“If you want to be successful in business (in life, actually), you have to create more than you consume. Your goal should be to create value for everyone you interact with. Any business that doesn’t create value for those it touches, even if it appears successful on the surface, isn’t long for this world. It’s on the way out.” So wrote Jeff Bezos in his final letter to shareholders, released last week. It’s a great sentiment, one I heartily agree with and wish that more companies embraced. But how well does he practice what he preaches? And why is practicing this so hard by the rules of today’s economy?

Bezos started out by acknowledging the wealth that Amazon has created for shareholders — $1.6 trillion is the number he cites in the second paragraph. That’s Amazon’s current market capitalization. Bezos himself now owns only about 11% of Amazon stock, and that’s enough to make him the richest person in the world. But while his Amazon stock is worth over $160 billion, that means that over $1.4 trillion is owned by others.

“I’m proud of the wealth we’ve created for shareowners,” Bezos continued. “It’s significant, and it improves their lives. But I also know something else: it’s not the largest part of the value we’ve created.” That’s when he went on to make the statement with which I opened this essay. He went on from there to calculate the value created for employees, third-party merchants, and Amazon customers as well as to explain the company’s Climate Pledge.

Bezos’ embrace of stakeholder capitalism is meaningful and important. Ever since Milton Friedman penned the 1970 op-ed in which he argued that “the social responsibility of business is to increase its profits,” other constituencies — workers, suppliers, society at large, and even customers — have too often been sacrificed on the altar of shareholder value. Today’s economy, rife with inequality, is the result.

The Limits of John Mackey’s Brand of Enlightened Capitalism

While I applaud the goal of understanding “who gets what and why” (which in many ways is the central question of economics), I struggle a bit with Bezos’ math. Let’s walk through those of his assertions that deserve deeper scrutiny.

How much went to shareholders?

Our net income in 2020 was $21.3 billion. If, instead of being a publicly traded company with thousands of owners, Amazon were a sole proprietorship with a single owner, that’s how much the owner would have earned in 2020.

Writing in The Information, Martin Peers made what seems to be an obvious catch: “Instead of calculating value by looking at the increase in Amazon’s market cap last year — $679 billion — Bezos uses the company’s net income of $21 billion. That hides the fact that shareholders got the most value out of Amazon last year, far more than any other group.”

But while Peers has put his finger on an important point, he is wrong. The amount earned by shareholders from Amazon is indeed only the company’s $21.3 billion net income. The difference between that number and the $679 billion increase in market cap didn’t come from Amazon. It came from “the market,” that is from other people trading Amazon’s stock and placing bets on its future value. Understanding this difference is crucial because it undercuts so many facile criticisms of Bezos’ wealth, in which he is pictured as a robber baron hoarding the wealth accumulated from his company at the expense of his employees.

The state of “the market” has become a very bad proxy for prosperity. Those lucky enough to own stocks are enjoying boom times; those who do not are left out in the cold.

The fact that Bezos is the world’s richest person makes him an easy target. What we really need to come to grips with is the way that our financial system has been hijacked to make the rich richer. Low interest rates, meant to prop up business investment and hiring, have instead been diverted to driving up the price of stocks beyond reasonable bounds. Surging corporate profits have been used not to fuel hiring or build new factories or bring new products to market but on stock buybacks designed to artificially boost the price of stocks. The state of “the market” has become a very bad proxy for prosperity. Those lucky enough to own stocks are enjoying boom times; those who do not are left out in the cold.

Financial markets, in effect, give owners of stocks the value of future earnings and cash flow today — in Amazon’s case, about 79 years’ worth. But that’s nothing. Elon Musk is the world’s second-richest person because the market values Tesla at over 1,000 years of its present earnings.

The genius of this system is that it allows investors and entrepreneurs to bet on the future, bootstrapping companies like Amazon and Tesla long before they are able to demonstrate their worth. But once a company has become established, it often no longer needs money from investors. Someone who buys a share of a hugely profitable company like Apple, Amazon, Google, Facebook, or Microsoft isn’t investing in these companies. They are simply betting on the future of their stock price, with the profits and losses coming from others around the gaming table.

In my 2017 book, WTF? What’s the Future and Why It’s Up to Us, I wrote a chapter on this betting economy, which I called “supermoney” after the brilliant 1972 book with that title by finance writer George Goodman (alias Adam Smith). Stock prices are not the only form of supermoney. Real estate is another. Both are rife with what economists call “rents” — that is, income that comes not from what you do but from what you own. And government policy seems designed to prop up the rentier class at the expense of job creation and real investment. Until we come to grips with this two-track economy, we will never tame inequality.

The fact that in the second paragraph of his letter, Bezos cites Amazon’s market cap as the value created for shareholders but uses the company’s net income when comparing gains by shareholders to those received by other stakeholders is a kind of sleight of hand. Because of course corporate profits — especially the prospect of growth of corporate profits — and market capitalization are related. If Amazon gets $79 of market cap for every dollar of profit (which is what that price-earnings ratio of 79 means), then if Amazon were to raise wages for employees or give a better deal to its third-party merchants (many of them small businesses), that would lower its profits, and presumably its market cap, by an enormous ratio.

Every dollar given up to these other groups isn’t just a dollar out of the pocket of shareholders. It is many times that. This of course does provide a very powerful incentive for public companies to squeeze these other parties for every last dollar of profit, encouraging lower wages, outsourcing to eliminate benefits, and many other ills that contribute to our two-tier economy. It may not be Amazon’s motivation — Bezos has always been a long-term thinker and was able to persuade financial markets to go along for the ride even when the company’s profits were small — but it is most certainly the motivation for much of the extractive behavior by many companies today. The pressure to increase earnings and keep stock prices high is enormous.

These issues are complex and difficult. Stock prices are reflexive, as financier George Soros likes to observe. That is, they are based on what people believe about the future. Amazon’s current stock price is based on the collective belief that its profits will be even higher in the future. Were people to believe instead that they would be meaningfully lower, the valuation might fall precipitously. To understand the role of expectations of future increases in earnings and cash flow, you have only to compare Amazon with Apple. Apple’s profits are three times Amazon’s, and free cash flow is four times, yet it is valued at only 36 times earnings and has a market capitalization less than 50% higher than Amazon’s. As expectations and reality converge, multiples tend to come down.

How did Amazon’s third-party sellers fare?

[We] estimate that, in 2020, third-party seller profits from selling on Amazon were between $25 billion and $39 billion, and to be conservative here I’ll go with $25 billion.

That sounds pretty impressive, but how much of a profit margin is it really?

Amazon doesn’t explicitly disclose the gross merchandise volume of those third-party sellers, but there is enough information in the letter and in the company’s 2020 annual report to make a back-of-the-napkin estimate. The letter says that Amazon’s third-party sales represent “close to 60%” of its online sales. If the 40% delivered by Amazon’s first-party sales come out to $197 billion, that would imply that sales in the third-party marketplace were almost $300 billion, and $25 to $39 billion in profit on $300 billion works out to a profit margin between 8% and 13%.

But is Amazon calculating operating income, EBITDA, or net income? “Profit” could refer to any of the three, yet they have very different values.

Let’s generously assume that Amazon is calculating net income. In that case, small retailers and manufacturers selling on Amazon are doing quite well since net income from U.S. retailers’ and manufacturers’ overall operations are typically between 5% and 8%. Without knowing which profit number Amazon’s team is estimating, though, and the methodology they use to arrive at it, it is difficult to be sure whether these numbers are better or worse than what these sellers achieve through other channels.

One question that’s also worth asking is whether selling on Amazon in 2020 was more or less profitable than it was in 2019. While Amazon didn’t report a profit number for its third-party sellers in 2019, it did report how much its sellers paid for the services Amazon provided to them. In 2019, that number was about $53.8 billion; in 2020, it was $80.5 billion, which represents a 50% growth rate. Net of these fees, income to Amazon but a cost to sellers, we estimate that seller revenue grew 44%. Since fees appear to be growing faster than revenues, that would suggest that in 2020, Amazon took a larger share of the pie and sellers got less. Of course, without clearer information from Amazon, it is difficult to tell for sure.

Meanwhile, Amazon took in another $21.5 billion in “other income,” which is primarily from advertising by sellers on Amazon’s platform. That grew by 52% from 2019’s $14 billion, again suggesting that Amazon’s share of the net is growing. And unlike some forms of advertising that bring in new customers, much of Amazon’s ad business represents a zero-sum competition between merchants bidding for top position, a position that in Amazon’s earlier years was granted on the basis of factors such as price, popularity, and user ratings.

How about employees?

In 2020, employees earned $80 billion, plus another $11 billion to include benefits and various payroll taxes, for a total of $91 billion.

There’s no question that the $91 billion that Amazon paid out in wages and benefits in 2020 is meaningful. Some of those employees were very well compensated, others not so well, but all of them have jobs. Amazon is now one of the largest employers in the country. It is an exception to the tech industry in that it creates a large number of jobs—and not just high-end professional jobs—and that some of the jobs it creates are in locations where work is scarce.

That being said, Bezos’ description of the amount earned by employees is misleading. In every other case, he makes an effort to estimate the profit earned by a particular group. For employees, he treats the gross earnings of employees as if it were profit, writing, “If each group had an income statement representing their interactions with Amazon, the numbers above would be the ‘bottom lines’ from those income statements.”

