by ERIC D. BEINHOCKER
a summary by Bobby Lopez
This book will argue that wealth creation is the product of a simple, but profoundly powerful, three-step formula— differentiate, select, and amplify—the formula of evolution
We are accustomed to thinking of evolution in a biological context, but modern evolutionary theory views evolution as something much more general. Evolution is an algorithm; it is an all-purpose formula for innovation, a
To picture Epstein and Axtel’s model, imagine a group of people shipwrecked on a desert island, except that both the island and the castaways are simulations inside a computer. The computer island is a perfect square with a fifty-by-fifty grid overlaid on top
The virtual island has only one resource—sugar—and each square in the grid has different amounts of sugar piled on it. The heights of the sugar piles range from four sugar units high (the maximum) to zero (no sugar). The sugar piles are arranged such that there are two sugar mountains, one mountain at the northeast corner and one at the southwest corner, each with sugar piled three and four units high.
The game begins with 250 agents randomly dropped on the Sugarscape.
Each virtual person, or “agent,” is an independent computer program that takes in information from the Sugarscape environment, crunches that information through its code, and then makes decisions and takes actions. In the most basic version of the simulation, each agent on Sugarscape can only do three things: look for sugar, move, and eat sugar.
Each agent has vision that enables it to look around for sugar, and then has the ability to move toward this source of energy. Each agent also has a metabolism for digesting sugar.
Thus, each agent had a basic set of rules that it followed during each turn of the game:
- The agent looks ahead as far as its vision will allow in each of four directions on the grid: north, south, east, and west (the agents cannot see diagonally).
- The agent determines which unoccupied square within its field of vision has the most sugar.
- The agent moves to that square and eats the sugar.
- The agent is credited by the amount of sugar eaten and debited by the amount of sugar burned by its metabolism. If the agent eats more sugar than it burns, it will accumulate sugar in its sugar savings account (you can think of this savings as body fat) and carry this savings through to the next turn. If it eats less, it will use up its savings (depleting fat).
- If the amount of sugar stored in an agent’s savings account drops below zero, then the agent is said to have starved to death and is removed from the game. Otherwise, the agent lives until it reaches a predetermined maximum age.
- In order to carry out these tasks, each agent has a “genetic endowment” for its vision and metabolism. In other words, associated with each agent is a bit of computer code, a computer DNA, that describes how many squares ahead that agent can see and how much sugar it burns each round. An agent with a slow (good) metabolism needs only one unit of sugar per turn of the game to survive, versus an agent with a fast (bad) metabolism, which requires four. Vision and metabolism endowments are randomly distributed in the population; thus,
- Each agent also has a randomly assigned maximum lifetime, after which a computer Grim Reaper comes and removes it from the game.
- Finally, as sugar is eaten, it grows back on the landscape like a crop, at the rate of one unit per time period.
(2) Results 1: The Rich get Richer. But not because of genetic, nor birth, but as an emergent property: the confluence of all factors
As time passes, however, this distribution changes dramatically. Average wealth rose as the agents convened on the two sugar mountains but the distribution of wealth became very skewed, with a few emerging superrich agents, a long tail of upper-class yuppie agents, a shrinking middle class, and then a big, growing underclass of poor agents.
The Pareto distribution is where the so-called 80-20 rule comes from, as roughly 80 percent of the wealth is owned by 20 percent of the people.
The wealth distribution that the simple Sugarscape model produced was just this kind of a real-world Pareto distribution.
First, we can ask, is it nature? —does it have something to do with the genetic endowments of the players? That is, are all the agents with great eyesight and slow metabolisms getting all the wealth? The answer is no.
Are all the agents born on top of sugar mountains getting all the wealth and those with the bad luck of being born in the badlands staying poor? The answer to this is no as well.
How, then, from these random initial conditions do we get a skewed wealth distribution? The answer is, in essence, “everything.” The skewed distribution is an emergent property of the system. It
All this means that even in Sugarscape, there is no simple cause-and-effect relationship driving poverty and inequality. Instead, it is a complex mix of factors. It is not easy to come up with solutions for the poverty problem even in the highly simplified world of Sugarscape.
(3) Results 2: Population growth causes: wealth growth, improved genetics, population swings, and the rich got richer
Epstein and Axtell decided to give each agent a tag indicating its age and whether it was male or female. Once an agent reaches “child-bearing age,” and if that agent has a minimum amount of sugar savings, he or she is considered fertile. Each period, fertile agents scan their immediate neighborhood of one square to the north, south, east, and west. If they find another fertile agent of the opposite sex, they reproduce. The DNA of the resulting baby agent is then chosen randomly, half from the mother, and half from the father. Thus the child’s vision and metabolism characteristics will be some mix of the two parents. In addition, the baby agent inherits wealth from both parents, receiving an amount equal to half the father’s wealth plus half the mother’s wealth.
The baby agent is born in an empty square next to its mother and father, so if the parents live in a rich or poor sugar neighborhood, the child agent will start its life there as well.
Then three things began to happen:
- Over time, both average vision and average metabolic efficiency began to climb, as the most fit members had more and more offspring. As the average of these attributes rose, so too did wealth.
