Crystallized Imagination

What Information Theory, Genetics, Thermodynamics, and Networks, can teach us about what the economy is and how it evolves.

By Cesar A. Hidalgo

To Iris, Anna and Mridu

PART I: The Complexity of Dreams

“We talk much of imagination. We talk of the imagination of poets, the imagination of artists, and I am inclined to think that in general we don’t know very much exactly what we are talking about.

It is [that] which penetrates into the unseen worlds around us, the worlds of science. It is that which feels and discovers what is, the real which we see not which exists not for our senses.”

Ada Lovelace

“Imagination: Faculty of the mind which forms and manipulates images. mid 14.c”Online etymology dictionary

“Imagination will often carry us to worlds that never were. But without it we go nowhere.”

Carl Sagan

“This world is but a canvas to our imagination.”

Henry David Thoreau

Chapter I: The Secret to Time Travel

The chair where I waited for my daughter’s delivery was not that comfortable. My wife Anna and I had arrived at MGH[footnoteRef:1] at 6:30pm that Saturday. We had stayed at home waiting for her contractions to evolve and decided to hit the hospital when these were only a few minutes apart. Her contractions intensified when she was in triage, but the epidural she received a couple of hours later brought her the numbness she needed to rest. It was now 2am. The night was peaceful. All that we could hear was the infrequent noise of the pump that inflated her blood pressure monitor. The room was lit by the lights of a few monitors and by the streetlights bouncing off the Charles River. In that dim light all that I could see was Anna resting peacefully in her bed. I held her hand as I waited for the delivery of my daughter in a plasticky armchair, that as I told you, was not that comfortable. [1: Massachusetts General Hospital]

At 3am the nurse and doctors told us that it was time for Anna to start pushing. Anna was fully dilated and Iris—my daughter—was making one of the most important trips of her life. It took only 26 minutes for Anna to push Iris into the hands of the focused but nervous medical student that received her. Twenty-six minutes sounds like a short time for delivery, and it is. Yet, I will argue that the trip that Iris made that night was not a 26-minute trip down a few inches of birth canal, but an eighty thousand year trip from a distant past to an alien future. In twenty-six minutes Iris travelled from the ancientness of her mother’s womb to the modernity of the 21st century society. Birth is in essence, time travel.

Up until that night, with the exception of a few sounds, Iris had experienced a world that was no different from the one experienced by babies eighty thousand years ago. She had been carried inside her mother’s womb hearing mostly the voice of her parents. During a period of 40 weeks Iris developed oblivious to the complexity of the world that buzzed around her. That obliviousness changed that night.

Iris was born at 3:26am in a room that was not illuminated by sunbeams, but by fluorescent and incandescent light bulbs. Her paternal grandparents, who were anxiously waiting to hear about her delivery, saw her face for the first time through an email attachment. The music that filled the delivery room minutes after Iris was born[footnoteRef:2] did not come from the birds or the trees, but from the speakers of a tablet computer that obeyed the orders of an algorithm that chose a song for us. Iris’ trip that night was only a few inches long and lasted a few minutes, but in a deep sense, it was an eighty thousand yearlong journey. That night, she traveled from a distant past and into a present that was literally fantastic. [2: Raindrops, by Cillo. It played in Pandora and I thumbed up the song that night.]

Iris’ trip is special to me, although this form of time travel is not uncommon. Being born in the 21st century is an alien experience for most babies. The 21st century is a world that is quite different from the one where our species evolved. It is a surreal world populated by objects that were dreamed before they were constructed.

The delivery room where Iris performed her journey was full of tangible objects. Yet, the tangibility of the beds, pillows, scales and monitors that populated Iris’ delivery room was not what made the destination of her trip modern. The tactile nature of the reality embodied in mundane objects, such as blankets and night tables, can mislead people to believe that our world is made mostly of atoms, when instead, it is made of a much more mysterious intangible. That intangible is information.

During the last million years the atoms that populate the earth have not changed, but the world has changed. The difference between the world where Iris was born and the world of early hominids, does not reside in the physicality of matter but in the way in which matter is arranged. That physical order is information. Iris’ night delivery was not facilitated by objects, but by the information embodied in these objects. Her night delivery was not illuminated merely by light bulbs, but by the understanding of electricity, energy, and materials that is embodied in light bulbs. Iris was not kept warm that night by a random collection of threads, but by blankets that weaved together matter, knowledge and imagination. Paradoxically, Iris was born in a non-fictional world that although tangible, is made of fiction. This is a world that was not different from the one where she evolved, except by the way in which matter was arranged.

The fact that the objects that populate our world are made of information, and imagination, may seem obvious. Information is as old as life itself. After all, the replication of information rich molecules, such as DNA and RNA, is not the replication of matter, but the replication of the information that is embodied in matter. Yet, what is different between us and other species is that we have developed the ability to encode large volumes of information outside our bodies. Unlike ants, who can excrete a few predefined messages as pheromones, humans can encode information in the form of objects, whether these are the books we read or the physical and digital artifacts that populate our world. The information that we encode in objects, however, is not haphazard. It is primarily information that was begot in our minds. So our ability to build the houses that we imagine, and the software that we dream about, is a simple answer to the eternal question: what is the difference between us—humans—and all other species? The answer is that we—humans—are able to create physical instantiations of the objects we imagine, while other species are stuck with nature’s inventory.

In the next pages I will argue that the process of economic development is nothing more than the process by which our civilization develops its capacity to accumulate physically embodied information and the capacity to process information—which I define as knowledge. This is the information that is embodied in the physical and digital objects that we make, and in the human networks that allow us to create these objects. In fact, I will argue that our ability to create objects that are made primarily of information is what endows our species with fantastic capacities. Our ability to arrange atoms according to our imagination is what allowed a few of us to walk on the moon, and it is what allows many of us to enjoy long distance communication, food refrigeration, and long distance travel.

The ability of ancient civilizations to construct objects plays a central role in the way in which we understand history and the development of human civilization. Our understanding of the civilizations that predate writing is derived largely from the artifacts they left behind, since these long predate written records[footnoteRef:3]. In fact, our ability to encode information in objects predates the invention of writing by at least sixty thousand years[footnoteRef:4], and that of cave paintings by at least thirty thousand years. The development of human civilization is the development of our ability to embody information in objects, and it is hence, a process that begun with the creation of simple concoctions of bits and atoms, as simple as arrows and spears. [3: Dating the origins of human language is difficult, since oral expressions largely predate writing. The earliest records of written language date back about 8,000 years, so archeologists wanting to estimate the origins of spoken language need to consult other forms of evidence, such as the complexity of the products found in the archeological record. The idea that complex products can be used to date the origins of human language is based on two lines of argument. First, an individual able to produce a complex tool, such as a spear or a carburetor, is an individual that understands how different parts fit together, much in the same way that human languages allow us to combine different words into sentences, and sentences into narratives. In other words, individuals that can build complex tools are likely to be individuals that have an internal way of representing each of the parts involved in the construction of the tool, and can think about the sequences of actions required to put these pieces together. The mental process required to ideate and assemble a complex product can be thought of as a primitive grammar, and it has been found to define similar patterns of brain activity. Making complex products is enabled by a similar combinatorial capacity than the one present in human languages. So, even though the combinatorial capacity of product creation does not necessarily imply the use of a human language, it is reasonable to assume that these two combinatorial cognitive capacities emerged together.The second line of argument supporting the use of complex objects to date the origins of language relates to the diffusion of the knowledge required to make complex objects, such as arrows and spears. The proliferation of arrows and spears can be seen as an indication of the existence of early forms of human language, since learning how to build an arrow is different than learning how to use a rock to crack open a walnut. Simple tasks can be learned easily through observation and imitation, while the production of complex objects involve nuances that are communicated much more effectively among individuals that share a language. For instance, individuals sharing a language can more effectively learn how to safely handle the poison used to make arrowheads, or how to fasten a stone ax to a wooden handle. Current archeological records show that Homo sapiens were fashioning tools that were as complex as that of some modern hunter-gatherers as far back as 100,000 to 70,000 years ago. This shows that our ability to crystallize imagination largely predates our ability to write about it, and date the origins of language, and complex products, to the time before we left Africa. This suggests that it is not the use of simple tools what separated our great ancestors from other species, but their ability to create complex objects, objects that were superior to those found naturally in the environment. (Reference EO Wilson, The Social Conquest of Earth, and Yuval Harari, From Animals to Gods)] [4: EO Wilson The Social Conquest of Earth, Yuval Harari, From Animals to Gods.]

