, the company must be dominant within some cognizable market, and must engage in behavior that unfairly maintains that dominance or leverages it into an unfair advantage in another market. Over the past few decades, the Department of Justice has used this authority sparingly. The DOJ adopted Section 2 guidelines in a lengthy report in September 2008 designed to further reduce such interventions, and although that document was revoked early in the subsequent administration, it was never replaced, and the historical hesitancy seems to continue today.
The last major technology case brought by the DOJ was the Microsoft case, just over 20 years ago. That case had a transformative impact on the internet ecosystem, and was instrumental for the success of the Firefox web browser (and thus my employer Mozilla) and ultimately for the creation and flourishing of entire industries.
By internet standards, 1998 was an aeon ago. American antitrust seems to be stuck. Decades of court cases and precedent, following the legal work of Robert Bork and the economic theories referred to as the Chicago School, have made it significantly hard to prosecute antitrust claims against single-firm conduct and vertical mergers. Other legal systems have not made this shift. The result is more room for the European Commission and the Bundeskartellamt in Germany to take an aggressive position on growing centralization and individual corporate practices, including in particular in the tech sector.
I’ve been getting to know antitrust and competition attorneys much better over the past couple years. Along the way, I picked up the saying, “Antitrust theories never really die — they just go out of style.” Fortunately, some of the pre-Chicago school theories seem to be coming back in vogue.
On both single firm conduct and merger review, antitrust in the U.S. is experiencing a bit of a renaissance right now. It’s coming from many different parts of the sector, including advocacy organizations, most notably Citizens Against Monopoly; think tanks like the Center for American Progress; and academics from multiple perspectives like Carl Shapiro and Hal Singer. This renaissance has also appeared in major national media, with Farhad Manjoo of the New York Times an early leader. More recently, calls for intervention have emerged from some of the highest-profile political layers. Change of some form seems inevitable.
I’m hopeful the changes that come out of this maelstrom will be constructive and lasting ones, meaningful recognition of gaps in the operating economic theory behind American antitrust (such as Hal Singer’s well-articulated innovation gap) that can be closed with effective and efficient solutions.
The most recent antitrust decision in the U.S. doesn’t quite reflect this evolving legal and political climate. The Supreme Court case Ohio v. American Express raised the bar for antitrust enforcement in two-sided markets, making the necessary economic showings more complex and difficult under some circumstances. It’s not clear to me how this case will play out in other contexts, and I think perhaps immediate reactions that it will raise the bar for antitrust across the board including in tech may not play out as predicted. But it does seem to constitute yet another hurdle, whether large or small, for competition authorities in the United States.
To me, Ohio v. American Express doesn’t reach the really interesting part of centralization and competitive harm in the context of tech and tech platforms. What’s fundamentally different about tech isn’t the size of the business nor the 2-sided nature of the market; those both have equivalents offline, and we have jurisprudence to know how to evaluate them.
The unique feature of digital platforms in the context of competition is the nature of the vertical integration of distinct and interconnecting digital services, the technical ability to control that interconnectivity through product and business decisions around integrating code bases and offering APIs — and how those decisions can run counter to long-standing assumptions of interoperability and openness on which the internet was built. Even before we started calling everything a platform, that was how tech was designed: to interface with and operate alongside software and services built by other businesses, relying on settled norms and the mutual benefits of interoperability. Unfortunately, those norms are no longer settled, nor the mutual benefits guaranteed, in the digital economy prisoner’s dilemma we have today.
This is also one more reason why classic competition thinking doesn’t quite fit for the tech sector. Competition problems aren’t measured accurately by looking at market share. I’ll go even one step further: Being big isn’t inherently harmful if a dominant digital platform is truly open to its vertical and horizontal competitors, because they can keep it on its toes. Possession of data isn’t always nine-tenths of the law because of the non-rivalrous nature of data and the possibility of offering third-party APIs to provide meaningful access and empowerment to others.
At the same time, it’s not sufficient to create a low-concentration market to preserve the internet’s competitive characteristics in the way they work today. Imagine that, instead of the internet experience we had today, we had five silos of technology stacks, fully vertically integrated with no interoperability across them. If the choice we face as consumers was one of a few entire single-firm, homogenous internet experiences rather than the heterogenous experiences we can choose today, what would antitrust law think? The only cognizable “market” would be labeled as something like “internet services”, and we could have five companies at roughly 20% market share each (give or take). The HHI measure for that market would be 2000 — far short of the threshold for significantly concentrated.
Perhaps antitrust lawyers and economists could overcome that limitation and find a way to measure the lost economic productivity/innovation of new entrants who would be stifled if unable to build on top of the dominant digital platforms. If successful, this might be enough to support an antitrust intervention to open up the integrated stacks. But I don’t believe we have the right legal and economic tools for that today. And even if we develop them, it’s much (much!) harder to build interoperability back into a system that has been built closed than to guide it to be open as it goes forward.
A technology ecosystem consisting entirely of vertical stacks that don’t interoperate might measure as a functional market. But it wouldn’t be the internet. It wouldn’t be the disruptive, generative, creative, world-shaping force it is today.
Effective interoperability is a key piece of what makes the internet unique and amazing, and a necessary condition for the radical competition, incomparable innovation, and stratospheric investment we associate with the internet economy. Although it’s not an obvious takeaway from antitrust law as it stands today, competition authorities who understand their mandate as making markets work to their fullest ought to make the promotion of interoperability a top priority.