No, Bezos, employee earnings are their top line. Just as a company has gross income before expenses, so do employees. The bottom line is what’s leftover after all those expenses have been met. And for many of Amazon’s lower-paid employees — as is the case for lower-paid workers all over the modern economy — that true bottom line is negative, that is, less than they need to survive. Like workers at other giant profitable companies like Walmart and McDonald’s, a significant fraction of Amazon warehouse employees require government assistance. So, in effect, taxpayers are subsidizing Amazon because the share of the enterprise’s profits allocated to its lowest-paid employees was not enough for them to pay their bills.

Amazon was the “worst offender” among a rogues’ gallery of high-tech companies that use aggressive tax avoidance strategies.

That points to a major omission from the list of Amazon’s stakeholders: society at large. How does Amazon do when it comes to paying its fair share? According to a 2019 study, Amazon was the “worst offender” among a rogues’ gallery of high-tech companies that use aggressive tax avoidance strategies. As noted in The Guardian, “Fair Tax Mark said this means Amazon’s effective tax rate was 12.7% over the decade when the headline tax rate in the U.S. has been 35% for most of that period.” In 2020, Amazon made a provision for taxes of $2.863 billion on pretax income of $24,178 billion, or about 11.8%. This may be legal, but it isn’t right.

Amazon is clearly moving in the right direction with employees. It introduced a $15 minimum wage in 2018, ahead of many of its peers. And given the genius of the company, the commitment to workplace safety and other initiatives to make Amazon a better employer that Bezos highlighted in his letter are likely to have a big payoff. When Amazon sets out to do something, it usually invents and learns a great deal along the way.

“We have always wanted to be Earth’s Most Customer-Centric Company,” Bezos wrote. “We won’t change that. It’s what got us here. But I am committing us to an addition. We are going to be Earth’s Best Employer and Earth’s Safest Place to Work. In my upcoming role as Executive Chair, I’m going to focus on new initiatives. I’m an inventor. It’s what I enjoy the most and what I do best. It’s where I create the most value. … We have never failed when we set our minds to something, and we’re not going to fail at this either.”

I find that to be an extremely heartening statement. At Amazon’s current stage of development, it has the opportunity and is beginning to make a commitment to put its remarkable capabilities to work on new challenges.

Stakeholder value means solving multiple equations simultaneously

I was very taken with Bezos’ statement that “if any shareowners are concerned that Earth’s Best Employer and Earth’s Safest Place to Work might dilute our focus on Earth’s Most Customer-Centric Company, let me set your mind at ease. Think of it this way. If we can operate two businesses as different as consumer ecommerce and AWS, and do both at the highest level, we can certainly do the same with these two vision statements. In fact, I’m confident they will reinforce each other.”

One of my criticisms of today’s financial-market-driven economy is that by focusing on a single objective, it misses the great opportunity of today’s technology, summed up by Paul Cohen, the former DARPA program manager for A.I. and now a professor at the University of Pittsburgh, when he said, “The opportunity of A.I. is to help humans model and manage complex interacting systems.” If any company has the skills to do that, I suspect it will be Amazon. And as Bezos wrote elsewhere in his letter, “When we lead, others follow.”

Amazon is also considering environmental impact. “Not long ago, most people believed that it would be good to address climate change, but they also thought it would cost a lot and would threaten jobs, competitiveness, and economic growth. We now know better,” Bezos wrote. “Smart action on climate change will not only stop bad things from happening, it will also make our economy more efficient, help drive technological change, and reduce risks. Combined, these can lead to more and better jobs, healthier and happier children, more productive workers, and a more prosperous future.” Amen to that!

In short, despite my questions and criticisms, there is a great deal to like about the directions Bezos set forth for Amazon in his final shareholder letter. In addition to the commitment to work more deeply on behalf of other stakeholders beyond customers and shareholders, I was taken with his concluding advice to the company: “The world will always try to make Amazon more typical — to bring us into equilibrium with our environment. It will take continuous effort, but we can and must be better than that.”

It is in the spirit of that aspiration that I offer the critiques found in this essay.

Originally published on the O’Reilly blog as “Checking Jeff Bezos’s Math.”

How Much Value Has Amazon Created for the World? was originally published in Marker on Medium, where people are continuing the conversation by highlighting and responding to this story.

Note: This is a republication of an article that I published last week on LinkedIn Pulse exploring the limits of Amazon’s vaunted “customer obsession” when pursued at the expense of suppliers, and contrasting it with the philosophy of Ingram, the book wholesaling platform that gave Amazon its start.

A few days later, Jeff Bezos published his 2020 letter to Amazon shareholders, which almost reads like a response to my article. Though that could not have been possible given that his shareholder letter must have been well underway when my article first appeared, the update to Amazon’s thinking is very relevant. In his letter, Jeff seems to have recognized some of the ways that Amazon’s customer-primacy focus is insufficient. He wrote:

“If you want to be successful in business (in life, actually), you have to create more than you consume. Your goal should be to create value for everyone you interact with. Any business that doesn’t create value for those it touches, even if it appears successful on the surface, isn’t long for this world. It’s on the way out.”

Jeff went on to provide calculations about how much value Amazon has created for each of the constituencies the company touches. While some of his math doesn’t really hold up (that will be the subject of another post), it’s clear that he is thinking hard about the issue of how to balance competing priorities, and that’s a really good thing. Given the history of Amazon’s brilliant innovations in pursuit of the goals that it sets out, its new ambition to be “Earth’s Best Employer and Earth’s Safest Place to Work” in addition to “Earth’s most customer centric company” is likely to lead to some remarkable changes. I still hope to see Amazon aspire to be Earth’s most supplier-centric platform as well.

With that, here’s my original article.

Jeff Bezos is the world’s richest person, and Amazon, the company he founded, one of the world’s most admired and valuable. Two recent books, Invent and Wander: The Collected Writings of Jeff Bezos, with an introduction by Walter Isaacson, and Working Backwards, by longtime Amazon executives Colin Bryar and Bill Carr, offer lessons from the company’s enormous success.

The Family Business, by Keel Hunt, due out April 20, tells the story of another company, Ingram Industries, which, not coincidentally, played an indispensable role in enabling Amazon’s initial success as the world’s largest online bookstore. Ingram is a family owned business, founded in 1857 as a sawmill in Wisconsin but reinvented multiple times, eventually becoming a transportation and distribution company based in Nashville. 50 years ago, it branched out into book wholesaling, later adding video (and in the heyday of that industry, packaged software.) When Amazon was founded In 1995, it was essentially a web front-end to Ingram’s warehouses and its database of virtually every book that was commercially available. Even today, the Ingram Content Group is a key part of the hidden infrastructure of publishing and bookselling in the US, including Amazon.

Reading these corporate biographies in parallel provides a lot of food for thought. I spend a lot of my time these days studying marketplaces and the technology platforms that enable them: Amazon, Google, Shopify, Alibaba, and of course, my own O’Reilly learning platform. I’m interested in what makes marketplaces succeed and what makes them fail. And in particular, I’m trying to understand how modern technology-based platforms decide the central question of economics: who gets what and why?

When I was asked to write a foreword for The Family Business, I agreed to do so on the strength of my own company’s long association with Ingram — it was a key enabler of O’Reilly’s publishing business as well as of Amazon’s and still remains an important partner — but also because the decisions that Ingram has made as a company have something important to teach us about today’s economy, lessons that are complementary to those that are taught by Amazon. Ingram, like Amazon, teaches us about the necessity of constant corporate reinvention. It teaches us the advantages that come from being a platform — an enabler of other businesses. But it also teaches some lessons that are in stark contrast to those from Amazon and other Silicon Valley giants.

The core narrative of Silicon Valley is of the invention of a new, magical user experience so transformative that it draws hundreds of millions of users: a storefront from which you can order any product with one click, a search engine that gives access to all the world’s information, a phone that is “insanely great,” an app that summons a car and driver to pick you up within minutes wherever you are and take you wherever you want to go. Exponential user growth is seen as the ultimate measure of success. Today, Silicon Valley companies look to be valued at billions of dollars on that metric alone, when some of them can hardly be called businesses, since they have no profits and may even lack a plan for earning any.

Jeff Bezos founded one of the first of the internet’s hyper growth companies, but he understood that the reality is far more complex than simply growth in users. In 2001, he supposedly drew Amazon’s strategy on a napkin. The picture looked something like this:

Jeff pictured a flywheel in which sellers provide a big selection of products, and the unique Amazon customer experience of unparalleled access to those products drives more traffic, drawing even more sellers. Growth of a super-scale business allows a lower cost structure, allowing Amazon to lower prices for customers, which drives an even better customer experience, which drives more traffic, draws more sellers and more products, around and around, faster and faster.

Companies like Amazon, Uber and Lyft, and even Google and Netflix, are marketplaces, connecting and enabling both buyers and sellers. Amazon connects buyers to hundreds of millions of products; Uber and Lyft connect riders with drivers, and Google and Netflix connect readers and viewers with content providers.

One of the big problems in these hyper-scaled marketplaces is building up both sides of the market at the same time. Uber and Lyft demonstrate just how expensive it is. They invested billions of dollars in marketing, and even today arguably sell their services for less than it costs to provide them in order to acquire more passengers.

It’s a lot easier if you only have to build one side of the market. When Amazon launched in 1995 as “the world’s biggest bookstore,” it didn’t have to spend money assembling a critical mass of books, publishers, and authors. Ingram had already done that. Starting in 1970, Ingram had been connecting publishers and bookstores, such that any bookstore — not just Amazon — had access to every book in print. Jeff’s revolutionary insight, the one that launched Amazon, was that the web made it possible to create a friendly online interface to Ingram’s enormous catalog and that technology could be used to radically simplify the process of ordering and delivering. And the flywheel began to spin.