Although the ability of people to make products is central to our understanding of human civilization, the ability of countries to make products of increasing complexity is not a central theme in our understanding of economic development. When we look back in history we talk about civilizations that learned to dominate Stone, Bronze[footnoteRef:5] and Iron. These are civilizations that used their dominion over these materials to fabricate weaponry, jewelry and household items. Twenty first century nations are not separated by their dominion over simple metals, but still differ largely in their ability to build physical and digital products. These are no longer bows and arrows, but engines, generators, synthetic materials, video games, and mobile phones, among a myriad of other products. [5: Bronze Age civilizations, separated by long stretches of space and time, had remarkable similarities. Both, ancient Egyptians and Aztecs dedicated much of their efforts to the creation of pyramids that were mostly solid. Both ancient Egyptians and Aztecs had a two-calendar system involving a religious calendar and an operational calendar that regulated activities such as harvesting and tax collection. Both Aztecs and Egyptians had pictographic writing systems that where augmented with phonograms—pictures that represented sounds. And certainly, both ancient Egyptians and Aztecs knew how to smelt copper and tin to make bronze, but were unable to crystallize their ideas in Iron.]

Our world is marked by differences in the ability of countries to make products, and hence, it is marked by difference in the ability of countries to embody information into matter. This puts the few countries that are more successful at embodying information at the cutting edge of modern civilization. These are the countries that are able to design and create almost every product that we have imagined, and by doing so, get to enjoy most of the wonders of modernity[footnoteRef:6]. [6: Certainly, one could argue that not all products increase welfare. Think of an atomic bomb, for example. This line of criticism, however, can be misleading because it fails to separate the use of a technology with the technology itself. In fact, most technologies are double edge swords. Knives can be used to make a tomato salad, or to kill a fellow human being. A wrench can be used to repair a car, or to knock someone unconscious. A pillow can be used to rest at night, or to smother someone to death. The distinction between the uses of technologies, which can be positive or negative, and their ability to open new possibilities, which is a positivistic statement about the space of what is possible, is well exemplified in a quote that Richard Feynman liked to repeat when talking about nuclear energy. This is: “To every man is given the key to the gates of heaven. The same key opens the gates of hell.”]

Yet, most countries are not at the cutting edge and witness the benefits of our species’ ability to embody information mostly from the sidelines. These countries get to enjoy the benefits of the information that is embodied in these products mostly when they are lucky enough to have valuable rocks or mineral fuels that they can trade for them. When that is not the case, their inclusion into the global economy is marginal.

Our evolving ability to make products is at the core of the process of economic development. Yet, as I mentioned, neither our ability to make products nor the difficulties involved in developing this ability are at the core of the practice of economic development. During the twentieth century the practice of economic development went through two big panaceas. Both panaceas boomed and failed. The first panacea was inspired by the Marshall plan, which was the plan through which the United States supported the reconstruction of Germany after WWII. The Marshall Plan was wrongly interpreted as evidence that financial aid was all that was needed to make a prosperous country out of a pile of rubble[footnoteRef:7]. The overestimation of the Marshall Plan’s effects gave rise to a first panacea involving the creation of institutions that lent money to developing countries with the hope that the removal of financial constraints would help developing countries advance their economies. The proposal was even rosier, since the hope was not only to jump start development through finance, but also to do so at a profit for those willing to risk the funds. Cleary, money wasn’t enough and development through loans failed to reproduce what pundits thought the Marshall Plan had accomplished. [7: Collier the Bottom Billion (CITE, page)]

Later in the twentieth century the practice of development went through a different panacea. This second panacea involved formal institutions. These are the institutions created with the intention of governing human behavior, such as political systems, and the laws that governments impose on their citizens.

The institutional debate of the second half of the twentieth century was colored largely by the tension between the different institutional models that were in conflict during the cold war. Countries wanting to prosper needed to adopt the institutions of the already prosperous western nations: open up to trade, privatize inefficient public sector industries, provide autonomy to their central banks, allow their currencies to float, and elect their governments democratically.

The big experiment of this second panacea was the dissolution of the Soviet Union. As the Soviet Union collapsed, western powers pushed for what they called a “shock therapy”[footnoteRef:8]. A shock therapy implied a quick reformation of a country’s formal institutions that mimicked the institutions that had evolved over centuries in the west. This shock therapy neglected the strength of informal social institutions, which are the inherited social behaviors that are not enforced by law. The naïve assumption was that social institutions changed as the incentives provided by formal institutions changed. Yet, through repeated failures we have learned that history is more stubborn than we originally thought. Markets and democracies are not strong enough to produce virtuous social institutions, but virtuous social institutions are strong to produce virtuous markets and democracies. [8: Jeffrey Sacks, The End of Poverty]

As a case in point consider the collapse of the Soviet Union. This was seen by some as the perfect opportunity to show the world that a drastic change in a country’s formal institutions could put a developing nation in the right track. As the iron curtain melted, prefabricated laws and institutions where imported. The imported institutions, however, failed to produce the changes that pundits expected. Almost a quarter century has now passed and it is hard to argue that the post-soviet Russia that has been ransacked by private oligarchs, and its democratically elected officials, is in a better shape than the soviet republic of Russia. If we consider communist Russia as the proof that communism and central planning did not work, then, we should consider post-communist Russia as the proof that neither “democracy”[footnoteRef:9] nor capitalism work either. [9: One might argue that what Russia has at this time is not necessarily democracy, so it is important to take this comment with a pinch of salt.]

If there is any consensus left in the field of economic development, it is that economic development is a complex process that we don’t fully understand. Yet the lack of consensus in the field does not imply a lack of new theories and explanations. During the last decade a new stream of thought, focused on the ability of countries to make complex products has begun to timidly tickle into the economic development mainstream[footnoteRef:10]. Much like archeologists, who focus on the ability of a civilization to make artifacts, this stream of thought focused on the ability of a country to make products. These are the products that embody the modernity of the destination of Iris’s time travel, the products that pack increasing amounts of information. An umbrella term for this emerging body of literature is economic complexity[footnoteRef:11], since this is the stream of thought that focuses on the evolution of a country’s ability to make complex products. [10: The development of this literature started with work that questioned some assumptions that although naïve, had been engrained in economics. In Economic Development as Self-Discovery (2003), Hausmann and Rodrik call attention to the fact that for entrepreneurs to be able to produce a product or provide a service, they need to first discover the cost of producing at a given location. These discovery costs, where costs that were absent from mainstream economic models. In “What You Export Matters” (2007), Hausmann, Hwang and Rodrik classified products into “rich” or “poor” country goods based on the level of income of the countries that produced them. They used this continuous metric to show that poor countries producing “rich” country goods exhibit higher levels of aggregate economic growth (GDP per capita growth). In “The Product Space Conditions the Development of Nations” (2007), Hidalgo et al. showed that a country’s future exports were more likely to be products that were similar to the ones a country already exported. This showed that the ability of countries to upgrade their productive structure was highly constrained by the mix of products that a country was currently making. Finally, in The Building Blocks of Economic Complexity (2009), Hidalgo and Hausmann showed that information about income was not needed to explain future growth, since most of the relevant information was contained in the structure of the network connecting countries to the products they export (see also Hidalgo (2009)). The elimination of measures of income provided a change of language that eschew away from terms such as rich country goods and poor country goods by focusing on the complexity of products instead. Much of this book will be dedicated to explore what the complexity of products means.] [11: This new stream of literature has also been called New Structural Economics, although I prefer to avoid the “new” in the branding since at some point, it won’t be new. The use of the word “new” to name economic sub-disciplines is often conventional, and devoid of poetry and explanatory power. In some cases, it is used to the extreme. Economics has trade theory, a new trade they, and recently, a new new trade theory. ]

But why would anyone want to create a theory of economic development by looking at the products that people make, instead of the people that make them? Isn’t that backwards?

