CHAPTER THREE

SUBSTITUTIONAL EQUIVALENCES

The Drivers of Societal Phase Change

LET’S BEGIN LOOKING at substitutional equivalences with an example of a structural transformation that might seem on its face to be an unlikely driver of societal phase change, but that nicely illustrates the substitutional equivalences that power it: bank robbery.

Robbing banks at gunpoint is dangerous work that frequently results in long prison sentences or getting shot or killed. And the returns on the investment get lower all the time. Gone are the days when Willie “the Actor” Sutton could rob one hundred banks for a collective haul of $2 million—$30 million in today’s dollars.¹ Because of the use of credit cards, banks no longer keep huge hoards of cash on hand; the average bank robber today makes off with only about $5,000 per heist. But it is still true that it makes sense to rob banks, because, in Sutton’s famous words, “that is where the money is.”² The modern, high-return way to rob a bank is to steal small amounts of money from thousands of customer accounts using computer zombies. The process is managed remotely, preferably from a foreign country that turns a blind eye to cybercrime.

Information Age tools made this possible. First, banks provided millions of customers with relatively insecure systems that allowed them to transfer money in and out of their accounts via the Internet. Then, in the late 2000s, low-cost versions of Zeus malware—a Trojan horse that runs on Microsoft Windows—became available.³ Cybercriminals planted it on bank customers’ computers and used it to collect their account information, track their keystrokes, and discover their passwords.

A crime organization called Citadel set up a “botnet” ring that consisted of 1,500 servers around the world. Those servers, the virtual equivalents of Mafia consiglieri, managed 1.2 million computers on which Zeus-based malware had been installed. Taking a little bit of money at a time, Citadel stole an estimated $500 million.⁴ To stop them, Microsoft engineers joined forces with cyber-trained investigators in the FBI, who coordinated their efforts with eighty international partners. In 2015, Dimitry Belorossov (a.k.a. Rainerfox), a Russian citizen, was arrested in Spain. One of the key technologists behind Citadel,⁵ he was tried, convicted, and sentenced to four-and-a-half years in jail. Setting aside the respective lengths of their prison sentences, Sutton and Belorossov couldn’t be more different, even though they both worked in the same profession.

Information Age tools had reached a high level of sophistication with respect to data capture and manipulation. Simultaneously, the banking environment had created millions of points of attack. A critical point was passed and a new type of bank robbery was created, for which we had few law enforcement techniques to defend ourselves. What we know about armed robbery gives us very little insight into how to deal with cyber-theft.

The rules and the tools had changed; local police were of little or no help. At first, even national and international law enforcement organizations were stymied. Putting an end to this new kind of heist required cyber-trained investigators, help from Microsoft, and the efforts of numerous international partners.

THE POWER OF SUBSTITUTIONAL EQUIVALENCE

Substitutional equivalence is the engine that drives societal phase change. Throughout history, new ideas, new technologies, and new inventions have enabled humanity to construct new forms of institutions, new types of relationships, and new kinds of space that replace former ways of doing things. However, we immediately add that these substitutional equivalents are not the same as what they replace. Typically, they work more efficiently in some manner and often include new features. But they operate using new rules, tools, and processes.

When new methods and forms perform in an almost identical way to what they replace, they are substitutionally equivalent, but a structural transformation has not necessarily happened. For example, if you replace a handwritten ledger with a spreadsheet, you are substituting a more useful and efficient method of bookkeeping for an old one, but a structural transformation has not occurred—the spreadsheet is merely a new medium for the same information and it is put to the same use. The same rules apply. But if the intelligent technology–powered substitution enables the creation of a new institutional form that follows a different set of rules and performs in a different fashion from the old one, then a structural transformation has taken place. Add up enough of them, and you have a social phase change.

Seventy-five years ago, most consumer transactions were paid for with cash. Today, most of us use credit or debit cards. Things are very different when one pays with a credit card instead of cash, for both the buyer and the seller. When you use cash, the merchant gets his money instantly, which he then deposits into his bank. When you pay with a credit card, he gets his money from the bank that is associated with the card. If a cash-using customer comes up short, the merchant doesn’t let her take the product home. If the credit card–using customer is short on cash, the merchant doesn’t have to know or care. Why? Because unless the customer is committing credit card fraud, the merchant knows he will receive his money.