By 2001, when Jeff drew his flywheel diagram, Amazon was already selling electronics and music CDs as well as books, and before long, it was the interface to virtually anything its customers might want to buy. Amazon also created its own, much faster, real-time distribution layer, while continuing to rely on Ingram (and other wholesalers of different kinds of products) for those products that have less demand. As its flywheel spun faster and faster, Amazon took in more and more products and vendors, built more and more infrastructure for warehousing and delivery, and became the master of logistics that we see today. With the success of its third-party marketplace, millions of sellers now compete to offer hundreds of millions of products.

As explained in Working Backwards, one key to Amazon’s success has been its relentless focus on the consumer. Customer obsession is the first of Amazon’s leadership principles. In his 1998 Letter to Shareholders, Bezos wrote, “Our customers have made our business what it is, they are the ones with whom we have a relationship, and they are the ones to whom we owe a great obligation.”

And there’s the rub. Because Amazon understands so well that delighting the customer with lower prices, faster delivery, and a better customer experience drives its growth, it can sometimes forget that it operates a two-sided marketplace in which its merchants also matter. Rather than considering its merchants as among those “with whom we have a relationship, and … the ones to whom we owe a great obligation,” Amazon seems to view them as a resource to be exploited, an inexhaustible fount of redundant supply to whom no obligation is owed.

This is the Achilles heel of Silicon Valley. Focus on the user, taken as the only gospel, becomes a liability. Amazon faces antitrust investigations in both Europe and the US based not just on strongarm tactics against competitors but against its merchants. Google is likewise being investigated for competing against the web sites whose content it was originally created to help consumers search. Uber co-founder and CEO Travis Kalanick’s palpable disdain for one of his drivers led to a massive PR backlash and his ouster from the company.

Amazon’s treatment of its merchants seems like a curious blind spot in a company that has been so prescient, so innovative, and so capable of creating value for those in its ecosystem. Looking at Bezos’s flywheel, it should be clear to the company that merchants are as important to the flywheel as customers.

Why does this happen? Unlike many critics of Silicon Valley, I don’t think it’s because the leaders of these companies are making decisions solely motivated by profit as is so often claimed by their critics. In fact, Jeff Bezos, Larry Page and Sergey Brin, and Mark Zuckerberg are profoundly thoughtful individuals working to do the right thing. The problem is that they are working within an economic system that values growth above all else, disdains small businesses as inefficient, and tilts the playing field against them.

In the dominant economic narrative for the last 40 years, growth and “economic efficiency” (reflected in lower prices) are the key measures of progress. This favors giant firms at the expense of smaller ones, leading to a two-track economy in which some “superstar firms” soak up all the profits while others are left behind. Unfortunately, that means workers in a large part of the economy and our society are among those left behind. Worsening economic inequality is not something that just happens. It has been designed into the modern economic system.

Silicon Valley represents the apex of this type of thinking. Investors and entrepreneurs celebrate “blitzscaling,” hypergrowth before profits, to build an unshakeable lead, even a monopoly position, before competitors realize what they are up to. The rewards of such a position are enormous. Payoffs in the stock market can be far greater than could ever be achieved through earned operating income. Tesla’s stock, for example, currently has a price/earnings ratio of 1106. What that means in plain language is that the company’s stock market valuation represents more than 1100 years — yes, over a millennium! — of its current level of profit. Amazon’s valuation is somewhat more reasonable, representing only 77 years of today’s profits. Ironically, the better known and more expected the stream of future profits, the lower the valuation seems to be. Apple, with a profitability rate 5 times that of Amazon, only has 34 years of profit reflected in its stock price. Meanwhile, 81% of all IPOs in 2020 were for companies with no profits at all! (In 1980, by contrast, 91% of IPOs were for profitable companies. In 1999, right before the dotcom bust, the number was down to 14%. Count that as a warning.)

Since most of the compensation for Silicon Valley companies’ most valuable and sought-after employees is in stock options, growing that valuation inevitably becomes one of the company’s core objectives, even when the company consciously values other factors ahead of growth. The machine requires growth, or else it breaks down. It’s much like driving an SUV rather than a smaller vehicle. Regardless of your values or where you are going, when you pull up to the pump, you need to fill up a bigger tank.

As is true across most of the modern economy, which sacrifices many an externality on the altar of growth, distorting a complex system by favoring one element at the expense of others is bad in the long term. For example, while Amazon has made the market for many goods more efficient when it comes to lower-cost goods, it has raised the risk of counterfeit goods. Jeff Bezos has told his team that “[other companies’] margin is our opportunity,” and accordingly, Amazon works to eliminate anyone it considers a middleman between the consumer and the ultimate source of supply. As Business Insider pointed out, though, this didn’t eliminate costs so much as it “shifted them to different, often hard to police and control places instead.”

“Before online retail, supply chains relied on friction to achieve quality. Becoming a vendor to Walmart required years of work and experience. Those vendor relationships were precious and would last for decades. Because of how hard it was to build one, Walmart could trust on the network of vendors to keep up the quality. In turn, they were invested in vetting their suppliers. Friction in the system meant the supply chain could be trusted. And if anything went wrong, there was a clear path to follow to find the responsible party.”

For Amazon, competition with its merchants also means that those merchants have an incentive to look elsewhere for a better deal. Over the years, Amazon has rebuffed competitors from Ebay to Walmart. Shopify, a platform company that provides infrastructure for companies to operate their own ecommerce sites, is the first rival that has begun to catch up to Amazon, with Gross Merchandise Volume now about $120 billion to Amazon’s $490 billion (versus $100 billion for Ebay, and Walmart in the “single digit billions.”) One executive at Shopify said to me, “Amazon went down the wrong path enough for us to exist.” [Correction: the original version of this story incorrectly listed Ebay’s market cap of $38 billion instead of its GMV.]

What does all this have to do with Ingram?

Ingram is a private company. That means it doesn’t have a public stock price that allows it to receive decades of future earnings today. In this sense, it’s an old-fashioned company, which provides a service and makes its money in the form of each year’s profit. A dollar of earnings is worth a dollar to the company, not $1100 (Tesla), not $77 (Amazon), not $34 (Apple) or $37 (Google or Microsoft.)

Unexpectedly, this allows a company to take a longer-term, more balanced view. If you can achieve an astronomic valuation on user growth alone, it is easy to convince yourself that any improvement that delights users and speeds user acquisition is worthwhile, whether it be lower prices, faster delivery, or more corporate efficiency to enable those things, even if it is at the expense of other elements of the flywheel, such as the merchants who sell on your platform or the drivers who deliver the packages or the passengers to their destination.

Ingram doesn’t have “users.” It is a B2B platform. Both sides of its marketplace are businesses: that is, publishers and bookstores (in the segment of its business that we have always dealt with.) And it has to thoughtfully balance the needs of both of them. It can’t sacrifice one to please the other. And it doesn’t have to do so to please Wall Street. Ingram’s management understands that the businesses on both sides of its marketplace are its customers, and obsesses about both of them.

Ingram’s innovation began with support for booksellers. In 1973, the company provided a weekly microfiche feed of new titles, radically improving the ability of small bookstores to keep up with the output of the fast-growing publishing industry. However, much of what has driven Ingram over the years is innovation designed to support its suppliers (authors and publishers). For example, when Ingram introduced print-on-demand via its Lightning Source subsidiary in 1997, it radically improved the economics for publishers of slow-moving books, like university presses. Since unit costs decrease with volume, publishers were always tempted to overproduce, ending up with massive inventory costs, and often, inventory write offs. Before O’Reilly went print-on-demand with Ingram for many of our titles, we routinely wrote off millions of dollars a year of unsalable inventory. For many small publishers, this is the difference between life and death as a company.

Ingram focused tirelessly on making the book supply chain more efficient for everyone — remaining invisibly in the background, such that today, small bookstores selling online can have books dispatched from Ingram’s warehouses with the bookstore’s own branding.

There’s no question that Amazon has also introduced many services that benefit the supplier side of its marketplace. But Amazon’s innovations on behalf of the supplier side often come with costs designed to soak up their margin. Merchants on the platform are expected to compete fiercely with each other for attention. Amazon’s huge and fast growing advertising business, for example, can be seen as a tax on merchants. Before the addition of this lucrative business, merchants mostly had to compete on product quality and price. Now, they must also pay to play. It’s hard to see how this doesn’t end up being bad for the consumer as well. Where once we were sure that Amazon was showing us the products that represented the best combination of high user ratings, popularity, and low price, now we are first shown paid placements.

We have learned a lot from Amazon that guides our work at O’Reilly, but we have also learned an enormous amount from Ingram. As I said in the Foreword that I wrote for The Family Business:

“[Ingram] is a company that has always served its partners, growing as we grow, and never at our expense.

Print publishing is a much smaller part of O’Reilly’s business today (though Ingram delivers a far larger part of it, particularly since 2005, when O’Reilly president Laura Baldwin made a deal with John Ingram to switch us over to print-on-demand using Ingram services — another example of the transformative power of Ingram’s infrastructure investments). Today, the largest part of our business is O’Reilly’s online learning platform, a subscription-based digital marketplace for over forty-five thousand business and technical books, thirty thousand hours of video, live online training, interactive coding environments, technology certifications, and more, provided by hundreds of content partners and used by five thousand enterprise clients.

