In 2016, hackers broke into the servers of the ride-sharing company Uber. The haul included the personal phone numbers and email addresses of fifty million Uber riders and the license numbers of some 600,000 drivers. Rather than report the breach to regulators or disclose it to the public, the company paid the hackers $100,000 to delete the data and keep the breach quiet.
Since then, we’ve seen a cascade of egregious cases of corporate misuse of Americans’ personal data, including Facebook’s Cambridge Analytica scandal and the massive Equifax records breach. As a result, official Washington is finally sending signals that it intends to do something serious about internet privacy. The Federal Trade Commission’s Republican chairman, Joseph Simons, is urging Congress to grant the FTC new enforcement powers to prosecute the worst offenders. A recent report by the Government Accountability Office warned that the U.S. lacks comprehensive legislation governing the use of online personal information by private-sector companies. In late February, the House and Senate began hearings on how to craft such a law.
The combination of Big Data and Big Business means that marketers can estimate with increasing accuracy just how much you, personally, are willing to pay, and charge you accordingly.
Yet the debate over internet privacy has so far ignored what may be the most significant privacy issue of all: price discrimination. Perhaps you missed it, for example, when in 2017, an Uber executive admitted in an interview with Bloomberg News that the company had taken its familiar “surge pricing” model to a whole new level. Under the old model, it raised prices for everyone in a certain location when local demand became strong. The new fare system, called “route-based pricing,” is essentially micro-surge: the company sets rates according to what it thinks each individual customer is willing to pay based on factors including how poor or affluent their destination is.
Uber is hardly alone in its attempt to engage in ever finer degrees of price discrimination. Marketers have always offered different prices and deals to different kinds of customers. Sometimes this is benign, as when companies give discounts to students or veterans, for example. But the combination of Big Data and Big Business is making possible something different in kind: giving some of us far worse prices and terms of services than others, based not only on our membership in different demographic groups but also on our individual characteristics, as revealed by our online activities. When it comes to the greatest abuses of our personal data, the debate should not just be about who has access to it. It also needs to be about the already acute problem of how corporations use our data to discriminate in the marketplace—not only against consumers, but also as producers and sellers—and how to keep it from getting much worse.
You probably know by now that companies use your data to try to sell you more stuff. “Behavioral advertising,” as it’s called, certainly can be creepy. I once heard of someone in the United Kingdom who, after searching the internet for information about stage 4 breast cancer, started seeing ads for funeral homes. But, at least within limits, most people don’t seem to particularly mind seeing targeted ads. Personally, if I’ve spent an evening googling the properties of bicycle snow tires, I won’t feel particularly violated when I start seeing banner ads about the virtues of, say, Schwalbe Ice Spiker Pro tubeless-ready winter bike tires, available from Amazon for $121.47. If I start seeing more listicles touting “Nine Insanely Affordable Caribbean Winter Vacations,” I can live with that, too, even if the algorithm is mistaken in concluding that my interest in bicycle snow tires implies an interest in flying to Aruba. At least this kind of behavioral advertising is better than seeing endless ads on cable news channels touting cures for obscure maladies I don’t have or care if I do. (“Ask your doctor about Peyronie’s disease.”)
By contrast, what I really don’t want is for the price I pay for bicycle snow tires, or vacation packages, or prescription drugs, or anything else, to be inflated just for me based on what my internet behavior reveals about my location, personal habits, health issues, tastes, or ability to pay. Nor do I want my online searches for cancer cures or alcohol treatment centers to lead my health insurer to cancel my coverage or my employer to quietly lay me off. Similarly, I don’t want to be discriminated against because some algorithm somewhere has sliced and diced me into a tossed salad of demo- and psycho-graphic segments, e.g., “middle-aged men in Lycra” (referred to as MAMILs by one market research firm), who pedal bespoke bikes through the snow to their professional jobs. In other words, I don’t want marketers using my personal data to decide that I am too affluent or too poor, eager or indifferent, averse or reckless, complacent or desperate, to hold out for a better deal.