Cash is anonymous and, unless the police have arranged to give you marked bills, it is virtually untraceable. When you use a credit card, you give up a great deal of privacy, because the system can track your spending. Though the buying power of cash may change with the broader economy, when you use a dollar to buy something that costs a dollar, you pay a dollar for it, and the merchant receives a dollar. When you buy that same item with a credit card, you may opt to borrow the money and pay interest on it. The merchant may receive a little less than a dollar because the credit card system charges him a fee for its services. Using a credit card is different from paying with cash, but we use it to perform the same functions. Indeed, the credit card has rendered cash obsolete in many ways. In that sense, it is substitutionally equivalent.

To repeat the list from chapter 1, three types of substitutional equivalences are powering the Autonomous Revolution:

1. Information Equivalence. In which the information underpinning an institution can be used to restructure that institution

2. Intelligence Equivalence. In which traditionally human acts and capabilities, both intellectual and manual, can be replaced by computer-based operations

3. Spatial Equivalence. In which activities that are traditionally based in the real world, such as social interactions, commerce, entertainment, meetings, even travel, now operate in the virtual world

Now let’s take an in-depth look at each of them.

INFORMATION EQUIVALENCE

Intelligent machines have created numerous forms of substitutional equivalence.

The effects of the structural transformations that we are experiencing in our business and economic institutions and in our work and home environments are primarily being driven by information and intelligence equivalences. Information equivalence is so prevalent in our institutions and affects such a large swath of our economy that we will begin there.

Our institutions are composed of information and physical processes. For example, a farm has a very large physical component and a fairly small information component, whereas a bank has a very large information component and a relatively small physical one. In the most extreme case, a modern social network such as Facebook manages more than 2 billion users on six continents with just 4,000 or so employees, who are based in a handful of buildings. When you add intelligent machines to institutions where the information component is dominant, dramatic changes take place. The plumbing profession will likely stay pretty much the same throughout the Autonomous Revolution, but retail banks (which are really information systems in disguise—“information proxies”) will change beyond recognition.

Regarding the phrase information proxy, think of an office building and its contents—desks, paper files, computers, networks, and people. This structure is really an information storage and processing network. Information is stored on computer and paper files and in people’s brains. Information is exchanged over a digital network, by moving paper files from desk to desk, and in face-to-face conversations. Intelligent machines and people process it.

If an information system can gather this information by itself, store it in a computer-readable form, process it using simple rules or via artificial intelligence algorithms, and respond without human involvement, then the office building and its contents can be replaced by a rack of computer gear. There is no need for people, file cabinets, desks, or more than a few square feet of physical space. As mentioned before, many customer call centers today already fit this description.

The physical structure, its contents, and its people are information proxies. The substitional equivalent is a rack of computer gear.

A retail store is an information proxy combined with a physical delivery system. When consumers go to Walmart, they engage in a number of information exchange processes. Customers find out what items are available and at what prices. They may want to inspect the merchandise to determine its quality and try it on to make sure it is the right size. When customers make a purchase, they pay cash or provide credit card information. Then they carry the merchandise to their car to complete the physical delivery process.

Retail shopping is being replaced by informational and spatial equivalences. Instead of going to the store to shop, cyber-shoppers get product, price, and availability information online. They can even try on an item of clothing virtually, using digital imaging. The computer takes their credit card information (an intelligence equivalence of what a flesh-and-blood retail clerk does), and FedEx or UPS provides delivery.

For many commodities, the retail substitutional equivalence of the future will consist of shopping online with rapid delivery by autonomous vehicles from large warehouses located near urban centers. The same functions will be carried out, but will use entirely different tools, rules, and processes.

Surprisingly, many institutions that do not appear to be heavily dependent on information content for their success are, in fact, quite vulnerable to disruption by information equivalent infrastructures. For example, on the surface, the hospitality industry would seem to be dominated by physical infrastructure. But the industry depends on two very important information components: the idea of a hotel and brand awareness. In the past, when travelers thought about where they would stay while visiting a distant city, the mental model that they called up was a hotel. The second important information component had to do with trust. Consumers assumed that if they stayed in a branded hotel, they would experience a certain level of quality. The Airbnb model has changed the rules and tools by substituting a different set of information equivalents for those assumptions. Now travelers think about staying in the homes of total strangers. And they trust a computer system—Airbnb—to find them one that is clean, safe, convenient, and fairly priced.