In managing this marketplace, we continue to take our inspiration from Ingram’s generous enablement of its partner ecosystem, even as we use technologies from the digital realm like machine learning and personalization to help our customers navigate the best and most useful content and learning experiences provided both by our own team and by our partners. At O’Reilly, one of our mottos is “Create more value than you capture. I like to think that we learned to think that way at least in part because of our long partnership with Ingram.”

One of the most important lessons we learned from Ingram is what we might call supplier-focused innovation. Many of the new features on the oreilly.com platform didn’t come about because we were thinking about new features to delight our customers (though that was also one of our goals), but because we were trying to figure out new ways to make money for our marketplace suppliers — the authors, publishers, and other content producers who co-create the value of our platform.

We launched our platform in 2000 (seven years before Amazon entered the ebook market with the Kindle) after I received a pitch from one of the many startups in the ebook aggregation space. I realized that while it aimed to offer a great deal to consumers, it wasn’t sustainable. As I did some back-of-the-napkin math for the startup over dinner, I pointed out that their offer was for the grand total of six cents per book per subscriber, which we would then have to share with our authors. If you believed, as I did, that ebooks were going to be a primary modality for customers, that wasn’t going to cut it. We needed a business model that would preserve incentives for authors to keep producing new books (and later, many other types of materials.)

Years later, we introduced one of our most popular modalities on the platform, live online training, when Laura Baldwin realized that we needed to create new sources of revenue for our authors. Our customers turned out to love it, but the idea came from thinking about how we could offer new opportunities to our creators.

So yes, read the books about what we can all learn from Amazon. It is rightly one of the bellwether companies of the 21st century. But don’t forget what we learn from the giants of an earlier era, like Ingram, before winner-takes-all became the name of the game. A platform has an obligation to serve all of its partners and stakeholders, not just its users.

A Tale of Two Platforms was originally published in Marker on Medium, where people are continuing the conversation by highlighting and responding to this story.

I had no idea I’d played a role in bringing the idea back into popular discourse, and when Will asked me to comment, I wrote to him in email:

“I had no idea I’d played a role in popularizing the phrase!

It’s never been central to my thinking. It’s cute, and reminds people not to take for granted the deal that they are entering into. But it is a deal.

I think very differently about the value exchange between consumers of an advertising driven product, the product’s owners, and the advertisers. This is a two sided market. The primary market is an unpriced market in which producers and consumers of content are matched up by interest, and the consumers ‘pay’ for the value received with their attention, which is then monetized on the other side of the market by providing a paid opportunity to advertisers to also bid for that attention. Both sides of the market receive value. And in an efficient ad market, like pay per click advertising on Google search (vs. display advertising), the advertisers are bidding for attention just like the free content providers.

Thinking about it this way helps us to understand many of today’s markets, some of which are explicitly priced, and others which appear to be free. Uber is also a two-sided market, for example, where both drivers and riders are offered a free app in return for matching them up when the passengers are looking for a ride. The paid transaction then occurs between the two parties.

The reason I think of this as a special case of any other market is that it highlights that the fundamental questions of fairness in markets still apply. Is the exchange fair? (In many ad-driven markets, the exchange is more than fair, in my opinion. We get enormous amounts of free value in exchange.) Is the exchange aboveboard? It is here that many of the ad-driven services fall down, because they resell people’s data beyond the use that was anticipated and clear in the terms of the exchange. You can think of this as analogous to hidden fees in a paid transaction.

The ultimate question in any market is “who get’s what–and why.” That’s the title of a book by Nobel-prize winning economist Alvin E. Roth, a pioneer in market design. Creating “thick markets” where the parties trust each other is fundamental. And it seems to me that in today’s algorithmically driven big data platforms, we are seeing amazing discoveries about market design.”

I was happy to see that Will picked up on this concept late in the article (though without using my comments or attributing it to me) when he wrote:

“There are at least two alternative ways of viewing our relationship to Facebook that hold more promise for making that relationship a healthier and less exploitive one. The first is to view ourselves as customers of Facebook, paying with our time, attention, and data instead of with money. This implies greater responsibility on both sides. If we understood that Facebook and other ‘free’ online services exact real costs to things we value, we might use them more sparingly and judiciously. We might finally grasp that every time we grant new data permissions or sign on to a new privacy policy, we’re almost certainly giving up a lot. Even if we don’t have time to read the whole thing, let alone comprehend it, we could equate it in our minds to spending hundreds of dollars — and then make better decisions about whether that specific app or update is still worth it to us. Ideally this formulation of users as customers forces on Facebook and other apps the responsibility of earning their loyalty, convincing them that their service is worth the tradeoffs, and not violating their trust.”

The second alternative he provides was attributed to Jaron Lanier, who makes the case that we should regard ourselves (and ask to be remunerated) as workers for Facebook.

I like Will’s conclusion: “If we don’t like how Facebook is treating us, we shouldn’t throw up our hands and call ourselves the product of a system over which we have no control. We should act like people — customers, workers, citizens, whatever — who have the power to demand change.”

The article is chock-full of interesting insights. You should read the whole thing.

We must renew trust in our institutions!

I spoke a few months ago with Philip Howard, the head of the Computational Propaganda Project at the Oxford Internet Institute. In the course of our conversation about the use of bots and fake social media profiles to game Facebook’s algorithms in order to influence the US election, I asked him whether the US intelligence agencies were asleep about the possibilities of the Internet to spread micro-targeted disinformation.

Not all, he replied. They were well aware of the techniques the Russians had used, and even had used them themselves against other countries. They just never imagined, he said, that they would be turned on the US. They were the kind of thing that were used against banana republics, with corrupt institutions and low trust in government. No one would dare use them against the most powerful nation in the world.

His reply immediately brought to mind future US President John F. Kennedy’s Harvard honors thesis from 1940, which was later published under the title Why England Slept. It was a response to While England Slept, a collection of Winston Churchill’s speeches in the years leading up to World War II, in which Churchill had warned repeatedly against Neville Chamberlain’s policy of appeasement, which allowed Germany to re-arm and ultimately to initiate hostilities against Europe. I had recently seen both books lying on a table at a friend’s house where I had gone for dinner, and the titles had stuck with me.

“For five years I have talked to the House on these matters — not with very great success. I have watched this famous island descending incontinently, fecklessly, the stairway which leads to a dark gulf. It is a fine broad stairway at the beginning, but after a bit the carpet ends. A little farther on there are only flagstones, and a little farther on still these break beneath your feet…”

Winston Churchill, While England Slept

Kennedy asked, why did England not see that Germany was re-arming? Why did they think that their past glories kept them safe? Why were they blind to the changing nature of warfare that would render their island moat far less of a protection than they imagined?

We might ask similar questions. Why has America not seen the pervasive lack of trust in government, the media, and other institutions that made a social media attack possible? Why did so many of our political and business elites seek advantage by enabling and encouraging that lack of trust? How did we find ourselves “descending incontinently, fecklessly, the stairway which leads to a dark gulf”? And most importantly, what do we do about it?

A week later, I had another conversation, this time with Bill Eggers of Deloitte, who is doing research on the changing military and intelligence threats of the 21st century. He mentioned that one of the strategic asymmetries that he has identified is that in countries like China, close cooperation between government and the commercial sector gives the government access to vast amounts of data and the latest AI, data science, and social networking technology, while in the US, distrust of government means that companies give their data to government only when compelled to do so. There is a fear among military and intelligence planners that this will put the US at greater and greater disadvantage over time.

I had a somewhat surprising response to this observation. Rather than saying “good for our companies, who are standing up to our government,” I found myself asking instead “why is it that we trust companies like Apple, Amazon, Google, and Facebook more than we trust our own government? Our government is supposed to be of, by, and for the people!”

There’s been an idea growing in my mind that building institutions we can actually trust, at scale, is one of the great challenges of the 21st century.

The seeds of that idea had been planted by a constant drumbeat of articles about the blockchain and its supposed elimination of the need for trusted institutions. This quote from a 2014 article by CoinDesk, “Bitcoin: A Means for Global Independence,” gives the flavor of so many articles about blockchain: “innovative wallet technologies mean anyone can store vast amounts of value without the need to trust governments or banks.”

This strikes me as exactly the wrong way to frame the problem. It’s the cyber-equivalent to the narrative of Idaho survivalists, stockpiling guns and food for the coming apocalypse, or of billionaires building disaster boltholes in New Zealand. Self-sufficiency for the individual is not what we need! Like it or not, we are all in this together. The opportunity of a technology like blockchain isn’t to free us from trust in our institutions; it should be described as an experiment in building new kinds of collective institutions that we can trust. Projects like New America’s Blockchain Trust Accelerator, which seeks to connect “Blockchain-based pilot concepts with the right technologists, jurisdictions, and funders to achieve better governance outcomes,” seem like a step in the right direction.

Trust is the currency of civilization. Increases in trust, and new mechanisms for people to trust each other at a distance, are what has allowed human society to grow from small warring bands into great nation states, and to form world-spanning commercial institutions. It is now time for us to take the next step towards trust at global scale. And I wonder if, despite their current struggles, internet technology platforms are pointing the way forward.

We’ve all read the stories that begin, “If Facebook were a country…” Here’s the World Economic Forum’s version from 2016:

Since then, Facebook claims to be heading for three billion users; there are two billion Android phone users, one billion iPhone users; and Google owns four or five online properties with at least one billion users.