There are pragmatic and philosophical reasons that support the creation of a theory of economic development that hinges primarily on the ability of countries to make products—or create physical packets of embodied information. First, we will explore the pragmatic reasons, which hinge on practical considerations such as the availability and quality of data. Unlike measures of development based on surveys, or formal institutions, the production and commercialization of complex products represents data on economic development that is harder to cheat or forge. Countries can change their rules on paper with relative ease, and improve their scores in indicators based on surveys and formal institutions, such as those produced by the World Bank’s Doing Business Report[footnoteRef:12]. Yet, it is hard for countries to cheat on indicators based on the products that other countries report as imports. Becoming a competitive exporter of sophisticated products, such as medical imaging devices or helicopters, is not easy, and represents hard data on the ability of a country to pack knowledge and information into the products they make. Much like the transformation of ancient civilizations from Stone to Bronze, and from Bronze to Iron, the transformation of an economy from the export of basic commodities to high tech software and hardware tells us something deep about its people. This brings us to the second pragmatic reason why creating a view of economic development based on products is informative: which is the indirect information about an economy that data on products can provide. [12: http://www.doingbusiness.org/methodology (accessed October 7, 2013)]

The ability of countries to make complex products has been marginal, yet present in past theories of economic growth and development. As we review in detail in part III, past theories of economic development were constructed on the accumulation of coarse factors, such as physical capital, human capital and social capital. Certainly, the assumption was that these factors help countries produce economic activities that are more sophisticated, or that add more value. Yet, even though these theories have been useful to explain a number of phenomena, their focus on factors that are considered inputs, rather than the products that emerge as an economy’s output, constraints their empirical validation. Data on inputs factors is often much noisier and incomplete than the data on outputs that comes from industrial production. As a result, the validation of these theories has hinged on imperfect proxies of the real world, such as surveys to measure social capital and metrics on the years of schooling of a population to measure human capital. Certainly, what people answer on a survey, or the amount of time a child spent in a building labeled “school”, are poor reflections of the complexity of our reality. There should be better ways to capture these important aspects of the world and looking at the industrial output of a country is a good alternative.

In general, the ability of countries to make complex products gives us indirect information on processes that are otherwise hard to measure, including the formal and informal institutions of a population and the specific human capital they posses. As we will see, making complex products is difficult, since it requires the accumulation of vast amounts of productive knowledge, which is highly specific[footnoteRef:13] and that lives embodied in people and networks of people. The accumulation of productive knowledge is limited by the ability of people to form the networks where this knowledge is accumulated—such as firms and markets. If the ability of people to form these networks, and pack knowledge into them, is constrained by factors such as formal and informal institutions[footnoteRef:14], education, civil liberties, or infrastructure, then the ability of a country to make complex products will inform us about the quality and availability of these factors.[footnoteRef:15]. Hence, if we are uncomfortable with measures of development that hinge in the complexity of products, we can still consider them to be indicative of other social and economic processes that we might consider relevant but are hard to measure directly. [13: As we mentioned in the introduction, the idea that factors are more specific than what is usually assumed is an idea that is already present repeatedly in the writings of Wassily Leontief and Michael Porter.] [14: And it is. We will review this literature in part II] [15: In his book Trust, Francis Fukuyama uses this connection by linking social capital to the size of firms present in a country. For those interested in his views I suggest reading Chapter 3 of Trust: Scale and Trust.]

A third pragmatic reason to create a description of economic development centered on the complexity of the products that countries are able to make comes from the experience of East Asian economies. The rapid development of Japan, Singapore, Taiwan, South Korea and China, highlights the link between a country’s ability to make and export complex products and its ability to bootstrap itself out of poverty. Later, we will argue that the ability of countries to make complex products is in essence economic development since—among other things—it precedes economic growth. For the time being, however, it suffices to call attention simply to the obvious miracle represented by the rapid development of East Asian countries, and in particular, to the rapid development of China[footnoteRef:16]. [16: Rodrik, What’s so special about chinese exports? (2005)]

The case of China is particularly vexing for the western pundits that grew convinced of the need of a modern democracy as a pre-requisite for economic development[footnoteRef:17]. While I agree that civil and political liberties are important aspects of human development[footnoteRef:18], it is clear that the development of China, although lacking important civil liberties, has removed other important deprivations from the homes of millions of Chinese people. The industrial development of China, which hinges largely on a superb capacity to manufacture products, now at all levels of complexity, is a good example of the importance of developing the ability to make complex products. [17: It is also a problematic case for Francis Fukuyama, since in the mid 90’s he discounted China’s ability to sustain high levels of growth based on what he perceived as weak social institutions: a familial society with low levels of trust.] [18: As defined by Amartya Sen]

Yet, these pragmatic reasons do not represent the main motivation to create a description of the process of economic development centered on the ability of countries to make products. The main reason to focus on the creation of products, if we take the meaning of products in its broadest sense, is that by focusing on products we can create a theory of economic development centered on knowledge and information. Creating a description of economic development based on products is creating a description of the process of economic development based on the ability of our society to pack and unpack information into the physical bundles we call products. It is the creation of a view of the process of economic development that is based on the accumulation of information in its many forms, the crystallized forms that are represented by products, and the dynamic forms that are represented by the networks of humans that hold the knowledge required to make these products[footnoteRef:19]. So by focusing on products we are focusing on the accumulation of physically embodied knowledge and information, instead of more traditional forms of economic value. This is the accumulation of the knowledge embodied in the bodies of surgeons and of the machines that help them, the knowledge embodied on the minds of teachers and on the textbooks and software they use for their lessons. This is the knowledge embodied in the minds of film directors, fireman, and bakers, and the information embodied in the movie cameras, fire engines, and ovens they use. The evolution and development of economies is nothing other than the evolution and development of our species ability to accumulate and use information. This is the evolution of our ability to pack information into products, and to unpack information into the knowledge that resides in the networks of people that make these products. It is nothing more than the continuation of the most fundamental of all dualities: the duality between matter and information. [19: As a historical footnote it is worth noting that Marx also used the word crystallization to describe products in some of his writings: “A commodity has a value because it is a crystallization of social labor. The greatness of its value, or its relative value, depends upon the greater or less amount of that social substance contained in it; that is to say, on the relative mass of labor necessary for its production” (MARX in Value, Price, and Profit). For the reader that might be to quick to use this coincidence to make inferences of my political affiliation, or motives, I must say that my use of the word crystallization is not an endorsement of Marx, or a signal of my political affiliation. In fact, I learned about this quote at a time in which this book was almost completed, and I added it to make the historical linkages presented more comprehensive. For the reader that would like to learn about the difference between Marx’s use of the word crystallization to describe products, and mine, is that I see products as the crystallization of information, rather than social labor. This distinction should become clear after reading part I.]

In the following pages I will construct a description of the process of economic development based on the ability of countries to make products that embody the practical uses of knowledge and information. My exposition of this topic will be divided in three parts. The first part is an attempt to create a description of products based on their ability to carry information and the practical uses of human knowledge. The second part of the book returns its gaze to humans, and mainly, to the constraints that limit the ability of humans to accumulate the knowledge they need to make complex products. The third part of the book will use the description of products, and of the networks of people that create them, to expand current theories of economic growth and development.

Finally, for those readers that are interested in things other than economies, I will show that a theory of economic development based on the ability of countries to pack information into products will help unify the process of economic development with other natural processes, such as the development and evolution of biological organisms and statistical physics. Personally, I find this to be the most exciting stop in our journey, since I find there is grandeur in a view of the world that unifies different branches of science. As you will see in the next pages, the connections between biology, statistical physics, information theory and economies are not simply metaphorical. The connections are deep, and are related mainly to the discreteness of the networks that are needed to accumulate knowledge, and the constraints that this discreteness imposes in the ability of countries and organisms to accumulate knowledge and information.

Chapter II: The Physicality of Information

“To invent, you need a good imagination and a pile of junk.”