Yet this kind of market discrimination is the defining mega-trend of our ever more digitized commercial life. Attempts to expand its use and effectiveness are the overwhelming reason why corporations are so eager to scoop up our personal data in the first place. As Andrew Odlyzko, the former head of the University of Minnesota’s Digital Technology Center, has written, “The powerful movement to reduce privacy that is coming from the private sector is motivated by the incentives to price discriminate, to charge different prices to various customers for the same goods or services.”
Corporations have no intrinsic interest in invading your privacy. They really don’t care who your Facebook friends are or even how many of them you’ve slept with. No, the real reason corporations want more and more of your personal data is because they are after something that businesses have coveted for millennia but could only imperfectly pull off. Think of the haggling rug merchant in the bazaar, or the car salesman on the showroom floor. What they most want to know is the maximum you’ll pay today for whatever they have on offer.
In the past, a salesman had to rely on intuition and crude, often biased indicators, like how customers dressed or spoke, their expressed interest or need, their credit rating, and often their race or gender. Sometimes this resulted in low-income customers getting a lower price because the merchant figured that otherwise he couldn’t make the sale at all. But more often the favoritism went, and still does, to those who need discounts the least. Abundant studies show, for example, that white male car buyers tend to be quoted significantly lower prices than black and female car buyers. A typical salesman will offer “special deals” to customers he thinks are too savvy to pay the “regular” price, or who he thinks have the wherewithal to act on better deals elsewhere.
In the past, these judgments were inevitably shaded by the seller’s assumptions and personal prejudices. That’s still true, but now they are more and more likely to be based on correlations that some algorithm thinks it has discovered by crunching a trillion points of big data about you and “your kind.” No one likes being discriminated against when all they want to do is buy a car. But today, it’s getting harder and harder to avoid dealing with marketers who can estimate with increasing accuracy just how much you, personally, are willing to pay, and who charge you accordingly.
Until a few years ago, efforts to personalize prices using digital data about the customer were relatively primitive. In 2012, for example, a Wall Street Journal investigation found that Staples.com was quoting people higher prices if they lived in an area that lacked an Office Depot or other Staples competitor. The same year, researchers published evidence that Amazon was routinely charging some customers 20 percent more (and in some cases 166 percent more) than other customers for the same Kindle e-book based on the customers’ location. The same researchers also found that Google would recommend more expensive or cheaper models of digital cameras, headphones, and other products to different customers based on what Google’s algorithm concluded was their ability to pay.
By 2016, a ProPublica investigation revealed that Amazon was engaging in a different dimension of marketplace discrimination—one that affects both buyers and sellers and that deeply distorts the ability of markets to set fair and efficient prices. Amazon both provides a platform for third-party vendors and sells products directly on the same platform. In this way, not only does Amazon own the biggest store in the largest mall, it owns the mall itself.
What ProPublica found was that when consumers entered this virtual mall and searched for the best deal on, say, Loctite Super Glue, Amazon would prominently display offers available directly from Amazon rather than those offered by highly rated merchants who were selling the same glue for less.
This is just the beginning. When people try to sell their wares on Amazon, whether they are publishers trying to sell books or merchants trying to sell glue, they have to accept the terms Amazon offers. Indeed, these days many can’t reach the customers they need except through Amazon, which makes it very hard for them to say no when, for example, Amazon suggests it’s time to fork over more money so it doesn’t bury their offers at the bottom of every search. And because Amazon effectively has the ability to look into their cash registers, it has deep knowledge of just how much they can afford to pay. It can use this knowledge to wring more money from sellers.
On current trends, these forms of discrimination are poised to get far worse. One reason is the vastly increasing amounts of data that individuals and businesses generate online. Second is the rapidly increasing processing power available through machine learning, artificial intelligence, and other advances in computing, which enable more sophisticated, highly tailored means of discriminating. According to a report by Deloitte and Salesforce, 40 percent of brands that currently deploy AI are using the technology not just to personalize the customer experience but also to tailor pricing and promotions in real time.