It seems to me that the question these articles attempt to answer — the relative size of social networks and other internet services relative to today’s nation-states — is rather trivial. I want to ask instead: “if today’s platforms were nation states, how best can they use technology, big data, and their knowledge of their users to improve the lives of those users in the same way that nations must aspire to improve the lives of their citizens?” and “if today’s nation states were internet services, how would they act differently?” I’m going to focus here on the second question. Asking this question also helps us to see things that nation states may be doing, now or in the future, as they catch up to the digital capabilities of consumer internet services.

It’s clear that even despite the current backlash against the great tech platforms, trust in them remains high. Expressions of distrust in the media notwithstanding, we haven’t yet seen any kind of exodus from digital technologies. Billions of people still share on Facebook, perform Google searches, buy from Amazon, transact with banks. Billions of people voluntarily give up their location at every moment to their mobile carrier, to their mobile phone operating system provider, and to location-based application providers. Through our phone calls, our text messages, and our emails, we reveal virtually every personal and business relationship. While there are applications for anonymous communication, few bother to adopt them.

For all their flaws, we trust the great tech platforms more than we trust our own government. Governments have much to learn from their successes, and also from their failures.

As a thought experiment, imagine that the residents of a country are equivalent to the users of a social network or other internet service, and that the nation state is the equivalent of one of these services.

Some of the conclusions I’ve come to are uncontroversial; others might be provocative:

• Just as platforms are always trying to get new users, nations would always be trying to get new residents (that is, they would be pro-immigration), because they know that an economy is the result of people working together, exchanging goods and services, and that more people using their platform and their services is a good thing.
• Nations would think of themselves as platforms, whose goal is to harness the creativity of a their marketplace to create new services for their users, just as platforms like Apple, Amazon, Google, and Microsoft do. This is what a nation state does for its economy when it creates the rule of law, improves its infrastructure, polices the activities of participants on the service, and invests in basic research, and tries to compensate for failures in their market economies. I have in fact long argued that government is a platform, and needs to get good at it, using the tools of the 21st century.
• As Chinese economist Justin Yifu Lin wrote in his book Against the Consensus, governments would innovate by “observing the world as though seeing it for the first time through a newly born baby’s eyes, without preconceived notions.” This is what entrepreneurs do in envisioning radical new opportunities. There are new capabilities being developed every day, new opportunities to rethink the way things have been done in the past. We must seize them!
• They would understand their comparative advantage. Before introducing a new policy, they would (again quoting Justin Yifu Lin) try to “construct a causality model anew…by identifying the agents behind the phenomenon, the goals they wanted to achieve, the constraints they faced, and the options they had.” My friends Dan and Meredith Beam taught me that this is also the essence of a business model: “the way that all of the parts of a business work together to create customer value and marketplace advantage.”
• They would try to get their residents to commit more deeply — they would want their occasional users to become “daily active users” who understand how much value they get from the platform; that is, they would want their residents to become engaged citizens. And they would worry about net outflow of employees or users to competing countries just as platforms worry about losing customers or users to new entrants.
• They would try to draw the best talent. Sure enough, the competition for the best talent between Google, Facebook, Microsoft, Amazon, and other internet giants is also being played out by nation states.
• They would constantly develop new services for their residents, and constantly re-invent the old ones, eliminating obstacles to their effectiveness, measuring whether they have actually achieved their intended goals, and replacing them if they did not. This would not be an exercise carried out every few decades, or as the result of an election, but the daily work of government. (Teaching government how to adopt iterative, data-driven, user-centric practices is at the heart of the work of Code for America, on whose board of directors I serve. That work needs to be applied not just to technology development but to policy development.)
• They would be exploring the potential of AI and other algorithmic systems to create more equitable distribution of the fruits of the productivity gains from technology and better business and governance practices. (I am fascinated by the way that internet platforms are pioneering new algorithmic tools for managing marketplaces to create unexpected economic windfalls, and better outcomes for their participants. This is a longer topic than I can address here, and will be the subject of a separate piece. In the meantime, I did talk about it a bit in my keynote at Venturebeat’s Blueprint Conference two weeks ago.)
• They would collect enormous amounts of data on their citizens. If they are unscrupulous, they will use it to manipulate our behavior, and to keep us under the thumb of those in power. But why do we imagine only this dark alternative? It is precisely because of the power that governments have as the institution that sets the rules for society and creates the platform upon which society builds that we must update the institutions that hold them accountable. The creators of the US Constitution designed a unique and powerful set of checks and balances that were intended to make government more trustworthy despite the enormous imbalance of power. We need to embrace this same path in the 21st century.

I’m not holding up China as a model to emulate. Despite their apparent success on the Edelman Trust Barometer and in improving the economic lives of their people, they are using data to manipulate their people and to hold them in check. What I’m saying is that it seems inevitable that the US will emulate China, which is why we need to work hard as a nation to create a government that is worthy of our trust, and that doesn’t use the data it collects against us. I know that may be a utopian dream, but it’s one we need to work towards.

And it is in this regard that the current efforts to hold technology platforms accountable, and the efforts of those platforms to regain trust have so much to teach us. Jessi Hempel wrote a compelling report from the World Economic Forum in Davos earlier this year on discussions there about technology and trust. Read it.

Platforms are learning that they must be more transparent about what they do with the data they collect; they are learning today that manipulating users to increase sales or the use of their product may give short-term advantage, but that it erodes trust and usage over time.

A platform gains trust to the extent that it uses the data it collects on behalf of its users, not against them, providing benefits that they could not achieve alone. Platforms (and governments) are means for collective action, and they should be for collective benefit, not for extraction of value by the platform owner. This will be no less true in the future.

So too, politicians who who squander their people’s trust for short-term electoral advantage are not true leaders; they are destroying the foundation of their nation’s greatness, which is based on trusted institutions that allow people to work together towards common goals.

The Chinese philosopher Lao Tzu said it 2500 years ago:

“A leader is best
When people barely know that he exists,
Not so good when people obey and acclaim him,
Worst when they despise him.
‘Fail to honor people,
They fail to honor you;’
But of a good leader, who talks little,
When his work is done, his aim fulfilled,
They will all say, ‘We did this ourselves.’”

And:

“One who knows his lot to be the lot of all other men
Is a safe man to guide them,
One who recognizes all men as members of his own body
Is a sound man to guard them.”

As platforms evolve, basing themselves on new technologies like AI, blockchain, and marvels yet to be invented, they must stay true to this principle!

I write more about lessons from computer platforms for business, government, and the economy in my book WTF? What’s the Future and Why It’s Up to Us. For ongoing perspective on technology and the economy, subscribe to the O’Reilly Next:Economy Newsletter, or visit the WTF Economy site on Medium. And to learn about the tools of technology and 21st century business, subscribe to Safari, O’Reilly Media’s online learning platform.

Why America Slept was originally published in From the WTF? Economy to the Next Economy on Medium, where people are continuing the conversation by highlighting and responding to this story.

There’s a scene in Lin-Manuel Miranda’s Hamilton in which Thomas Jefferson, who has been away as ambassador to France after the American Revolution, comes home and sings, “What’d I miss?”

We all have “What’d I miss?” moments, and authors of books most of all. Unlike the real-time publishing platforms of the web, where the act of writing and the act of publishing are nearly contemporaneous, months or even years can pass between the time a book is written and the time it is published. Stuff happens in the world, you keep learning, and you keep thinking about what you’ve written, what was wrong, and what was left out.

Because I finished writing my new book, WTF? What’s the Future and Why It’s Up to Us, in February of 2017, my reflections on what I missed and what stories continued to develop as I predicted form a nice framework for thinking about the events of last year.

Our first cyberwar

“We just fought our first cyberwar. And we lost,” I wrote in the book, quoting an anonymous US government official to whom I’d spoken in the waning months of the Obama administration. I should have given that notion far more than a passing mention.

In the year since, the scope of that cyberwar has become apparent, as has how all of the imagined scenarios we used to prepare turned out to mislead us. Cyberwar, we thought, would involve hacking into systems, denial-of-service attacks, manipulating data, or perhaps taking down the power grid, telecommunications, or banking systems. We missed that it would be a war directly targeting human minds. It is we, not our machines, that were hacked. The machines were simply the vector by which it was done. (The Guardian gave an excellent account of the evolution of Russian cyberwar strategy, as demonstrated against Estonia, Ukraine, and the US.)

Social media algorithms were not modified by Russian hackers. Instead, the Russian hackers created bots that masqueraded as humans and then left it to us to share the false and hyperpartisan stories they had planted. The algorithms did exactly what their creators had told them to do: show us more of what we liked, shared, and commented on.

In my book, I compare the current state of algorithmic big data systems and AI to the djinni (genies) of Arabian mythology, to whom their owners so often give a poorly framed wish that goes badly awry. In my talks since, I’ve also used the homelier image of Mickey Mouse, the sorcerer’s apprentice of Walt Disney’s Fantasia, who uses his master’s spell book to compel a broomstick to help him with his chore fetching buckets of water. But the broomsticks multiply. One becomes two, two become four, four become eight, eight sixteen, and soon Mickey is frantically turning the pages of his master’s book to find the spell to undo what he has so unwisely wished for. That image perfectly encapsulates the state of those who are now trying to come to grips with the monsters that social media has unleashed.