The extent of the transformation that takes place as a result of information equivalence depends, first, on the amount of functionality that can move from one information equivalent structure to another. Second, the new information equivalent infrastructure must offer a significant perceived benefit. And third, the technology has to be in broad enough use that a large portion of the user base can move to the equivalent information infrastructure. For example, before a virtual currency can achieve broad use, large numbers of merchants have to be willing to accept it as a form of payment.

While tomorrow’s restaurants, plumbers, and landscapers will promote their businesses via social media, bill their customers electronically, and make use of digital tools in their operations, the essence of their businesses will likely remain the same. But some kinds of institutions are almost totally information proxies in disguise. Retail, banking, finance, and monetary systems are examples of institutions with extremely high information proxy content. One would expect them to be significantly transformed.

INTELLIGENCE EQUIVALENCE

Advances in artificial intelligence, deep learning, neural network processing, and big data have unleashed the forces of intelligence equivalence. Machines are now capable of intelligent behavior. In many applications, they can substitute for humans’ brains, minds, and senses.

For more than one hundred years, technologists involved in computation have speculated about and attempted to construct machines that would exhibit intelligent behavior. In 1914, the Spanish engineer Leonardo Torres y Quevedo demonstrated a mechanical device that could play simple king rook chess endgames.⁶

In 1921, the Czech author Karel Capek wrote R.U.R. (Rossum’s Universal Robots), which introduced the word robot to the world.⁷ Reading this prophetic play, written almost one hundred years ago, is a startling experience. Its disillusioned workers, displaced by robots, could equally well be members of today’s middle class.

In 1950, Alan Turing, one of the early investigators of machine intelligence, proposed a simple test to determine whether a machine could “think.” Now known as the Turing Test, it is a protocol in which three terminals are set up in isolation from one another, two operated by humans and one by a computer. One of the humans asks the computer and the other human a series of questions. If the questioner can’t tell which respondent is human and which is a machine after a certain number of tries, then the computer is said to have intelligence. By 1966, Joseph Weizenbaum, author Davidow’s first boss, had developed a program called ELIZA that appeared to pass the test.⁸

In the nearly seventy years that have passed since the creation of the Turing Test, artificial intelligence has passed through cycles of excitement and disillusionment. At a 1956 Dartmouth conference, where the term artificial intelligence was coined, Marvin Minsky predicted that the problem would be solved within a generation.⁹ He was wrong. Most early attempts to mimic intelligent behavior were based on systems that ran according to more and more complex sets of rules. But it turns out that it is both very difficult and very expensive to write rules that can cover every possible situation that a computer might find itself in. A real breakthrough would occur if machines could learn from their experiences and then reprogram themselves, without human assistance. This is precisely what has been happening in the past few decades.

In 1997, Deep Blue, a chess-playing computer developed by IBM, beat the Russian grandmaster Garry Kasparov in a six-game match.¹⁰ Kasparov said that he had sensed a thinking presence inside his computer opponent. Then, in 2016, Google DeepMind’s artificial-intelligence program, AlphaGo, defeated Lee Sedol, a Go champion, 4–1. Go is a more difficult game for a computer to play than chess, and AlphaGo’s victory is perhaps the best harbinger of what is to come. While Deep Blue relied on hard-coded functions written by human experts for its decision-making processes, AlphaGo used neural networks and reinforcement learning. In other words, its system studied numerous games and played games against itself so it could write its own rules.¹¹

The lesson here is that it is now possible to use inexpensive computer power to develop intelligent processes. Until recently, intelligent systems learned how to behave in specific situations. When presented with a new situation, they would have to learn how to behave from scratch. Being able to generalize a response based on past experience comes very close to cognitive behavior—a once-impossible barrier that now appears to be crumbling. The Economist recently reported that certain computers, trained to play a number of games, have been able to come up with viable strategies for playing different games they have never seen before.¹²

The semiconductor industry has only just begun to unleash the power of Moore’s Law (the regular doubling of semiconductor chip performance) on neural networks. Google’s DeepMind system used Nvidia’s newly announced P100 chip, containing 1.5 billion transistors, to power its system. The chip enabled Google to build neural networks that were five times deeper than in the past—and the deeper the networks, the more intelligent the behavior.¹³ The key point here is that neural network systems will benefit from high rates of progress in the semiconductor industry. In the years to come, each generation of machines will be cheaper, faster, and more capable than the one that came before.

A second key advance is the amount of data we have at our disposal. The numbers of data sets that can be used to train new systems will grow exponentially.

