Despite growing concern about the power of Big Tech, regulators have struggled to hold these companies and their executives to account. Here’s why, and what we can do about it.
On Tuesday, US regulators announced a major new investigation of Google, accusing it of illegal monopolization. This is the first major US investigation of a technology company since the case against Microsoft over 20 years ago. Since then, regulators in both the US and EU have largely failed in regulatory lawsuits against technology companies, or been able to impose only minimal fines for misconduct.
In new research published this week, I argue that the power of these companies, and the difficulty governments face in holding them to account, is the result of a phenomenon I call “Janus Faces.” These companies are so many things at once—so many different types of company rolled together—that they can slip through the cracks in our regulatory system, which depends on being able to clearly define what market a company is in and who its customers are.
Here’s an example: In 2007-2008, Google acquired an advertising company called DoubleClick. The merger review centered on whether search-based advertising, which Google already dominated, and display advertising, in which DoubleClick was the leading player, were part of the same online advertising market, in which case Google would be buying its way into monopoly (a reason to bar the merger), or distinct sectors with the merger able to go through.
Google successfully persuaded the panel of the latter, by promising in part that it had no intention of merging the user datasets on which the two advertising models relied. This was a non-binding commitment, and the datasets were merged the following year. That is how our Google searches came to inform the display ads we see on other, non-Google websites. By that time, it was too late for regulators to intervene. Presenting search and display advertising as separate markets made Google and DoubleClick’s merger appear less of a source of market consolidation than it ultimately turned out to be.
That’s not just a matter of new economy companies upending old economy distinctions. I find that technology companies deliberately cultivate ways to straddle such categories, in order to make themselves more difficult for regulators, the media and the public to pin down.
What can be done about it? In that 2007-2008 merger, Google’s argument was accepted by four of the five members of the panel that decided the case. But the dissenting opinion, from Federal Trade Commissioner Pamela Jones Harbour, is telling. Harbour argued that rather than attempt to draw boundaries between online markets with ever-shifting definitions, regulators should treat Google, and all companies like it whose revenues derived from monetizing user data by selling it to advertisers, as companies in a new data market.
Harbour correctly predicted that it was DoubleClick’s data that Google sought to acquire and that the datasets would ultimately be combined. In the future, she argued, any mergers that would contribute to a company acquiring or consolidating monopoly in user data ought to be either prohibited, or require companies to enter into legally binding commitments about how each merger partner’s datasets would be used.
Twelve years on, this remains a great idea, not least because it solves the conundrum of what existing market platform companies are in by classifying them as a new type of data company. Whenever any companies whose business model involves monetizing user data want to merge or acquire smaller companies, regulators in the country where the merger has to be approved could require the parties to sign and submit a binding data plan as part of merger review. If the plan shows that the merger would not give one company a monopoly in data, the merger can go ahead. Later, if the merged company wants to make changes to how it uses the data, it would have to seek new regulatory approval, and companies could be made to divest from certain markets if their proposed plans amounted to monopolization. Many of the mergers that have helped turn competitive markets into ones dominated by a single player—Facebook’s purchases of rival startups WhatsApp and Instagram come to mind—would be unlikely to clear such a threshold, or to be worth it to companies to pursue if they could not harvest the data from their targets anyway.
Reframing these companies as data companies would also allow us to revive solutions we have adopted in the past. Data is the raw material of our time, which flows through and powers our economy just like oil, gas and water have done. The companies that supply access to these older natural resources are treated as utilities, whose status as essential infrastructure means that they are subjected to special public interest obligations. Recognizing the technology platforms as data utilities might justify a public interest ban on companies operating simultaneously as players in their own marketplaces. This approach is known as “structural separation.” For example, India has just introduced legislation that bans retail platforms from selling products on their own websites. This prevents Amazon from using its data on what users like to buy to sell own-brand products that out-compete other retailers.
In addition, framing the platforms as data companies brings privacy concerns about their power together with the concerns about monopolization. Since competition law has more enforcement ability than privacy law, this can help make these privacy issues easier to address. It also fixes a growing problem with competition law, which has become increasingly focused on maintaining narrow economic objectives like low prices, instead of on achieving public interest objectives like fairness.
That was not always the case. In the early 20th century, when the first competition laws were enacted to regulate monopolies in the oil and railroad sectors, protecting the market was only one justification. The other was protecting democracy from the influence of such powerful corporate actors. “The Constitution,” argued the anti-monopolist U.S. President Theodore Roosevelt, “does not give the right of suffrage to any corporation…There can be no effective control of corporations while their political activity remains.” At a time when Facebook and Twitter are facing questions about their role in elections, and the US Supreme Court is expanding the scope for corporate money in political life, this public interest approach to monopolies is particularly relevant.
Finally, treating the platform companies as data utilities allows the law to act on behalf of all of us who use technology platforms, who aren’t “customers” in the strict sense, because we don’t pay to use these websites. That is the logic behind the EU’s General Data Protection Regulation, which has allowed the EU to take action on behalf of European citizens against American corporations over whom they otherwise have limited jurisdiction. By recognizing that citizens and their data constitute the raw material of platform power, data regulation has the ability to challenge that power itself.