Thomas A. Edison

A few months ago an article on the front page of a Chilean newspaper business’ section caught my eye. The article talked about a Chilean having bought the world’s most expensive car. The car, a Bugatti Veyron, had a sticker price of more than 2.5 million dollars, and represented one of the most flamboyant acts of conspicuous consumption I have ever seen.

After a quick web search I estimated the per-kilo price of the car, which turned out to be roughly $1,300 U.S. dollars, or $600 dollars a pound[footnoteRef:20]. To put this into context, we can look at the per-kilo price of gold and silver. Depending on the day, the price of a kilo of pure silver is about $1,000, while that of a kilo of gold is around $50,000[footnoteRef:21]. For comparison, consider that the per kilo price of a regular car ranges from $10 for a Hyundai Accent, to $60 for a top of the line BMW—like the M6. So, although the Bugatti is not worth its weight in gold, its worth more than its weight in silver and a Hyundai is worth at least its weight in Bronze. [20: My friend—and undergraduate advisor—Francisco Claro suggested this calculation to me a few years ago. His example at that time was a fighter jet.] [21: Based on data on January 14, 2013 16:45 NY Time, on goldprice.org, the exact number was $53,586.]

Now, you may argue that comparing a kilo of Bugatti and a kilo of silver is pure nonsense, since there is not much you can do with an actual kilo of Bugatti. Yet, this nonsensical has much to teach us about the connection between physical order and the information that is packed in a product.

Imagine for a second that you just won a Bugatti Veyron in the lottery. Pumped up, you decide to take your new car for a test drive. In your excitement, you crash the Bugatti into a wall, escaping unharmed but a little sad, since you did not have any car insurance. The car is a total wreck; you have blown it to smithereens. Now, how much is that kilo of Bugatti worth?

The answer to this question is perfectly obvious. The sticker price of the car evaporated in the seconds it took you to crash it against that wall, but not its weight. So where did the value go? The pre-crash Bugatti had a great resale value, as least as measured by its price, whereas the resale value of the post-crash Bugatti is not that hot. The resale value of the Bugatti evaporated in the crash, not because the crash destroyed the atoms that make the Bugatti, but because the crash changed the way in which these were connected. As the parts that made the Bugatti were pulled apart and twisted, the information that was embodied in the Bugatti was largely destroyed. This is another way of saying that those two million dollars were not stored in the car’s atoms, but in the way in which these atoms were arranged[footnoteRef:22], and that is information[footnoteRef:23]. [22: There are many good references discussing the connection between thermodynamics and economics that we discuss here. A good discussion of this can be found in Eric Beinhocker’s The Origin of Wealth (2005) Harvard Business School Press.] [23: You might argue that there is much more to the value of a Bugatti than its physical order or information. I agree with you, and suggest you to keep on reading. I will be adding these additional considerations gradually.]

So it looks like the value of the Bugatti is connected to physical order[footnoteRef:24], which is the same as information. But what is information? Defining information is not easy, but I will give it a shot. According to Claude Shannon[footnoteRef:25], the father of information theory, information is a measure of the minimum volume of communication required to uniquely specify a message. Now, I know this sound like a mouthful, and it does not appear to connect directly to the Bugatti example, but in the next lines I will unpack Shannon’s idea and connect it with the information that is embodied in the Bugatti[footnoteRef:26]. [24: Since order is a word that has many different meanings, I would like to clarify the meaning of the word order that I will be using. Going forward I use order to mean physical order, meaning the way in which the parts of a system are arranged. By definition, this is the same as information. It is the order you wish to have in your closet, but not the order that involves a command; such as the order you place in a restaurant. Physical order is what differentiates the Bugatti before the crash from the wreck that was left after the crash. Physical order cannot transform lead into gold, but in the case of the Bugatti can transform steel into silver, and in the case of the Hyundai Accent, it can transform steel into Bronze.] [25: Norbert Wiener arrived at the same formula in his formulation of information theory, which he coined cybernetics.] [26: Please note that the information packed in the Bugatti is information in its physical meaning of order, which is different from the computational meaning of communication. Yet both of these definitions have a deep mathematical equivalence that allows us to move from one to the other.]

Since most of us will never own a Bugatti Veyron, I will use a more accessible product instead: a tweet. For those that are not familiar with twitter[footnoteRef:27], I’ll just say that twitter is a micro-broadcasting platform where users post messages of up to 140 characters. A tweet is just one of these messages, or a little packet of information. So now that we know what a tweet is, we are ready to ask Shannon: how much information is contained in each tweet? [27: Either because you don’t have access to the technology, you don’t use it, or because you are reading this in a future in which twitter is no longer popular.]

An important disclaimer we need to make before introducing Shannon’s definition of information is that Shannon separated information from meaning[footnoteRef:28]. Information is the minimum volume of data we need to specify a message, any message, and whether this message is a tweet that is all A’s, or the funniest tweet you ever saw, is irrelevant from a pure information theory perspective—but it will become relevant later. [28: James Gleick, The Information]

To understand the information content that is contained in a tweet, consider two hypothetical twitter users: Abby and Brian. Both Abby and Brian have a book that contains every possible tweet and are using this book to play the “guess game”. In this game, Abby randomly chooses a tweet from her book and asks Brian to guess it using only yes or no questions. According to Shannon, the amount of information that is embodied in a tweet is equal to the number of yes or no questions that Brian needs to ask to guess Abby’s tweet with 100% accuracy[footnoteRef:29]. But how many questions is that? [29: This is the simplest possible case we can use to illustrate Shannon’s theory, since it assumes all tweets and characters are equally likely. In reality, all characters and strings of characters are not equally likely. It is much more likely for a tweet to contain the sequence of characters “http://” than the sequence of characters “qwzykq”. If Brian knows about these differences he could exploit them to reduce the number of questions he needs to ask in order to guess the tweet. If you are uncomfortable with these simplifying assumptions, assume that Abby and Brian come from different planets, and that the only thing that Brian knows about Abby’s alphabet is that it is based on 32 different characters.]

For simplicity, we will assume that Abby and Brian are using a set of 32 characters and that both know this. You can think that they are using the English alphabet in lowercase plus a few extra characters, such as the space ( ), the colon (:), the slash (/), the comma (,), the period (.), the at (@) and the hash (#). Also, this time without any loss of generality, we will assume that Brian and Abby both have a table that maps each character to a number (a=1, b=2, etc…).

A sure way for Brian to guess Abby’s tweet is to guess each character, and the best way to search for each character is to use each question to cut the number of possible characters in half. If Brian decides to guess the tweet by guessing each character, then his first yes or no question should be: “Is the first character larger than 16?” If Abby answer’s no, then Brian will know that the first character on Abby’s tweet is in the first fifteen rows of his character table. By asking that question Brian reduced the number of possible first characters in Abby’s tweet by half. His second question does the same for the possible characters that remain, and it is: “Is the first character larger than 8?” If Abby says yes, Brian will now that the first character on Abby’s tweet is between character 9 and 16 of his character table. Now, you should be able to guess what Brian’s next question will be? This is “Is the first character larger than 12?”

As you can see, with each question Brian is cutting the number of possible characters in half. After five question Brian should always be able to guess a character, since there are 32 possible characters and we need to divide 32 by 2 five times in order to cut down the set of options to just one. Finally, since there are 140 characters in a tweet, Brian will need 140x5 guesses, which is a total of 700 yes or no questions, or bits, to uniquely identify Abby’s tweet[footnoteRef:30]. [30: Notice that this 700 number was the one that was sitting on top of the 2700, which is the total number of possible tweets. The general formula in this case is N log2(S), where N is the number of characters and S is the size of the alphabet. This is equivalent to log2(SN) where SN is the total number of possible tweets. In general, notice that the information content of a message goes as the base two logarithm of the number of possible messages. This is because the most efficient way to search for a message, or uniquely identify it, is to iteratively cut the search space in half.]

Shannon’s information theory is the basis of modern communication systems. By quantifying the information content of messages Shannon helped develop the technologies needed to encode and decode messages, and hence, create digital communication technologies. So in a profound way we can say that the ghost of Shannon lives in every tweet, email and computer file, since all of these make use of a theory that evolved in his brain[footnoteRef:31]. But as we will see, Shannon’s theory can also help us understand the amount of information that is embodied in a product. [31: Not denying the contributions of his contemporaries such as Warren Weaver, Alan Turing and Norbert Wiener.]

