6. America, Land of the Free Shipping

I first met Jeff Bezos at a 1996 retreat for CEOs of the venture capital firm of Kleiner, Perkins, Caufield and Byers. This may sound like a Davos- or Bohemian Club–style conclave of movers and shakers, but nothing could be farther from the truth. Most of the thirty or so attendees were relative newcomers to the Silicon Valley scene. Jeff was one of the first to realize that the opening of the Internet to commercial use might create significant business opportunities. Prior to that, it was restricted to government and research institutions for official business, and access was controlled by the Defense Advanced Projects Agency (DARPA).

With a bachelor’s degree in electrical engineering and computer science, and a series of technical jobs on Wall Street under his belt, Jeff had landed a position with a secretive but highly successful investment firm founded by an affable Columbia professor ten years his senior—Dave Shaw. Yes, the same Dave Shaw who pioneered programmed trading, founded D. E. Shaw and Company, and earned the Wall Street nickname King Quant. But one day, for reasons only he and Dave know, Jeff quit to seek his fortunes elsewhere. He and his new wife piled their possessions in their car and took off for Seattle. During the ride, he worked on a business plan while, I must assume, his wife drove.

Jeff was a high-energy, first-time entrepreneur with a boisterous laugh and a quirky smile. It seemed to me, as a fellow CEO of an early Internet startup, that Jeff exhibited an astounding lack of caution. I was always amazed that he would make multimillion-dollar business commitments with complete confidence that, when the time came, he would be able to raise the money required. And he was always right. Jeff’s idea was to start an online bookstore, which he gave the odd name of Amazon. As I recall, when we met, he was struggling with the minor problem that Amazon was too small to warrant the attention of book publishers, not to mention that he had no way to buy or sell books. We were both trying to figure out how to take payments over the Internet as, at the time, no respectable bank would process credit card transactions transmitted over some sort of public computer network they had never heard of. If you didn’t see or talk to your customers, how could you know they were who they claimed to be?

Lacking the capital and the connections to invest in warehouses and inventory, Jeff did the next best thing: he made a deal with the largest wholesaler at the time, Ingram Book Group. Ingram stocked and shipped books in small quantities to independent bookstores around the country, and the wholesaler also served as a resource for major chains if they ran short of stock locally and needed a quick delivery. The advantage, from Jeff’s perspective, was that Ingram would drop-ship orders as small as a single book, though I suspect the company wasn’t very happy about it.

My startup idea was that goods didn’t have to be sold online at fixed prices. So along with two partners I started Onsale.com, the first auction site on the Internet.1 But Jeff learned an important lesson from Dave Shaw that I had missed, at least at first: the real value wasn’t in the inventory, it was in the data.

Jeff recognized that the same basic principles that D. E. Shaw and Company applied to securities transactions could be applied to information provided by people. For him, at least initially, the fact that he was dealing with physical products was incidental, or at least secondary; the inventory and logistics were matters that could be subcontracted to third parties for a fee. The real essence of his business was the accumulating book reviews and purchasing histories of his customers. He recognized that the time it took to read a book represented a higher cost for most buyers than the nominal price of the product. Time spent digging into a read you didn’t like was wasted. So why not let his customers curate his products the way an experienced bricks-and-mortar bookstore salesperson might do? He correctly guessed that the opportunity to pontificate publicly and help others was reward enough for their efforts.

Stating the obvious, Amazon has not only revolutionized virtually all aspects of the book industry—with the possible exception of the writing—it has also grown to become one of the world’s largest sellers of just about everything. But there’s another way to look at Amazon’s remarkable rise.

When we describe Amazon as an “online retailer,” we think of it as a digital analog of a physical store. But there’s an alternative way to describe Amazon: as an application and expansion of D. E. Shaw and Company’s securities-trading strategies to the buying and selling of retail goods.