A third reason is the growth of tech platforms. Whether you are a merchant selling wares on Amazon, a driver selling rides on Uber, a homeowner renting out rooms on Airbnb, or a publisher posting content on Facebook, you are in a dependent relationship with a dominant corporation that can use its deep knowledge of your business to figure out how much it can get away with charging you.
Whether you are a merchant selling wares on Amazon, or a driver selling rides on Uber, you are in a dependent relationship with a dominant corporation that can use its deep knowledge of your business to figure out how much it can get away with charging you.
A final, highly important reason is the increasing degree of corporate concentration found throughout the economy. Engaging in egregious price discrimination doesn’t work very well when customers can easily take their business elsewhere. But for monopolistic corporations—which increasingly know that you have no real choice but to deal with them—price discrimination is both possible and highly lucrative.
Health care is a prime example. According to a study published in 2016 in the medical journal BMJ, hospitals charge about twenty times more to perform anesthesiology on an uninsured patient than on a patient covered by Medicare. Meanwhile, according to a study by the Congressional Budget Office, the price for a hospital stay is 89 percent higher when charged to commercial insurance plans and their customers than when a Medicare patient stays in the same bed for the same amount of time. The reason they can get away with this is monopolization. The rapid pace of mergers has more and more hospital markets dominated by a single dominant health care platform that controls not only most hospital beds but also the local doctors’ practices and labs as well. (See “The Case for Single-Price Health Care” in our April/May/June 2018 issue.) Now these giant health providers are merging with insurance companies, even as insurance companies merge with huge drug retailers like CVS. Imagine what all this concentrated market power can do with all the data out there on the internet about your health issues.
So what’s the solution? One approach is to imitate the Europeans. Last May, the European Union put into effect a set of internet privacy rules known as the General Data Protection Regulation (GDPR). A key provision of these regulations requires companies to obtain an individual’s explicit consent before storing or processing any of his or her personal online information. California has passed a similar law that will take effect in 2020, reflecting the idea’s popularity among progressives and privacy activists.
Yet this approach to protecting privacy has already had an unintended consequence. As the Wall Street Journalreported last April, before the GDPR even went into effect, Google and Facebook executives had become unexpectedly enthusiastic about the very regulations designed to contain their powers of surveillance. Before long, the reason became obvious. Google, Amazon, and Facebook had to spend massively on new digital infrastructure, but they found it comparatively easy to get their end users to click on the required privacy waivers. By contrast, smaller competitors, whether online publishers and websites or digital advertising firms, found compliance with the GDPR regulation prohibitively expensive and otherwise difficult to pull off. Investments in new tech start-ups shriveled, and according to the German internet search firm Cliqz, the market share of smaller online advertisers fell by roughly 30 percent.
So the EU inadvertently made the giant tech platforms even stronger and their dwindling number of competitors even weaker. Mark Zuckerberg himself captured the way such ill-advised regulation can foster monopoly when he told Congress last year, “A lot of times regulation by definition puts in place rules that a company that is larger, that has resources like ours, can easily comply with but that might be more difficult for a smaller start-up.”
A better approach would begin by attacking price discrimination directly.
Indeed, it’s the traditional American approach to dealing with monopolistic corporations that know too much about their customers’ business. It’s useful to recall that railroads, especially at the peak of their power at the turn of the twentieth century, had a lot in common with today’s tech giants. They were privately owned corporations that controlled, and in many places monopolized, a key network industry upon which virtually every American depended as both consumers and producers. Because of that, the men who owned railroads had access to information that gave them the ability to make or break different individuals, businesses, and whole regions depending on the terms of service they offered.
Western farmers, for example, found that the railroads charged them freight rates right up to the point that their profits dwindled to almost nothing. Meanwhile, the plutocrat John D. Rockefeller could coerce railroads into granting Standard Oil huge discounts beneath what rival shippers paid. The effect of such price discrimination was to foster inequality. “The great majority of local and personal discriminations are in favor of the strong,” noted the political scientist Arthur T. Hadley in his classic 1895 study of railroad economics. “As such they do great harm to the community by increasing inequalities of power.”