Shannon’s theory tells us that information is related to search, since the amount of information that is embodied in Abby’s tweet is equal to the number of yes or no questions that Brian needs to ask in order to search for it in a list containing all possible tweets[footnoteRef:32]. But how many yes or no questions do we need to identify a Bugatti? [32: Note that the number is arrived by assuming an optimal search strategy that assumes no information about the possible tweet or the regularities of the languages used to write it. Brian could have tried to guess the tweet by asking less informative yes or no questions, such as: “is the tweet: Why Washington politics are as successful as Washington sports teams?” Yet a search strategy based on guessing fully formed tweets is extremely inefficient, since instead of cutting the search space in half with every question, it will only by eliminating one of the 2700 possible tweets. On the other hand, if Brian knows that the tweet is not gibberish, and it is written in English, he could design strategies to guess the tweet in less than 700 questions, since the language contains correlation that make the next letter non-uniformly distributed once the previous letters are revealed. ]

The case of the Bugatti is not as simple as that of a tweet. After all, it involves positioning gazillions of atoms and not just 140 characters. Also, in the case of the Bugatti, we are not searching for any possible configuration of atoms, but for those that produce something that is like a Bugatti, or at least similar enough for us to be unable to tell the difference. Unlike the case of twitter, where we consider all tweets to be equivalent, the case of the Bugatti is one in which we consider groups of configurations that are similar. For example, one group is the one containing all configurations that are equivalent to a Bugatti’s wreck, no matter whether the wreck resulted from crashing the Bugatti into a wall or driving it down a cliff. Another group contains all configurations that are equivalent to what we would call a Bugatti in perfect condition. This group includes configurations in which the tires have been rotated or where the windshield wipers have been exchanged.

By creating groups we are rephrasing our problem, from the information required to identify a unique element of a set to the information required to identify subsets of configurations that are equivalent according to a criterion. As Shannon famously said “The fundamental problem of communication is that of reproducing at one point either exactly or approximately a message selected at another point”. Making a distinction between unique elements and sets of elements is important because it helps us differentiate between the information needed to transmit a message with content or to transmit gibberish. Technically speaking, an image in which the color of each pixel is chosen at random is uncompressible, and hence, requires more information to be transmitted than the image of a face or a landscape. Yet, if we consider all such random images to be equivalent, the information needed to transmit them becomes tiny. In fact, we can transmit a random image from an arbitrary size easily using less information than that needed to write one tweet[footnoteRef:33]. Ultimately, these equivalences tell us two things. The first one is that when we talk about information we are not looking to exactly specify a message or object, but to specify it with enough detail to identify a message or object that is equivalent to the one we had in mind. The second one is that much of what we interpret as information is not the raw counts of bits needed to transmit a message, but the correlations that exist within a message or an object, which technically can reduce the amount of bits needed to transmit it. These are the correlations that make music predictable enough to dance to it, or that make the picture of a landscape seem smooth and harmonious, and quite different from a collection of random pixels. Hopefully, the discovery of these correlations within this text motivates you to keep on reading this book. [33: In pseudo-code, we can write an expression encoding an image of one million times a million pixels as: for x=1:1,000,000; for y=1:1,000,000, Image(x,y)=random_number; end; end.]

So let’s use these last insights to get back to the Bugatti example. To quantify the amount of information that is embodied in the Bugatti assume you are now in a game show where the Bugatti is hiding behind a mystery door, but where you don’t know if it is in perfect condition or not. The game lets you ask yes or no questions about the position of each molecule in the Bugatti’s body, but not questions such as “Is the Bugatti a wreck?” These are cumbersome questions such as: “Is the molecule in position (1.3545422334, 3.756545121) a rubber molecule?” but for our though experiment they will have to do. In this game the amount of information that is physically embodied in the Bugatti is the number of yes or no questions that you need to identify whether the Bugatti is in perfect condition or not. Now, it should be perfectly clear that the number of yes or no questions needed to make sure that the Bugatti is not in perfect condition is much smaller than the number of yes or no questions needed to make sure the car is in perfect condition. This should be clear because there are many ways for the Bugatti to be a wreck. After all, you don’t know if the Bugatti was wrecked from the front, or the back; or whether it has a small boo-boo caused by the opening of a door in a parking garage[footnoteRef:34]. So to make sure the car is in pristine condition you will need to ask enough questions to map out the Bugatti in full detail. Since we need to answer many more yes or no questions to identify a Bugatti in pristine condition, than a wreck, the moral of the story is that a Bugatti in perfect condition carries more information than a Bugatti wreck, and that information is destroyed in a car crash. This tells us that although the Bugatti is made of many atoms, what distinguishes it from a wreck is information. And hence, we can say the Bugatti is made of embodied information, much like everything else. [34: A scratched door is not equivalent to a wreck, but is also not equivalent to a car in absolutely perfect condition. So the number of yes or no questions you would need to ask to check whether the car is in perfect condition would need to include making sure that the car does not have any scratches.]

Shannon’s theory helps us quantify the amount of information that is embodied in a tweet or in a Buggati. Taking Shannon to the extreme we can say something like: “a Bugatti is made of ~1027 atoms and one terabyte of information, while a Bugatti wreckage is made of the same number of atoms but only a few hundred megabytes of information.”[footnoteRef:35] Yet, there are some important questions that Shannon’s theory leaves unanswered, at least when it comes to Bugattis. One of these questions is: what are the relevant groups that we need to consider when searching for the arrangement of atoms that we call products? After all, grouping Bugatti’s with rotated tires in one group and grouping wreckages in another group makes perfect sense to us, not because I have explained it, but because we—humans—have a great sense of intuition. So to develop a deeper understanding of the information that is contained in a product we need to ask the question: how do we define the groups of arrangements of atoms that are relevant as products? As you will see, by defining these groups we will begin to bring back the idea of meaning into our definition of information. [35: Don’t quote me on these numbers.]

Grouping Bugattis with rotated tires on one group, and wreckages on another, made perfect sense to us because we intuitively know that the members of each of these groups can perform different functions. A wrecked Bugatti and a pristine Bugatti mean something different to us because these are groups of atoms that we associate with different functions. So we will hijack the word meaning to indicate configurations that convey the same context specific information to the recipient. In the case of twitter, this will group in the same category the tweets “Baby I love you?” and “I love you baby”, since a recipient of either of these tweets would identify them as the same. Hence, both of these tweets would be performing the same function[footnoteRef:36], telling your baby that you love her. [36: What it means for something to be useful is a philosophically difficult question that I am avoiding to discuss here. I will touch upon this topic in the next chapter.]

The association between the information embodied in a digital or tangible object, and the functions associated with it can take some time to comprehend, so I will a second example. Consider a library. In a library we can order books by title, topic, publication year, size or language. Each of these orders contains information, but the information embodied in each order facilitates different functions. In the case of the library these are search functions. For instance, ordering books by topic and publication year can help us quickly identify the earliest books on quantum mechanics, while ordering the books by the last name of their authors can help us quickly find all of the works of Mark Twain. The latter is true whether we order books in alphabetical order or in reverse alphabetical order. This tells us that there is a correspondence between alphabetical order and reverse alphabetical order that makes both of these orders equivalent, at least with respect to the search function they help perform. So going forward, we define the information content of a product, not merely as the information that is embodied in the way in which its atoms are arranged, but as the information embodied in any of the configurations of atoms that are able to perform a given function. This is true whether this is the search function that helps us find a book in a library, or the transportation function that allows us to ride a group of atoms down the road.

The case of the Bugatti is a bit more nuanced that the one of the library because the set of functions that people expect from a vehicle are larger and more complex than the ones that people expect from a library catalog. They are also, much more social. In the case of the Bugatti, you can think that its Chilean owner did not have only transportation in mind, but also was hoping to look like a douche while cruising down the streets of Santiago[footnoteRef:37]. [37: Although not quite a Bugatti, a BMW can perform a similar function at a more modest price.]