When Amazon was processing your order by transmitting it to Ingram for delivery, it was engaging in the same sort of arbitrage as Dave Shaw’s supercomputers: two simultaneous transactions that were guaranteed to make a profit, as long as they both settled. The first transaction was with you—an agreement to sell you a book at a particular price and deliver it within a mutually agreeable time frame (otherwise known as a futures contract), and a separate transaction was with Ingram to purchase a book for delivery to a specified destination. By constantly adjusting its “ask” price to you, Amazon locked in a known price spread as its gross profit. And like the programmed trading algorithm, the whole scheme worked because Amazon had better information than you did. Specifically, it knew where and how to purchase the item at a better price (that is, at Ingram), something you were not privy to nor could take advantage of directly.2

Jeff understands now, as he did then, that the power was in the data. He’s spent nearly two decades amassing an unprecedented array of statistics about individual and collective purchasing habits, including detailed personal information on more than 200 million active buyers. He mainly did this by losing money, but that’s sure to stop once your continued patronage is statistically assured, the potential pool of new buyers starts to shrink, and Jeff decides to curtail his investment and expansion efforts. (In business terminology, when the cost of new customer acquisition converges with the customers’ lifetime value.) As with other monopolies, once suppliers have no choice but to deal with Amazon, which is certainly true for the book industry, and customers are as loyal as courtesans, there’s a myriad of ways to extract the premium attendant to market dominance. (Contrary to popular belief, monopolies are not illegal. It’s how they use their market power to restrict competition that sometimes causes them to run afoul of the law.)

But how does Amazon put this cornucopia of information to productive use? For starters, by adjusting prices on the fly to achieve certain business objectives. As long as investors are willing to permit the company the leeway to grow at the expense of profits, Amazon wants to build your confidence that when you buy from them, you are getting the best, or at least a very good, price. The easiest way to meet this objective is to actually do it. So the company constantly monitors competitive prices, and adjusts its own prices accordingly.

If you’re an Amazon regular, you’ll notice that the prices of items in your shopping cart inexplicably change over time, sometimes by trivial amounts. Whole businesses have sprung up to monitor these fluctuations to help you snag the best price.3 The question is why. The seemingly random character of these changes suggests that they are the result of an automated process, most likely website crawlers pulling prices from other sites, or even on Amazon itself, for, in addition to selling directly, Amazon is a marketplace where others can sell the same goods. A price study at the University of Michigan in the summer of 2000 found that the price of a DVD on Amazon varied by up to 20 percent depending on the user’s browser and account. When a group of customers accused Amazon of charging them different prices for the same items, the company attributed the discrepancies to “a very brief test to see how customers respond to various prices.”4

Most people think it’s illegal for companies to charge different prices to different customers, but there’s nothing illegal or inappropriate about such practices as long as certain criteria—such as race, gender, and sexual orientation—are not used to discriminate. Amazon is not unique in this regard. A 2005 University of Pennsylvania research report discusses how grocery store loyalty card programs offer differently valued discount coupons at checkout depending on such factors as how brand-loyal you appear to be.5 In other words, if you were inclined to buy the item anyway, why almost literally give away the store?

The problem here is that all this wonderful laissez-faire is a prelude to à prendre ou à laisser—take it or leave it.6 The free flow of information around the Internet creates winner-take-all markets, and online retailing is no exception.7

Before the Internet, there were two points of friction that enabled a retailing market vigorous enough to profitably accommodate multiple sellers of identical goods. The first was information. How much more difficult was it to comparison shop, when you had to drive to a competitor’s store or search for their ads in the local newspaper, hoping to find the same item?

The other point of friction for retail goods used to be the effective cost of delivery. So what if the lamp you want to buy is available for less at a store one hundred miles away—it’s not worth the drive. In principle this could also be a problem on the Internet, because the cost to ship an item from a warehouse in New Jersey to New York City should be less than the cost to ship the same item to San Francisco. But Amazon has solved that problem as well, by using its economies of scale to pre-position the goods locally near major population centers, a luxury unavailable to most current or potential future competitors.

These two friction points are closely related, and Amazon has conflated them to brilliant advantage. Separating the cost to you into a product price and a shipping price is a fiction subject to manipulation. Mincing the total price into components is a time-honored way of obscuring true costs, though in the end it all boils down to one total figure.