While discrimination based on broad criteria like race, gender, or religion remains very much illegal, the old prohibitions against more individually tailored forms of discrimination have mostly been repealed.
Americans responded to this problem by passing a variety of state and local laws prohibiting railroads from engaging in the most egregious forms of price discrimination. Then, in 1887, Congress created the Interstate Commerce Commission, the nation’s first independent federal regulatory agency. For nearly a century thereafter, the ICC went after railroads that engaged in unjust market discrimination against individuals, places, or lines of business. An early example was an 1891 ruling by the ICC that railroads could not deny African Americans traveling across state lines access to first-class cars. Meanwhile, the ICC made sure that small towns, small businesses, and farmers on the prairie didn’t face discrimination by insisting that railroads offer roughly the same terms of service to everyone, everywhere.
As new network industries came online, the government applied the same anti-discrimination principles to them. Airlines, for example, were regulated until the end of the 1970s under the same common carrier principles as railroads, which prevented them from offering inferior service and higher prices in what people today call “flyover” America. Similarly, regulators applied a public utility model to make sure that electric companies didn’t favor some businesses with lower rates than others or discriminate against different households in different neighborhoods.
Americans in the past further buttressed their anti-discrimination laws by prohibiting corporations from doing what Amazon does today when it vertically integrates into being, among many other things, a huge package delivery company. If you were a railroad, you couldn’t also be a retailer, or vice versa, because that would mean other retailers could never match you on shipping costs. Similarly, if you were a bank, you could not also be a manufacturer or chain store owner because what you knew about other businesses’ finances would give you an unfair advantage. And if you were a telecommunications company, you couldn’t get into the business of selling your customers’ personal information to advertisers.
It could have been much different. The original American Telephone and Telegraph Company, for example, might well have wound up selling data to marketers about who its customers were calling and even about what its surveillance of their phone conversations revealed about their health, income, and preferences. AT&T might even have gone a step further by offering free telephone service to customers who agreed to have their calls periodically interrupted by ad messages with special prices tailored just for them. But unlike Facebook and Google, AT&T was not allowed to be both a provider of communications infrastructure and a vertically integrated behavioral advertising agency, and thus its ability to engage in or foster market discrimination was highly limited.
Since the late 1970s, however, while discrimination based on broad criteria like race, gender, or religion remains very much illegal, the old prohibitions against more individually tailored forms of discrimination have mostly been repealed. Under the thrall of economists promising lower prices through “deregulation,” policymakers in both parties decided that it was time, for example, to lift restrictions on price discrimination by railroads, trucking firms, and airlines. This is why it now costs more to fly between many midsize and smaller cities in middle America than it does to fly clear across the country between New York and San Francisco. By the 1980s, meaningful antitrust was gone, too, along with “fair trade” laws that once constrained price discrimination at the wholesale level. And this just as the internet would soon give enormous built-in advantages to corporations that could leverage network effects to become monopolies of unprecedented size and power.
What would it look like if we restored the same anti-discrimination principles that we once used to contain the power of railroads and other network monopolies? The most recent major example came in 2015, when the Federal Communications Commission promulgated “net neutrality” rules prohibiting internet service providers from favoring some customers and discriminating against others when it comes to the speed and price of moving data across their networks. Yet the FCC has since reversed itself. Meanwhile, price discrimination by Google, Facebook, Amazon, and other giant tech platforms remains entirely unregulated.
Controlling these platforms like we once controlled railroads would begin by requiring them to publicly list their terms and prices, and to justify any that discriminated against or in favor of different users. Discrimination against individuals would be flatly illegal. Discrimination against different classes of customers would sometimes be permitted, but would have to be shown to serve the public interest. The ICC allowed railroads to charge different rates for transporting high-value items like watches or perishable food than for hauling low-value items like coal. This pricing brought in more revenue than if all classes of freight were charged the same and thereby helped railroads to meet their fixed costs. But where any such class-based price discrimination existed, it had to be explained and justified to the public.