So the Bugatti example teaches us that physical order, or information, is partly what makes the Bugatti comparable to silver and gold. It also tells us that each good does not only have information, but a “meaning”, which is represented by the practical and social functions that the good helps us perform. Yet, functions are not the only way to differentiate between alternative configurations of atoms. We can also differentiate between different configurations of atoms by looking at the amount of effort, or energy that is required to construct each of them[footnoteRef:38]. [38: Some orders are not only harder to create, but also more delicate. Consider the order required to make the Bugatti run. This is an order that can be disrupted by a broken transmission. A broken transmission, however, will not disrupt the recycling process, so we can say that the order required to make a car go is more delicate than the order required to prepare the parts of the same car for recycling. The orders that are more delicate are, in general, orders that require more information, since the fact that they can be disturbed by small changes means that the number of equivalent configurations that pertain to the same group as them is small.]

To get a hold on the differences in the amount of information that is embodied in a product, and it’s relation to the energy needed to create a product, consider a do-it-yourself bookcase and a do-it-yourself car. Certainly, most people are fine assembling their own IKEA furniture, at least after pulling out a handful of hair or two. Most people, however, would stay away from buying a car from the Internet if this was shipped dissembled in a box with a set of instructions. This is because finishing cars requires embodying more information than bookcases, and this tells us that the information that is embodied in a car requires a greater effort to be embodied than the information that is embodied in a bookcase. It is important to note, however, that information and effort are not perfectly equivalent. We need more effort to assemble one thousand bookcases than a single bookcase, but a thousand bookcases do not carry more information than a single bookcase, since the last 999 bookcases are redundant with the first one.

We began this chapter by wrecking an imaginary Bugatti to illustrate that products are not made of atoms, but information. This is the information that is contained in the way in which the atoms that make a product are arranged. The Bugatti example is tangible, and it gives us a tactile experience. Yet, what is true for the Bugatti is also true for digital products, whether these are as simple as a tweet or as sophisticated as a Shakespeare play. In the case of a tweet, or a Shakespeare play, we might fail to see the physicality of the information, but information is always physically embodied. If that tweet is stored in your hard-drive, its physicality is embodied in the spins of the electrons that represent each of its 700 bits. If the tweet is being transmitted through Wi-Fi, its physicality is embodied in the electromagnetic waves that have been modulated to transmit the message. Information, whether tangible or digital, is always physically embodied. In fact, talking about physically embodied information is redundant, since there is no such thing as information that is not physically embodied. Yet, information is weightless and immaterial. It exists in a magical duality with matter that leads people to sometimes believe information is not physical. My favorite example to explain this duality is to consider a deck of cards. Every time we shuffle a deck of cards we are changing the information that is contained in it, but not its weight. Information is not a substance, yet it cannot exist divorced from matter. The subtlety of this duality is often elusive, so I will use the redundant phrasing “physically embodied information” often throughout the text, even though saying: “physically embodied” is technically redundant when talking about information.

Going back to products we find no difference in physicality between tangible products, like a Bugatti, and intangible products, like a Shakespeare play. Both types of products exit as physically embodied information. Yet, digital products differ from tangible products in the ease with which we can reproduce, amplify and create them. There is a great difference between our nascent digital world, which is being built on methods that help us quickly embody information in electrons and photons, and the world where we could only embody information in tangible macroscopic products, by bluing molecules and atoms. Whispering dance moves to photons or rearranging electrons in a circuit is a more flexible, energy efficient and quicker way of embodying information than rearranging atoms, even though it has its own limitations[footnoteRef:39]. This difference might grow to be as large as that between the world of modern humans and that of our distant evolutionary cousins. Yet, before we get ahead of ourselves and get sucked into the rabbit hole of the futures opened by modern information technologies, we will continue to explore some general aspects of the information that is physically embodied into tangible and digital products. The goal we are trying to achieve is to develop a description of the process of economic development that is based on our species ability to accumulate physically embodied information. Shannon’s theory brings us a long way, but there is certainly more to a product than the information that is embodied in the way in which its atoms are arranged or the functions that it can perform. Think of the information that is embodied in medicine. It is hard to argue that the information that is embodied in a pill can be reduced to that which is contained in its tiny active chemical component. This invites us to explore other aspects of the information that is embodied in a product. The first one of these aspects is the origins of the information embodied in them. The second one is the ability of the information that is embodied in products to amplify the availability of the practical uses of the knowledge that was required to make them. So here is where we will go next. The next stop in our journey is an exploration of the origins of the information that is embodied in a product. This will help us understand what’s so special about the apples that do not grow on trees. [39: Virtual houses do not protect us from the rain.]

Chapter III: The difference between apples and apples

“That’s all the motorcycle is, a system of concepts worked out in steel. There’s no part in it, no shape in it, that is not out of someone’s mind”

Robert M. Pirsig (1974)[footnoteRef:40] [40: Zen and the art of motorcycle maintenance (page 174)]

“I paint objects as I think them, not as I see them.”

Pablo Picasso

Consider two types of apples. Those that you buy at the supermarket, and grow on trees, and those that you buy at the Apple store, and are designed in Silicon Valley. Both are traded in the economy, and both crystalize order, whether in silicon chips or biological cells. The main difference between them is not their number of parts, or their ability to perform functions—edible apples are the result of tens of thousands of genes that perform sophisticated biochemical functions. The main difference between apples and apples, is that the apples we eat existed first in the world, and then in our heads, while the apples we use to check our email existed first in someone’s head and then in the world. Both of these apples are products, but only one of them—the silicon apple—is a crystal of imagination[footnoteRef:41]. [41: Why do I choose a crystal as a metaphor? In my opinion a crystal is the right metaphor because it is an ordered arrangement of atoms that is static. When we create products, we create tangible and digital objects that contain a frozen instantiation of a process that is much more fluid and dynamic: imagination. Once a car is built, it becomes the 2013 model, and it is basically frozen until the next model comes out. The same will be true for this book. Revisions to later editions—if any—will be unable to change the information that was physically embodied in the first edition. In that sense, the products that we create are crystals of imagination; they are static instantiations of our ideas. ]

Thinking about products as crystals of imagination is important because it highlights that products do not differ only on the information that they embody, but on the source of that information. Edible apples existed before we had a name for them, a price for them, or a market for them. They were present in the world, and as a concept, we simply imported them into our minds. On the other hand, iPhones and iPads are mental “exports” rather than “imports”, since they are products that were begot in the mind before they were distributed throughout the world. So the main difference between apples and apples does not reside on the existence of physical order, but on the source of their physical order.

Thinking about the world in terms of crystallized imagination is poetic, but also useful, since it provides a different way of parsing out the diversity of products that make the economy and provide us with an alternative perspective from which to understand important economic processes. Consider international trade. When we think of products as crystals of imagination global trade becomes nothing more than a global exchange of physically embodied ideas. The idea of crystallized imagination tells us that a country’s export structure carries information about more than just their abundance of capital and labor. Countries’ export structures tell us about the ability of people in that country to create tangible instantiations of imaginary objects, such as cars, espresso machines and motorcycles. In fact, the composition of a country’s exports informs us about the productive knowledge available in that country. A country that exports motorcycles is country shouting: If you imagine a motorcycle, we can figure out a way to build it[footnoteRef:42]. Countries that are able to create what people imagine are countries that can mine the depths of human ingenuity in search for new products, and as we will see, this is a capacity that is highly industry specific, hard to create, and that evolves and diffuses slowly. [42: Certainly, firms rather than countries are the ones that export. But since the export basket of a country is the combination of that of many firms, here I will discuss exports using countries for simplicity. ]

Thinking of product exports in terms of crystallized imagination tells us that we live in a world in which some countries are net importers of imagination, while others are net exporters of it, and that these countries are not the same than those with a positive or negative trade balance. For example, consider the export of a BMW. This Bavarian vehicle is clearly an export of imagination, since BMWs crystallize ideas of German design and engineering that are hardly embodied in Ecuadorian bananas. The balance of trade between two countries—monetary value of exports minus imports—is not always the same as their imagination balance. Saudi Arabia and Russia both have a positive balance of trade, due mostly to large reserves of mineral resources, but negative imagination balance, since their export economies are based mostly on exchanging oil and natural gas for luxury SUVs and sports cars.