This image also perfectly captures what we should be afraid of about AI — not that it will get a mind of its own, but that it won’t. Its relentless pursuit of our ill-considered wishes, whose consequences we don’t understand, is what we must fear.

We must also consider the abuse of AI by those in power. I didn’t spend enough time thinking and writing about this.

Zeynep Tufekci, a professor at the University of North Carolina and author of Twitter and Tear Gas, perfectly summed up the situation in a tweet from September: “Let me say: too many worry about what AI — as if some independent entity — will do to us. Too few people worry what *power* will do *with* AI.” That’s a quote that would have had pride of place in the book had it not already been in production. (If that quote resonates, watch Zeynep’s TED Talk.)

And we also have to think about the fragility of our institutions. After decades of trash-talking government, the media, and expertise itself, they were ripe for a takeover. This is the trenchant insight that Cory Doctorow laid out in a recent Twitter thread.

The runaway objective function

In April of 2017, Elon Musk gave an interview with Vanity Fair in which he used a memorable variation on Nick Bostrom’s image of an AI whose optimization function goes awry. Bostrom had used the thought experiment of a self-improving AI whose job was to run a paper-clip factory; Elon instead used a strawberry-picking robot, which allowed him to suggest that the robot aims to get better and better at picking strawberries until it decides that human beings are in the way of “strawberry fields forever.”

In the book, I make the case that we don’t need to look to a far future of AI to see a runaway objective function. Facebook’s newsfeed algorithms fit that description pretty well. They were exquisitely designed to show us more of what we liked, commented on, and shared. Facebook thought that showing us more of what we asked for would bring us closer to our friends. The folks who designed that platform didn’t mean to increase hyperpartisanship and filter bubbles; they didn’t mean to create an opening for spammers peddling fake news for profit and Russian bots peddling it to influence the US presidential election. But they did.

So too, the economists and business theorists who made the case that “the social responsibility of a business is to increase its profits” and that CEOs should be paid primarily in stock so that their incentives would be allied with the interests of stockholders thought that they would make the economy more prosperous for all. They didn’t mean to gut the economy, increase inequality, and create an opioid epidemic. But they did.

We expect the social media platforms to come to grips with the unintended consequences of their algorithms. But we have yet to hold accountable those who manage the master algorithm of our society, which says to optimize shareholder value over all. In my book, I describe today’s capital markets as the first rogue AI, hostile to humanity. It’s an extravagant claim, and I hope you dig into the book’s argument, which shows that it isn’t so far-fetched after all.

We need a new theory of platform regulation

In my book, I also wrote that future economic historians will “look back wryly at this period when we worshipped the divine right of capital while looking down on our ancestors who believed in the divine right of kings.” As a result of that quote, a reader asked me if I’d ever read Marjorie Kelly’s book The Divine Right of Capital. I hadn’t. But now I have, and so should you.

Marjorie’s book, written in 2001, anticipates mine in many ways. She talks about the way that the maps we use to interpret the world around us can lead us astray (that is the major theme of part one of my book) and focuses in on one particular map: the profit and loss statements used by every company, which show “the bottom line” as the return to capital and human labor merely as a cost that should be minimized or eliminated in order to increase the return to capital. This is a profound insight.

Since reading Marjorie’s book, I’ve been thinking a lot about how we might create alternate financial statements for companies. In particular, I’ve been thinking about how we might create new accounting statements for platforms like Google, Facebook, and Amazon that show all of the flows of value within their economies. I’ve been toying with using Sankey diagrams in the same way that Saul Griffith has used them to show the sources and uses of energy in the US economy. How much value flows from users to the platforms, and how much from platforms to the users? How much value is flowing into the companies from customers, and how much from capital markets? How much value is flowing out to customers, and how much to capital markets?

This research is particularly important in an era of platform capitalism, where the platforms are reshaping the wider economy. There are calls for the platforms to be broken up or to be regulated as monopolies. My call is to understand their economics and use them as a laboratory for understanding the balance between large and small businesses in the broader economy. This idea came out in my debate with Reid Hoffman about his idea of “blitzscaling.” If the race to scale defines the modern economy, what happens to those who don’t win the race? Are they simply out of luck, or do the winners have an obligation to the rest of us to use the platform scale they’ve won to create a thriving ecosystem for smaller companies?

This is something that we can measure. What is the size of the economy that a platform supports, and is it growing or shrinking? In my book, I describe the pattern that I have observed numerous times, in which technology platforms tend to eat their ecosystem as they grow more dominant. Back in the 1990s, venture capitalists worried that there were no exits; Microsoft was taking most of the value from the PC ecosystem. The same chatter has resurfaced today, where the only exit is to be acquired by one of the big platforms — if they don’t decide to kill you first.

Google and others provide economic impact reports that show the benefit they provide to their customers, but they also have to consider the benefit to the entrepreneurial ecosystem that gave them their opportunity. The signs are not good. When I looked at Google’s financial statements from 2011 to 2016, I noted that the share of its ad revenue from third-party sites had declined from nearly 30% to about 18%. There may be many reasons for this, but it certainly calls for some research. Amazon deserves similar scrutiny. Fifteen of the top 20 Kindle best sellers were published by Amazon.

There are other ways that these platforms have helped create lots of value for others that they haven’t directly captured for themselves (e.g., Google open-sourcing Android and TensorFlow and Amazon’s creation of Web Services, which became an enabler for thousands of other companies). Still, how do we balance the accounts of value extracted and value created?

We need a new theory of antitrust and platform regulation that focuses not just on whether competition between giants results in lower prices for consumers but the extent to which the giant platforms compete unfairly with the smaller companies that depend on them.

Augment people, don’t replace them

Enough of the news that expands on the darker themes of my book!

The best news I read in the ten months since I finished writing the book was the research by Michael Mandel of the Progressive Policy Institute that shows that ecommerce is creating more and better jobs than those it is destroying in traditional retail. “To be honest, this was a surprise to me — I did not expect this. I’m just looking at the numbers,” Mandel told Andrew Ross Sorkin of the New York Times. Here is Mandel’s paper.

This report nicely complemented the news that from 2014 to mid-2016, a period in which Amazon added 45,000 robots to its warehouses, it also added nearly 250,000 human workers. This news supports one of the key contentions of my book: that simply using technology to remove costs — doing the same thing more cheaply — is a dead end. This is the master design pattern for applying technology: Do more. Do things that were previously unimaginable.

Those who talk about AI and robots eliminating human workers are missing the point, and their businesses will suffer for it in the long run. There’s plenty of work to be done. What we have to do is to reject the failed economic theories that keep us from doing it.

That’s my call to all of you thinking, like me, about what we’ve learned in the past year, and what we must resolve to do going forward. Give up on fatalism — the idea that technology is going to make our economy and our world a worse place to be, that the future we hand on to our children and grandchildren will be worse than the one we were born into.

Let’s get busy making a better world. I am optimistic not because the road ahead is easy but because it is hard. As I wrote in the book, “This is my faith in humanity: that we can rise to great challenges. Moral choice, not intelligence or creativity, is our greatest asset. Things may get much worse before they get better. But we can choose instead to lift each other up, to build an economy where people matter, not just profit. We can dream big dreams and solve big problems. Instead of using technology to replace people, we can use it to augment them so they can do things that were previously impossible.”

Let’s get to work.

This article originally appeared as the New Year’s 2018 issue of the O’Reilly Next:Economy Newsletter. Subscribe to get news each week about technology and its impact on the economy.

What’d I miss? was originally published in From the WTF? Economy to the Next Economy on Medium, where people are continuing the conversation by highlighting and responding to this story.

That’s the title of my new book, due out on October 10 from Harper Business in the US, and on October 17 from Penguin Random House in the UK. The book won’t be on sale till October 10, but you can pre-order on Amazon now!

WTF? can be an expression of amazement or an expression of dismay. In today’s economy, we have far too much dismay along with our amazement, and technology bears some of the blame.

In the book, I share some of the techniques we’ve used at O’Reilly Media to make sense of and predict past innovation waves such as the commercialization of the internet, open source software, the internet as a platform, big data, open government, and the maker movement. I apply those same techniques to provide a framework for thinking about how today’s world-spanning platforms and networks, on-demand services, and artificial intelligence are changing the nature of business, education, government, financial markets, and the economy as a whole. I give tools for understanding how all the parts of modern digital businesses work together to create marketplace advantage and customer value, and why ultimately, they cannot succeed unless their ecosystem succeeds along with them.

The book is a combination of memoir, business strategy guide, and call to action. It covers the lessons that I’ve learned about technology platforms from four decades in the industry, and how they apply both to businesses and to the broader economy.

The core of the call to action is an exhortation to businesses to DO MORE with technology rather than just using it to cut costs and enrich their shareholders. Robots are going to take our jobs, they say. I say, “Only if that’s what we ask them to do!” I’ve had my fill of technological determinism. Technology is the solution to human problems, and we won’t run out of work till we run out of problems. Entrepreneurs need to set their sights on how we can use big data, sensors, and AI to create amazing human experiences and the economy of the future, making us all richer in the same way the tools of the first industrial revolution did. Yes, technology can eliminate labor and make things cheaper, but at its best, we use it to do things that were previously unimaginable! What is our poverty of imagination? What are the entrepreneurial leaps that will allow us to use the technology of today to build a better future, not just a more efficient one?