Finally, we are continually learning how to design better software.

The implications of all of this intellectual equivalence are staggering. Financial firms have hundreds of routine processes that can be automated. So do many other businesses. For example, WorkFusion makes a software that studies what employees do when they manipulate data on their computers and then assesses which tasks can be automated and which can be outsourced. Though its ostensible purpose is to help workers become more productive, over time, it’s easy to imagine how this process will replace them. When this happens, human work will have been absorbed into a workerless world—an archetypal example of intelligence equivalence.

Workers whose jobs are being transferred overseas have often complained that their companies force them to train their replacements. With services such as WorkFusion, employees are training invisible robots to replace them—an indignity that is no less infuriating for being invisible.

Machines are only going to get smarter. Suppose for a second that today’s intelligent machines are only capable of doing the work of a person of average intelligence, someone with an IQ of 100. Then imagine that, thanks to Moore’s Law, the technology in those machines improves by 40 to 60 percent per year. Suppose that this rate of technological progress raises the machine-equivalent IQ by 1 point per year. In a decade, those machines would have a machine-equivalent IQ that would empower them to do the work of more than 75 percent of the U.S. population.¹⁴

In fact, machines with the equivalent of 110 IQ points are here already. In certain applications, even the minds of highly educated doctors are no longer needed. In 2013, the FDA approved Johnson & Johnson’s Sedasys machine (since discontinued), which delivered sedatives to patients automatically, eliminating the need for an anesthesiologist.¹⁵ An emerging field in radiology is computer-aided diagnosis (CADx).¹⁶ A recent study published by the Royal Society showed that computers performed more consistently and accurately in identifying radiolucency (the appearance of dark images) than radiologists by almost a factor of 10.¹⁷

In 2014, the authors of this book took to the pages of the Harvard Business Review to warn about the growing numbers of Zero Economic Value citizens (ZEVs)—many of them possessing high skill levels—who will never find meaningful employment again, because they will keep losing the race with robot/artificial intelligence. We’ll discuss this sobering challenge in more depth later. For now, it’s important to know that many systems in use in the business world already incorporate both information and intelligence equivalent processes. Larger and larger numbers of worker-less environments will be the inevitable result.

SPATIAL EQUIVALENCE

Spatial equivalence is familiar to all of us. One form of it is social networking: Facebook, Instagram, Twitter, LinkedIn, Reddit, YouTube, and many more platforms now claim billions of users. Facebook alone has around 2 billion users worldwide. China has some 597 million social media users.¹⁸

Social networks—which we’ll discuss in chapter 8—provide a new way to organize affinity groups and an inexpensive way to communicate with them. But they have also transformed the social contract in often disturbing ways—such as the surrender of privacy and the proliferation of false information.

A second major form of spatial equivalence is online gaming, with its hundreds of millions of participants. It has largely replaced not only traditional board games but even, in extreme cases, outdoor recreation. It has brought with it a unique set of challenges, including compulsive and addiction-like behaviors. The power of these games will only increase with the rise of virtual reality, 3-D imagery, and more powerful computation engines. Gaming is also spreading its influence into other forms of recreation, increasingly bringing group participation, virtualization, and 3-D imagery to the rest of the Web, television, and film.

Already we are constantly interacting with virtual environments. While meeting someone in virtual space has become the social equivalent of meeting someone for a cup of coffee, the experience is not, in fact, the same. Moreover, human interactive experience must exist in time—and the amount of our time is necessarily limited. Thus, as our time in virtual space increases, the time we spend in the real world must decrease.

Many businesses now function primarily in virtual space. Google, Spotify, and Netflix are examples. When those businesses did not exist, some form of their services existed in physical space. People read printed newspapers, went to record stores to buy music, and to Blockbuster and other video stores to buy or rent movies. Before VCRs became common, they went to the movies.

The implications of these shifts are still not fully known. Even the rules of this new world are, as yet, only partially written. In the meantime, the lure of the virtual world—and the risk of severe and unintended social consequences that our immersion in it might bring—only increases. We will also look more deeply into this matter in chapter 8.

Substitutional equivalences put pressure on every corner of modern life. Understanding how they operate will be vital to determining how we should respond. For that reason, in the pages that follow, as we consider the various cultural aspects of the Autonomous Revolution, we will refer to substitutional equivalences regularly and will provide examples of how to respond to the challenges they present.