Indeed, there are many pairs of countries with balances of trade that run opposite to their balances of imagination. Consider Chile and Korea. The exports of Chile to Korea in 2010 amounted to roughly 4.3 billion dollars, with refined and unrefined copper representing 52% of this amount[footnoteRef:43]. The exports from Korea to Chile during that year were only about 3.3 billion dollars, but these included almost a billion and a half dollars in vehicles and vehicle parts[footnoteRef:44]. Chile’s trade balance with Korea is clearly positive, by as much as one billion dollars, but Chile’s imagination balance with Korea is clearly negative, when we consider the nature of the goods exchanged. [43: http://atlas.media.mit.edu/explore/tree_map/export/chl/kor/show/2010/] [44: http://atlas.media.mit.edu/explore/tree_map/export/kor/chl/show/2010/]

Exports from Chile to Korea in 2010 (from atlas.media.mit.edu)

Exports from Korea to Chile in 2010 (from atlas.media.mit.edu)

Another example is Brazil and China. In 2010 Brazil exported nearly USD 31 billion to China, and imported only USD 26.5 billion. Brazil is one of the few countries that enjoy a positive balance of trade with China (Chile is another). Yet, Brazil has a negative imagination balance with the largest Asian economy, since its trade involves mostly the exchange of iron ore and soybeans for electronics, chemicals and even processed metals.

Exports from Brazil to China in 2010 (from atlas.media.mit.edu)

Exports from China to Brazil in 2010 (from atlas.media.mit.edu)

So classic economic concepts, such as the balance of trade between two countries, seem incomplete once we reinterpret products as crystals of imagination. Yet, there are other examples of economic ideas that need to be revisited after realizing that the source of a product’s information might be of economic relevance. One of these is the way in which we classify industries and products[footnoteRef:45], which are essential groupings we use to create mental representations of the economy[footnoteRef:46], and hence, represent an essential building block of the language we use to describe economic processes. [45: To be clear, the difference between an industry, and a product, is the difference between the makers of a product and the product. For instance, a shoe is a product that is made by the footwear industry.] [46: These groupings are not equivalent to those discussed in chapter 2—the ones we used to explain the configurations that are equivalent in the case of the Bugatti.]

Classifying products and industries is essential to connect theoretical and empirical representations of the economy. Yet, finding classifications that work in both theory and practice is not easy because the empirical world is characterized by a diversity of products that is hard to categorize and classify. Products are quirky. They include pineapples, trashcans, flowerpots, bicycle racks, chimneys and ATM machines, among many others. Yet theories are constructed on numbers and concept that have trouble with this quirkiness, and hence, are not good at generating representations of products such as pineapples or rollercoasters. The problem is that to connect our theories with our data we need to represent both data and theories in the same language, and usually this requires us translating the language of one into that of the other. During the last centuries the usual compromise has been not to create a theory that includes pineapples, but to bundle pineapples with other goods in a coarse classification, such as agriculture. This coarsening pushes as to talk about industries and products in terms of raw materials, agriculture, services and manufactures, or in terms of capital-intensive industries or labor-intensive industries. Yet, this coarsening might not be the best way to parse the complexity of our world.

Consider industries traditionally classified as services, such as movies, music or video games. These are industries that crystallize the imagination of their creators, just like cars, guitars and video cameras, which would be considered manufactures, not services. From the perspective of crystallized imagination, however, a video game studio is conceptually closer to a guitar manufacturer than to a retailer, since the main difference between a video game and a guitar is that video game designers crystallize their imagination in bits, while guitar manufacturers do so in atoms. A video game studio is not an industry that provides a service, like serving coffee or providing lodging, but an industry that takes ideas and embodies them in a sharable physical form.

The distinction between services that serve, and services that manufacture using bits, is important in a world where people often associate “service” industries with higher levels of development, when in reality, services are a mixed bag. Think of the 21st century United States. The big gaps in income in the US are mostly gaps between the services that serve, or administrate, and those that crystallize imagination[footnoteRef:47]. Software manufacturers who concoct computer code to create new digital products are paid excellent salaries and are driving the technology boom of San Francisco, Boston and New York. At the same time, fast food workers, who also participate in the “service” sector, are struggling to make ends meet. Our traditional definition of services is a mixed bag—a bag that instead of mixing apples and oranges is mixing Silicon Valley engineers working on complex algorithms, with fast food employees, working on the concoction of a priori specified sandwiches. Talking about “services” is not longer an appropriate use of language in a world in which the occupations that fall into the service worker category have little to no relation[footnoteRef:48]. Certainly, talking about products and services is not the best way to parse out economic activities. It is just a convenient language we have grown used to. [47: San Francisco split by Silicon Valley's wealth http://www.latimes.com/business/la-fi-silicon-valley-backlash-20130814,0,7114762.story] [48: Certainly, there are more nuances in the academic literature than in the colloquial use of the word services that I describe here. For instance, it is common in some circles to talk about knowledge intensive business services (KIBS). I note, however, that my distinction does not focus on the knowledge that is required as an input to a service activity, but on whether these activities result in the embodiment of that knowledge into a crystal of imagination. ]

A better way of parsing the economy is to think of economic activities in terms of the information that they embody. When we consider the origins of that information, we will be talking about the imagination that an industry crystallizes or its relation to other industries that crystalize of imagination. Certainly, not all economic activities are involved in crystallizing imagination directly, since crystals of imagination need to be transported, financed and marketed, and these are processes that do not crystallize a much imagination. In fact, a large number of economic activities do not focus on embodying information in tangible or digital products, but focus on chaperoning the flow of embodied information. The existence of these chaperoning activities, however, does not contradict the fact that the economy is made of information, since all of these activities orbit around our ability to crystallize imagination and would not exist in its absence. There would be no economy in a world in which crystals of imagination do not exist, since there would be nothing worth transporting, trading or financing. Sure, we can think of an economy in which hominids just trade fruits, but this would be only hypothetical, since our civilization has been built for more than eighty-thousand years in the gradual exchange of objects that were begot by our minds[footnoteRef:49]. [49: Although subsets of the economy can make a living that way]

We can illustrate the centrality of the process of crystallizing imagination in the economy by looking at the paths connecting each economic activity to a crystal of imagination. Consider the retail industry. According to the National Retail Federation (NRF), retail represents roughly 8% of the GDP of the United States[footnoteRef:50]. Yet, retail depends critically on crystallized imagination since there would be no retail in the world if it weren’t for the clothes, cars, and furniture that are transacted in shops. So retail is an activity that is derivative of our species ability to crystallize imagination. [50: http://www.nrf.com/modules.php?name=Pages&sp_id=1214]

Another example of an activity that is derivative of our species ability to crystallize imagination is mineral extraction. Mining is not a service, yet it has a purpose only in a world where people use the atoms we harvest from the ground to crystallize imagination, no matter whether these are the lithium atoms needed to make rechargeable batteries or the oil used to transport these atoms to the centers of “crystallization”. The financial sector, on the other hand, is a sector allegedly involved in allocating the financial resources needed to help prioritize productive processes. While we may doubt this claim, we do not need to enter that discussion to realize that the financial sector helps lubricate the transactions of some crystals of imagination. For instance, the financial sector provides the mortgages required for homes—which are crystals of the imagination of architects and constructors—to change hands. Finally, consider the transport, logistics and communication sectors. These are sectors involved actively in the movement of crystals of imagination, or the atoms and bits required to make them. In fact, once we interpret the economy in terms of crystallized imagination the analogy between the Internet and the US highway system becomes even more obvious. Both, the Internet and the highway system help the crystals of imagination created by some to reach others[footnoteRef:51], and hence, are related to crystals of imagination in an analogous manner: helping the crystals of imagination made by some reach others. [51: As an example, we note that during 2012 Netflix and YouTube accounted for nearly 50% of all Internet traffic in the US. We note that there is an imperfect match between the bits involved in digital content and their online traffic due to services that help buffer content, such as the services provided by Akamai. These services can reduce Internet traffic without reducing the amount of content being distributed. ]

In sum, crystallized imagination is a simple way to conceptualize the informational soul of the world economy. The economy gravitates around our species ability to transform imaginary objects into a tangible or digital reality. Even though the embodying of information is not necessarily the activity that collects the major financial rewards[footnoteRef:52], all economic activities are connected either directly or indirectly to this unique human capacity. [52: The reader should not that it is wrong to equate the activities that generate individual financial rewards, with those that generate economic value. Often, the greatest financial rewards are not for those who create new crystals of imagination, but for those who control their trade or bet on others through financing.]