One of the major threads in the book is how the tech economy is increasingly dominated by platforms. Whether it’s Uber or Airbnb (the companies du jour) or Google, Facebook, Apple, Amazon, and Twitter, or financial markets, these platforms must create a “thick marketplace” of producers and consumers. When they try to extract too much value for the platform owner, they ultimately fail. (I first observed this pattern in the PC software ecosystem, and ever since have been warning internet companies against making the same mistakes Microsoft did.) Platforms must create more value than they capture if their ecosystem is to survive.

If a platform fails to create a sustainable ecosystem, entrepreneurs look elsewhere, and the platform loses its vitality. This is true for nations and economies as well.

In addition, we see that platforms are increasingly managed by algorithms, and that these algorithms always have a “fitness function,” the thing they are relentlessly optimizing for. Despite their seeming intelligence, algorithmic systems do exactly what we tell them to do, with unintended and sometimes frightening results. We have to be extremely careful what we ask for.

Uber optimizes for passenger pickup time, Google optimizes for relevance (in both search results and in ads), Facebook optimizes for engagement (and we see how that went wrong with fake news). What do our financial markets optimize for? That leads me to an account of financial markets as the first rogue AI, hostile to humanity. Income inequality, declining upward mobility, and job losses due to technology are not inevitable; they are the result of design choices we have made in the algorithms that manage our markets.

Robots are going to take our jobs, they say. I say, “Only if that’s what we ask them to do!”

I’m optimistic, though. There is plenty of work to be done, and plenty of uncompensated or poorly-compensated work (in caring and creativity and education) already being done, which can be the basis of a 21st century economy even as technology replaces many of the jobs we do today.

The fundamental design pattern of success with technology is to enable people to do things that were previously impossible. Companies that only use technology to do less by getting rid of people will be surpassed by those who use it to help them to do more. Companies must commit themselves to continuously educating and empowering people with the tools of the future, not just treating them as a cost to be reduced.

The great unanswered question is why technology is so often being used to eliminate jobs and to make a small segment of society ever richer rather than being used to do more, to solve the great problems we face today and to make society as a whole prosperous. I make the case that the answer is in the fitness function we have given to our economic algorithms, and the unexamined rules we use to guide our economy. Just as Google constantly updates its algorithms in pursuit of relevant search and ad results, and as Facebook wrestles with how to rethink its algorithms for user engagement in response to fake news, we must now rewrite the algorithms that shape our economy if we want our society to prosper in the age of the brilliant technologies of the 21st century, and use them to create a more human-centered future.

Here are some of the things that have already been said about the book. I’m blushing. I hope that the praise is justified, and that you find the book as stimulating as my advance readers have done.

But more than that, I hope that the ideas in this book teach and inspire a generation of entrepreneurs to build lasting companies with real impact, to work on stuff that matters, and to commit themselves to making a richer and more equitable world for all.

The book won’t be on sale till October 10, but you can pre-order on Amazon now!

WTF? What’s the Future and Why It’s Up To Us was originally published in From the WTF? Economy to the Next Economy on Medium, where people are continuing the conversation by highlighting and responding to this story.

We hear again and again that AI and robots are going to take away human jobs. My broken kettle says otherwise.

Yesterday, I set my electric kettle down awkwardly on the edge of the sink. Crash! It toppled over and smashed. I searched Amazon for a replacement, found several that were highly rated, and within minutes had placed an order. As a Prime customer, I had the option for same day delivery, by 6 pm, and so I brewed this morning’s tea without interruption or inconvenience.

Remember when it was amazing that Amazon offered free two-day shipping? Then free one-day shipping? Now for many products, it’s a matter of hours before your order is on your doorstep.

Amazon is constantly upping the ante. It doesn’t just cut costs. It uses technology to do more, delighting customers with better service and lower prices. And of course, Amazon’s customers respond by buying more products. Amazon grows faster, invests more, and delights more customers, who buy more, in a virtuous circle.

This is an example of what Jeff Bezos calls “the Flywheel.”

Far too many companies seem to miss the point that lower cost structure, lower prices, and a broader selection are only part of the flywheel. Unless it all adds up to better customer experience, the flywheel wobbles, and eventually loses its momentum.

What does all this have to do with AI and the jobless future? Take a look at this chart showing the increase in employment at Amazon during the three years that it went from 1,400 robots in its warehouses to 45,000:

I spoke recently with a fulfillment and logistics executive at Amazon who told me that robots have allowed the company to pack more products into its warehouses, and to speed up picking, so that it can put more products into rapid fulfillment. Amazon expects to hire another 100,000 workers in the next eighteen months, many of them in its fulfillment centers.

And that doesn’t include all the people working in actual delivery. Do you remember when the United States Post Office was seemingly on its last legs, cutting services and delivery hours? If you’re like me, you’re now getting multiple deliveries per day, and the postman might well show up in the evening, working overtime. Amazon Flex, Amazon’s peer-to-peer delivery service (akin to Lyft or Uber, but for delivery only) is apparently growing so rapidly that it may overtake Lyft as the second largest source of employment for on-demand drivers.

Nor does it include employment at the more than 100,000 small companies that use Amazon’s platform to sell and distribute their own goods. Many of these companies would have no access to the market without Amazon. (My brother is a good example. He runs a small distribution business out of a warehouse in Front Royal, VA, shipping used books, auto parts, and various imported products he thinks his customers might find interesting.)

Amazon reminds us again and again that it isn’t technology that eliminates jobs, it is the short-sighted business decisions that use technology simply to cut costs and fatten corporate profits.

This is the master design pattern for applying technology: Do more. Do things that were previously unimaginable.

Now, you might think that same-day delivery is a slender reed on which to hang dreams of a better future, one more technology solution to a “first world problem.” But the revolution in logistics will spread far beyond more quickly meeting the needs of people who already have everything. Zipline, the on-demand drone delivery startup, is a good example. They aren’t delivering consumer goods, but vital blood supplies and medicine, leapfrogging the need for 20th century roads and hospital infrastructure, bringing life-saving blood to clinics and small local hospitals anywhere in Rwanda within 15 or 20 minutes. (Post-partum hemhorrage is one of the leading causes of death in Rwanda because keeping every blood type available within quick reach of everyone who needs it has been prohibitively expensive.)

Keller Rinaudo on Twitter

First delivery of blood to a hospital https://t.co/1URysgL6iS

Along with other signals, such as Uber’s on-demand flu shot program, you can see the outline of a future healthcare system that brings care to the people who need it, wherever they are, employing not just robots but community health care workers upskilled with technology (including sensor-based telemedicine, augmented reality, and AI), providing better care at lower cost, putting to work vast numbers of people in the fastest-growing industry of the 21st century.

If, like Amazon, our healthcare system was laser focused on making life better for its customers, what might it do differently? Hospitals wouldn’t be using technology to reduce costs, jack up prices for access to the latest high-tech wizardry, and limit the amount of time doctors can spend with patients. They’d be letting the machines do what they do best — increase efficiency — so that people could spend more time with each other, providing richer, better, more human care.

As logistics and delivery become more automated, and costs continue to fall, what other kinds of services might be completely rethought, putting people to work delighting each other with currently unimaginable products or levels of service?

There is an enormous failure of imagination among those who think that we face a jobless future. The weavers of the 1811 Luddite rebellion, who smashed the machine looms that were threatening their livelihood, couldn’t imagine that their descendants would have more clothing than the kings and queens of Europe. Machines expanded the demand for the labor of weavers, augmented by machines, because it lowered the cost of fabric, and human creativity found new uses for that cheaper fabric, including decorating it with a constantly changing palette of color, cut, and design but also inventing entirely new kinds of uses, from surgical meshes to spacesuits.

Whenever one thing becomes commoditized, something else becomes valuable. As long as we use the productivity gains from technology to create value for society, and ensure that value is widely shared so that customers are able to afford the cornucopia of goods on offer, we will find new ways to put people to work.

Don’t just cut costs. Do more!

I discuss this and other design patterns for prosperity in my forthcoming book, WTF: What’s the Future and Why It’s Up to Us. Pre-order now.

Do More! What Amazon Teaches Us About AI and the “Jobless Future” was originally published in From the WTF? Economy to the Next Economy on Medium, where people are continuing the conversation by highlighting and responding to this story.

How I traced the falsity of one internet meme, and what that teaches us about how an algorithm might do it

I have a brother who is a big Donald Trump fan, and he frequently sends me articles from various right-wing media sources. Last week, he sent me a variant of the image above.

I immediately consulted Snopes, the fact checking site for internet hoaxes, and discovered that it was, as I expected, fake. According to Snopes, these are actually both electoral maps. Per Snopes, “On 11 November 2016, the Facebook page “Subject Politics” published two maps purportedly comparing the results of the 2016 U.S. presidential election with the 2013 crime rate in in the U.S…. The map pictured on the bottom actually shows a 2012 electoral map that was created by Mark Newman from the Department of Physics and Center for the Study of Complex Systems at the University of Michigan.” Snopes was unable to verify the source of the first map, but concluded (presumably by comparing with known electoral maps) that it is in fact an incomplete electoral map from the 2016 election.

Snopes, which uses human editors for fact checking, does a good job, but they can’t find every fake news story. Still, when a reputable fact-checking organization like Snopes or Politifact identifies a story as false, that’s a pretty good sign. And in fact, the fake news story has been taken down on Facebook, presumably in response.

Continuing my research, I used Google to search for other sources that might provide more insight on the relationship between the electoral map and crime rates. I quickly found this 2013 article from Business Insider, Nine Maps That Show How Americans Commit Crime. It shows a very different picture.