Examples of this technique in other fields are the “destination,” “delivery,” and “documentation” fees added to the sticker price of cars. Medical billing has raised this art of confounding consumers to a new level of absurdity, sending facility (hospital) charges and doctor bills separately with different due dates, so you rarely know how much a service is costing you at the time you actually pay.8 Even online, trying to compare the features and prices of different models of computers is virtually impossible even for experts, as you need to determine precisely what is included and what you will need to purchase separately. In this regard, Apple Computer is a model of consumer transparency.

Like the medical care industry, Amazon pioneered a new way to obscure the price you pay not only by separating the required information into two disparate parts but by literally making it impossible to discern your total cost until long after you have actually made a purchase. The innovation was to charge you a fixed annual shipping fee—Amazon Prime–regardless of how many purchases you have made or will make during that year. (Arguably, Amazon Prime is an updated variant of the buying-club strategy of charging an annual membership fee.) This conjures the oxymoronic illusion of free shipping that you pay for. Paraphrasing economist Milton Friedman, there’s no such thing as free shipping—someone always pays. Is this fee worth it to you? Would you be better off paying a higher price that includes delivery from an alternative vendor? Only Amazon knows for sure.

Persuading you to prepay for shipping not only discourages you from shopping elsewhere, it makes rational buying impossible. This is at least one motivation for Amazon’s commendable focus on customer satisfaction; as long as you’re happy with the company, there’s little reason to question its pricing practices or to comparison shop—even if you could.

But the company has taken the doctrine of information asymmetry one step further. Amazon’s network of warehouses is so extensive, it has adopted the remarkable policy of permitting its competitors to list their own products on Amazon’s website and use Amazon’s own facilities for fulfillment. You might think this is an egalitarian act that serves to level the playing field by giving the “little guy” access to the same advantages that Amazon enjoys. But in reality, this ingenious tactic gives Amazon two additional potential competitive advantages: it has a picture window into competitors’ sales and prices, and ultimately the strategy gives Amazon control over its competitors’ costs because it can adjust the rates it charges for these services. After all, no one said it had to charge competitors only their proportional share of the fulfillment costs. Want to take over the market for electric toothbrushes? No problem, charge your competitors more to stock and ship their inventory than it costs you to handle the same products.

The common thread behind these business tactics is to acquire an enduring information advantage over customers and competitors, deftly wrapped in a narrative of low prices, outstanding service, and fair play.

I think Amazon is an amazing company and Jeff Bezos is a great guy, but there’s another reason the financial markets value the company at more than six hundred times earnings (2013), when the average is around twenty times earnings: they look forward to the inevitable time when the company extracts monopoly prices after locking in its customers and scorching the earth of competitors. And this is as it should be. Shoppers aren’t stupid; they will go where they get the best all-around deal, including convenience, service, and other factors. They aren’t concerned with whether their short-term purchasing behavior may restructure the retailing landscape to the detriment of future consumers any more than the original residents of Easter Island worried about whether the trees they chopped down for firewood might contribute to a desolate, bleak landscape for their descendants.

But when the river of prices starts to rise and the profits pour in, the familiar competitive reference points with which to judge the value received will have long been submerged or swept away. Jeff’s decision to name the company Amazon—the name of the biggest river in the world, which sweeps away everything in its path—now makes more sense to me.

A focus on ensuring an information advantage was not the only lesson that Jeff learned from Dave Shaw. He also took to heart the immense benefits that advanced computer technology could bring to profiting from this data.

In principle, Amazon could have accomplished much of its success using only traditional retail practices behind the scenes—hiring product managers to monitor competitors and set prices, purchasing agents to select and order inventory, and warehouse workers to pick and ship orders. And indeed Amazon has plenty of all of these. But Jeff also invested heavily and early in automated systems to exploit his unique advantages. These aren’t the conventional data-processing systems of bricks-and-mortar competitors, which don’t have the opportunity to instantly adjust prices to market conditions and individual customer habits. And that’s where artificial intelligence, particularly machine learning systems, comes in.

Continuously testing and adjusting prices on an individual basis while responding to competitive threats is mind-bogglingly complex. Whether your goal is to reinforce the perception of low prices or to maximize profits, extraordinary speed and judgment is required, applied thousands of times a second across countless simultaneous transactions. Conducting this massive ballet requires a synthetic intellect. And that’s exactly what Amazon has built.