Finally, thinking about products in terms of crystals of imagination helps us understand the importance of the source of the information that is embodied in a product. Now we know that the information that is contained in products was born in the minds of the hairless apes that can create what they imagine. Complex products are not just arrangements of atoms that perform functions, but ordered arrangements of atoms that were originated as virtual ideas. It is therefore the origin of the information embodied in products what helps us resolve the paradox of the pill that I posed at the end of last chapter. The mental origins of the information embodied in products tell us that there might be important additions to the information that is embodied in physical order that rides implicitly in products. Hence, there is more information in a pill than the one crystallized in the structure of its tiny active chemical component. This additional information is more “meaningful” and less mechanical, in the sense that it is context dependent. It is information about the effects that the chemical compound will have in the human body, and on the mechanisms used to validate that information. This is information that we do care about when we buy a pill, but that it is not encoded in its active compound.

For the time being, however, we will conclude by asking: Why do hairless apes bother to make tangible instantiations of their imagination? Why do we put so much effort to create complex products? As we will see, it cannot be just because we are able to do so, but also it cannot be because of either greed or self-interest[footnoteRef:53], as Gordon Gecko moronically suggested. After all, crocodiles, zebras and all the wild animals are not limited by a lack of greed or self-interest, but are nevertheless unable to crystallize imagination as we do. So now that we have a better definition of what is the information that is embodied in products, we will continue our journey by exploring the reasons why we make products. The next section will continue to explore what hairless apes have to gain through their tireless commitment to transform dreams into reality. [53: The view of the world in which humans act to maximize utility is logically unsound when combined with its own empirical validation—the idea of revealed preferences. In simple terms, it is easy to understand the circularity of the argument by considering separately the theory—utility maximization—and its empirical validation—individual choices reflect individual preferences. If we consider the idea that individuals maximize utility to be a hypothesis, then to test it, we need a test in which one of the possible outcomes is to observe that individuals do not maximize utility. Revealed preferences is not that test, since it assumes that the actions taken by an individual are always aligned with the choice that maximizes her preference, and hence, is a test that by construction cannot result on an individual making a choice that does not maximize their utility. So together, the ideas of utility maximization and revealed preferences cannot be considered a proof of individuals acting on self-interest, or of anything else for that matters. Francis Fukuyama points this out as well in his book Trust (chapter 1, page 19). When referring to utility maximization, and the attitude of economists to this idea he writes: “Some economists try to get around this problem by broadening the definition of utility beyond pleasure of money to take account of other motivations such as the “psychic pleasure” one receives for “doing the right thing,” or the pleasure people can take in other people’s consumption. Economists assert that one can know what is useful by their choices—hence their concept “revealed preference.” The abolitionist dying to end slavery and the investment banker speculating on interest rates are both said to be pursuing “utility,” the only difference being that the abolitionist’s utility is of a psychic sort. At its most extreme, “utility” becomes a purely formal concept used to describe whatever ends or preferences people pursue. But this type of formal definition of utility reduces the fundamental premise of economics to an assertion that people maximize whatever it is they choose to maximize, a tautology that robs the model of any interest or explanatory power.” ]

Chapter IV: Knowledge Amplifiers

“Any sufficiently advanced technology is indistinguishable from magic.”

Arthur C. Clarke (1962)[footnoteRef:54] [54: Profiles of the Future: An Inquiry Into The Limits of the Possible (1962; revised 1973)]

“We don’t make most of the food we eat, we don’t grow it, anyway. We wear clothes other people make, we speak a language other people developed, we use a mathematics other people evolved and spent their lives building. I mean we’re constantly taking things. It’s a wonderful ecstatic feeling to create something and put it into the pool of human experience and knowledge.”

Steve Jobs, 1983[footnoteRef:55] [55: http://bits.blogs.nytimes.com/2014/01/24/the-30-year-old-macintosh-and-a-lost-conversation-with-steve-jobs/?_php=true&_type=blogs&_r=0]

In my talks, I often ask the attendees to raise their hands if they have used toothpaste that morning. I find this to be a good way to get audience participation since the embarrassment of not having used toothpaste encourages even the shyest attendee to raise her hand. After almost everyone has raised their hands, and some people have giggled, I ask audience members to keep their hands up only if they know how to synthesize sodium fluoride. As you can anticipate, all hands go down. This shows that products do not only give us access to embodied information, but also, to the practical uses of the knowledge that is required to make them. That is, they give us access to the practical uses of knowledge residing in the nervous systems of other people, and that is amplified by our ability to crystallize imagination.

In this chapter I will discuss the practical applications of our ability to crystallize imagination. I will focus on four aspects of products. First, I will discuss the ability of products to distribute the practical uses of the knowledge used in their production. Second, I will explore the role of products as a mean of creative expression. Third, I will describe products as a source of human augmentation. And finally, I will describe the importance of products to enable combinatorial creativity. Together these four abilities will help us understand the benefits of crystallizing imagination, and therefore, will help explain our want for crystals of imagination. By exploring the practical uses of products we will also learn that products are not just deposits of information, but also, that they represent a unique form of human communication[footnoteRef:56]. Products are unique in their ability to communicate, because unlike messages, they are not shared to be understood, but to be utilized. This utilization does not require much understanding. We do not need to know how a car or a computer works to reap their practical benefits. The same is true for soap. Finally, we will build on these ideas to refine our definition of the economy. In this last section, we will move our gaze away from definitions of the economy that focus on the allocation and distribution of resources, and instead, will focus on the ability of the economy to amplify the practical uses of knowledge and information. [56: Which echoes Marshall McLuhan definition of medium as used in his famous quote: “The Medium is the message” (McLuhan, Understanding Media, The extension of man). McLuhan defined medium as “any technology that ... creates extensions of the human body and senses”]

Going back to our toothpaste example we can note that when we are buying toothpaste we are not simply buying paste in a tube. Instead, we are buying access to the practical uses of three things: the creativity of the person who invented toothpaste, the scientific knowledge informing the chemical synthesis that is required to make toothpaste, and the productive knowledge required to synthesize sodium fluoride, put it inside a tube, and make it available across the planet. Something as simple as toothpaste gives us indirect access to the practical uses of the imagination and knowledge that exist, or existed, in the nervous system of people we probably never met.

The ability of toothpaste to provide us with access to the practical uses of knowledge residing in a stranger’s nervous system is quite magical. Yet the magic of toothpaste, much like that of any product, does not reside solely on its ability to gives us access to the practical uses of knowledge residing in the nervous system of others. Products are magical also because they endow us with capacities that escape our individuals’ abilities. Products augment us, and this is a great reason to want them.

Think of a guitar. Guitars allow us to sing with our hands by combining knowledge on Pythagoras’ twelve-tone scale with expertise on what is the right wood to build a guitar and how to shape it. If the guitar is electric, it will also embody knowledge of how musical waves can be captured using capsules, and how these sounds can be amplified into the sounds that many of us enjoy. All of these are capacities that are needed to make music, at least the kind of music that requires an electric guitar, but that do not need to be capacities of the musician. The musician access the practical uses of this knowledge through the guitar, and by doing so, he is augmented in his capacities as a musician.

Products are magical largely because they augment our capacities. Planes endow us with the ability to fly, ovens with the ability to cook, and toothpaste with the ability to keep our teeth until an older age.

So a good reason of why humans desire products is because products augment our capacities by providing us with access to knowledge that resides in the nervous system of other people. Yet, our need to create complex products does not emerge only from their ability to embody knowledge or augment us. There is an expressive component in the creation of products that we also need to consider. We crystallize imagination because this allows us to transform our ideas into a sharable reality. By default, thoughts are trapped in the prison of our minds and crystallizing our thoughts into products is what allows us to share them with others. A musici