Since Business Insider told me the source of the data (the FBI Uniform Crime Report) I could go verify it for myself. Sure enough, the data on the FBI site matched the Business Insider map.

I tell this story of two maps to emphasize that when people are discussing the truth or falsity of news, and the responsibility of sites like Facebook, Google, and Twitter to help identify it, they somehow think that determining “truth” or “falsity” is something that only humans can do. But as this example shows, there are many signals of likely truth or falsity that can be verified algorithmically by a computer, often more quickly and thoroughly than they can be verified by humans:

• Does the story or graph cite any sources? If no sources are given, it is far from certain that the story is false, but the likelihood increases that it should be investigated further. Note how the fake story with which I opened this article provided no sources, and how it was debunked by Snopes by finding the actual sources of the graphs.
• Do the sources actually say what the article claims they say? For example, it would have been entirely possible for Business Insider to claim that the data used in their article was from the FBI, but for there to be no such data there, or for the data there to be different. Few people trace the chain of sources to their origin, like I did. Many propaganda and fake news sites rely on that failure to spread falsity. Checking sources is something that computers are much better at doing than humans.
• Are the sources authoritative? In evaluating search quality over the years, Google has used many techniques. How long has the site been around? How often is it referenced by other sites that have themselves been determined to be reputable? (Google’s PageRank algorithm, which revolutionized internet search, was a variant of scientific citation analysis, where the importance of scientific papers is evaluated by the number of other papers that reference it, and the reputation of the individuals or institutions making those references. Previous search engines had used brute force matching of the words contained in a web page with the words that the user was looking for.) Most people would find the FBI to be an authoritative source. We don’t think about the tacit knowledge that lets us make that determination, and might be surprised that an algorithm lacking that knowledge might still be able to come to the same conclusion by other means. Yet billions of people have come to rely on Google’s algorithms to do just that.
• Do the sources, if any, substantiate the account? If there is a mismatch between the story and its sources, that is a clear signal of falsity. Last week, I wrote about an eye-opening experience with fake news. I’m no Donald Trump fan, so I was prepared to believe the headline I saw on Facebook: Mike Pence Gets ‘Booed Like Crazy’ at ‘Hamilton’. But something quickly became apparent when I watched the actual video embedded in the story: it didn’t match the description given in the article or the headline. As shown in the video, many people cheered Mike Pence as he entered the theater, and the most apparent “Boo” sounded like it was from the person holding the camera. By contrast, in the New York Times story about the same event, the description in the text closely matched the video. Many fake new stories contain jarring discrepancies between the headline and the story or between the story and its sources. Again, this is something that can be detected by a computer program (although comparing text to video may be at the outer edge of today’s capabilities.) Note that the program does not have to find absolute truth; it just has to cast a reasonable doubt, just like a human jury.
• Are there multiple independent accounts of the same story? This is a technique that was long used by human reporters in the days when truth was central to the news. A story, however juicy, would never be reported on the evidence of a single source. (The movie All the President’s Men, about the reporting of the Watergate scandal, made a powerful impression on me as a young man, as have many interactions with first rate reporters in stories that I myself have been involved in over the years since.) The Huffington Post’s ‘Booed Like Crazy’ was a quote from a tweet about the event. How many tweets were there from audience members reporting booing? How many reported cheering, or a mix of cheering and booing? Again, searching for multiple confirming sources is something that computers can do very well.
• If the story references quantitative data, does it do so in a way that is mathematically sound? For example, anyone who has even a little knowledge of statistics will recognize that showing absolute numbers of crime without reference to population density is fundamentally meaningless. Yes, there are more crimes committed by millions of people in New York City or Chicago than by hundreds in an area of rural Montana. That is why the FBI data referenced by the Business Insider article, which normalized crimes per 100,000 people, was inherently more plausible to me than the fake electoral maps that set me off on this particular quest for truth.

Note that when fake news is detected, there are a number of possible ways to respond.

1. The stories can be flagged. For example, Facebook (or Gmail, since much fake news appears to be spread by email) could show an alert, similar to a security alert, that says “This story appears likely to be false. Are you sure you want to share it?” with a link to the reasons why it is suspect, or to a story that debunks it, if that is available.
2. The stories can be given less priority, shown lower down, or less often. Google does this routinely in ranking search results. And while the idea that Facebook should do this has been more controversial, Facebook is already ranking stories, for example featuring those that drive more “engagement” over those that are more recent, and showing “more engaging” stories or stories related to ones we’ve already shared or liked. Once Facebook stopped showing stories in pure timeline order, they put themselves in the position of curating the feed algorithmically. It’s about time they added source verification and other “truth” signals to the algorithm.
3. The stories can be suppressed entirely if certainty is extremely high. We all rely on this level of extreme prejudice every day, since it is what email providers do to filter the email we actually want to see from the billions of spam messages sent every day.

As I wrote in my first article on the topic of fake news, Media in the Age of Algorithms, “The essence of algorithm design is not to eliminate all error, but to make results robust in the face of error.” Much as we stop pandemics by finding infections at their source and keeping them from finding new victims, it isn’t necessary to eliminate all fake news, but only to limit its spread.

Tonight, VP-Elect Mike Pence attended #HamiltonBway. After the show, @BrandonVDixon delivered the following statement on behalf of the show. https://t.co/Jsg9Q1pMZs

The Huffington Post Has a Fake News Problem!

I was really struck by the coverage of the Hamilton cast statement last night. The video is above. Please watch it before reading.

Donald Trump, of course, described the cast’s statement in the worst possible light.

I am sad that Trump’s reaction to a polite and respectful expression of First Amendment rights, a public statement on an important issue, should be to cast it as rudeness. It is even more sad that he has found time to say next to nothing in response to actual rudeness or far worse by his own followers. But I also wonder whether he actually watched the statement, or just read the coverage of it. Of the tens or hundreds of thousands of people sharing and resharing the coverage, I wonder how many actually watched either the statement, or the widely circulated video of the crowd “booing.”

In many ways even worse than Trump’s reaction was the coverage of the event. While I wasn’t there to verify what happened during the entire event, the New York Times did a good job capturing what seems to have actually happened (that is, their account, Mike Pence Saw ‘Hamilton.’ The Cast Had Something To Say, seems to match the video evidence that it describes), but consider this quote from a widely shared piece from The Huffington Post, entitled “Mike Pence Gets ‘Booed Like Crazy’ at ‘Hamilton’.”

There was some clapping and a few stray cheers, but in the video above, many of the theatergoers don’t sound very happy to have Donald Trump’s №2 man in the audience.

Now watch the video they are supposedly describing:

The New York Times,by contrast, describes the video accurately: “When Mr. Pence entered the Richard Rodgers Theater in Manhattan, he was greeted with a mix of clapping and booing, according to theatergoers who posted on Twitter.” (It actually seems to me that the loudest booing may have come from the person making the video.)

Think tactically for a minute. Is this an effective way to protest the policies of this administration? Or are you serving the cause by playing into the hands of those who want to discredit that protest? Report that the crowd booed (and as you can see, they mostly didn’t), and Trump gets to say that the protest was disrespectful.

Every story should have emphasized that the crowd cheered enthusiastically at the message of the play, at the cast’s statement at the end, and at various pointed moments and key lines during the performance. And it should have reported, as the Times did, that Pence himself applauded at key points in the play. As well he should. Let’s hope its message sunk in!

The follow-on coverage was an opportunity to amplify the message of the moment, not to confuse it and turn it back to a story of blind hatred!

At its heart, Hamilton tells the story of the American dream, where an impoverished immigrant can get ahead and make a huge difference to his country “by working a lot harder, by being a lot smarter, by being a self-starter.” It is a tale of ambition, of standing for values, and, as cast and performed, a testament to the multiculturalism that truly makes America great.

I tell this story to emphasize that it isn’t just Facebook that has a fake news problem, and it isn’t just Donald Trump and kids in Macedonia who are using social media to send the news spinning wildly away from the truth. When sites like the Huffington Post post partisan clickbait that is clearly untrue, they deserve to be shunned, not reshared.

In the context of the work that social media sites like Facebook need to do to improve their algorithms (something I wrote about last week in Media in the Age of Algorithms), it isn’t just a matter of determining which stories are true or false. It’s a matter of understanding which sites tell the truth, and which don’t, and lowering the algorithmic encouragement they give to those that fail to tell the truth. This is not that dissimilar to what PageRank and similar Google algorithms do, figuring out which sites are authoritative, and which are ripoffs.

Facebook has argued that they shouldn’t be in the position of determining truth. Others have argued that they need a team of editors who will make the determination. I have argued that they need to improve their algorithms so that they can do this at scale.

It seems to me that the process I went through in evaluating this story is something that can be done by an AI, or frankly, by much simpler algorithms. Does the headline match what is said in the story? Does the story match its sources? For that matter, does it provide sources to be evaluated? These are the kinds of recursive checking that computers are actually better at than most people.

You should be able to see here that “truth” as defined here isn’t a matter of matching a story to reality, but just matching a story to its own retelling. This is, after all, how detectives trip up liars, by looking for inconsistencies in the stories they tell.

“Truthrank” doesn’t have to be perfect to improve on the current situation.

Facebook and Google should be working together on this, and sharing their efforts with sites like the Huffington Post to use in evaluating their own publishing. (Though frankly, HuffPo and other low-quality sites like it will be penalized by an improved algorithm, which will give them incentives to improve on their own.)