Are customers willing to pay a penny more for tissue after their home team loses a closely fought football game? Are buyers in the winning city less sensitive about the price of champagne? Are you likely to purchase a spare iPhone charger at full price if you can get it on Tuesday, or are you the sort of person who needs a slight discount to put you over the edge? Do people who stream classic films before noon prefer reading romances to mysteries on their Kindle? And by exactly how much?

These questions may be difficult to answer, but at least you can detect some glimmer of logic behind them. Even to consider asking them requires a level of insight that few professional human marketers possess. A synthetic intellect, on the other hand, suffers from no such constraints. Perhaps customers with hyphenated names are willing to pay more for artificial flowers on weekdays than on weekends, those who live in apartments prefer books with blue covers to books with red ones, or MasterCard holders are less likely to return earphones if purchased along with other items. This is the unfathomable ocean explored by machine learning algorithms. Here’s an actual sample complaint on an Amazon discussion board: “I’ve been following the 42LV5500 [LG forty-two-inch HDTV], the price drops to 927 over night, then around 9am EST, it pops to 967, then around 5pm it drops between $5–$10, then drops again to 927 overnight. I noticed it in the morning, cancelled my order and repurchased at the 927 price. It is infuriating. They have followed this pattern for 3 days straight.”9

Today Amazon maintains the illusion, if not the reality, of having everyday low prices. In the future, nothing will deter such companies from presenting you, and only you, with precisely the offers and deals that will maximize the company’s profits.

Throughout this process, you will remain thoroughly in control. After all, it’s a free country—you can decide for yourself, take it or leave it, choose whatever path strikes your fancy. But while you may exercise freedom as an individual, collectively we will not. Synthetic intellects are fully capable of managing the behavior of groups to a fine statistical precision while permitting individuals to roam in whatever direction their predictable little hearts desire.

Amazon, of course, is simply a single example of a phenomenon that is quietly expanding into many aspects of our lives. Synthetic intellects of every variety discreetly bargain with us, take our measure, and note our interests. But there’s a fine line between simply bringing relevant opportunities to our attention and incenting us to take actions that benefit others. And the authority to craft these incentives, and thereby manage our collective behavior, is gradually moving from people to machines.

Today, coupons pop up on your smartphone as you drive past the local mall.10 Soon, you will wake up to text messages that offer a reserved parking spot close to your office if you agree to leave for work fifteen minutes earlier than usual, give you a free movie ticket if you skip watering your lawn, let you upgrade your iPhone a year sooner if you donate a pint of blood to your health maintenance organization before Friday.11

Our lives will be filled with individual propositions like these, managed by synthetic intellects whose goals are to optimize traffic, conserve natural resources, and manage health care. But there will be other systems with less lofty goals—to sell out the last doughnut as close as possible to closing time, to rent you a street-view apartment when a river view is also available, to route you through Salt Lake City while reserving the direct flights for higher-paying last-minute travelers.

But the part of these systems visible to you will only be the snowcap on the mount. Behind the scenes, they will also be furiously negotiating and bartering among themselves to accomplish their goals. How did the water department’s resource management system get those Fandango movie ticket coupons? By contracting with another system whose goal is to generate revenue for the movie theaters. How did your HMO’s blood plasma inventory system arrange for that iPhone upgrade? By bartering with another system charged with extending cell phone contracts.

The problem, as with the automated securities-trading systems competing in the same forums as human traders, is that you will be incessantly horse-trading with systems that have overwhelming advantages over you: speed; access to timely information; knowledge of exactly what the next person is likely to accept; and the ability to predict your own behavior better than you can. You’ll be playing against the house in a game where the dealer counts every card and knows exactly how the deck is sorted. You’ll be surrounded by Amazons in all aspects of your life, with no humans in sight.

This is a strange frontier, without precedent in the history of humankind. The new regime will creep in silently and unnoticed, as if on cat paws, while you marvel at how the modern world grows ever more convenient, customized to you, and efficient. But behind the scenes, enormous synthetic intellects will be shaving you the thinnest slice of the benefits that you are willing to accept, while reserving the lion’s share for … exactly whom?