- The new birth-death dynamics introduced population swings.
- The introduction of genetic inheritance as well as wealth inheritance across generations further accelerated the trend toward the rich getting richer and the poor getting poorer.
(4) Results 3: The introduction of trade resulted in: 1) increase in wealth; 2) clustering of trading by regions; 3) prices approached bur never reached equilibrium; 4) sub-Pareto optimal 5) rich got richier
They introduced a second commodity, called spice. Each square on the board now had a value for how much sugar it held and a value for how much spice it held.
Epstein and Axtell also tweaked their agents’ metabolisms so that they all required some of each commodity to survive.
As a final step, Epstein and Axtell made it possible for the agents to trade. There was no assumption of a market or an auctioneer as in typical Traditional Economic models. Instead, there was just straightforward bartering between individuals. As agents move around the Sugarscape, they encounter other agents. At each turn in the game, each agent looks one square to the north, south, east, and west and asks any players in its neighborhood if they want to trade. If one agent has a lot of spice and needs sugar and another agent faces the reverse situation, both agents could improve their circumstances by trading.
First result was that trading made the Sugarscape society much richer.
Secondly, there was some clustering in the trading networks by geographic region. The combination of geography and population dynamics created heavily trafficked trading routes, the computer equivalent of the ancient Silk Road, as agents shuttled back and forth between the sugar and spice mountains.
Epstein and Axtell could look inside each agent during each period of play and determine how much sugar or spice the agent was willing to buy or sell at a series of possible prices. What resulted was an almost textbook downward-sloping demand curve, along with an upward-sloping supply curve, even though Epstein and Axtell did not explicitly build anything about supply and demand into their model. Also, take into account that he intitial conditions were randomly set, but once the model got going, all the behavior was perfectly deterministic.
The actual prices and quantities traded (indicated by the dot in figure 4-4) never setded on the theoretically predicted equilibrium point (at the intersection of the supply and demand X in the figure).
Prices in Sugarscape move dynamically around an “attractor” (a term we will discuss in chapter 5) but never actually settle down into equilibrium.
Epstein and Axtell also found that there was far more trading than there would be if the system was reaching equilibrium, as in the real world.
The Sugarscape market, however, operates at less than Pareto optimality. There are always trades that could have happened, that would have made people better off, but didn’t. Again, this is because the agents’ trades are separated in time and space,
While trade in Sugarscape does “lift all boats,” making the society richer as a whole, it also has the effect of further widening the gap between rich and poor. In
In Sugarscape, there is only one reason to become a borrower—to have children. Epstein and Axtell introduced a rule that said that an agent can be a lender if it is too old to have children, or if it has more savings than it needs for its own reproductive needs. In turn, an agent can be a borrower if it has insufficient savings to have children, but has a sustainable income of sugar and spice.
What surprised Epstein and Axtell was not that significant borrowing and lending activity occurred, but that a complex, hierarchical capital market emerged. Some agents became simultaneously both borrowers and lenders—in effect, middlemen: Sugarscape had evolved banks! Certain really rich agents took on a wholesale role, lending to middlemen, who then made loans to the ultimate borrowers. In some simulations, the hierarchical chain grew to five levels deep.
These large-scale macro patterns grew from the bottom up, from the dynamic interplay of the local micro assumptions.
Economists have long recognized that certain products, such as e-mail, faxes, and telephones, share a property whereby the greater the number of people who use them, the more useful they become. This is called, appropriately enough, a network effect. Traditional Economics, however, has not historically had much to say about why these types of products tend to suddenly catch fire and take off in popularity.
Picture a thousand buttons scattered on a hardwood floor. Imagine you also have in your hand pieces of thread; you then randomly pick up two buttons, connect them with the thread, and put them back down. As you first start out, the odds are that each button you pick up will be unconnected, so you will be creating a lot of two-button connections. As you work away, however, at some point you will pick up a button that is already connected to another button and you will thus be adding a third.
In random networks, the phase transition from small clusters to giant clusters happens at a specific point, when the ratio of segments of thread (edges) to buttons (nodes) exceeds the value of 1 (i.e., on average, one thread segment for every button).One can think of the ratio of one edge to one node as the “tipping point” where a random network suddenly goes from being sparsely connected to densely connected.
Networks of nodes that can be in a state of 0 or 1 are called Boolean network Imagine a string of Christmas tree lights that blink on or off. Each bulb receives inputs from the two bulbs on either side of it telling it whether they are on or off. We can then imagine that each light bulb has a rule that it follows to determine what it does next period, based on the inputs from the other two bulbs
Basically, three variables guide the behavior of such networks.
1) the number of nodes in the network.
2) a measure of how much everything is connected to everything else. And
3) measure of “bias” in the rules guiding the behavior of the nodes.
Let’s look at each of these in turn and their implications for economic and other types of organizations.
a) Number of nodes in the network: Bigger is usually better
The number of states a network can be in scales exponentially with the number of nodes. A network with 2 nodes can be in four, or 22 , states: 00, 10, 01, and 11. Likewise, a network with 3 nodes can be in eight, or 23
The exponential growth in possible states creates a very powerful kind of economy of scale in any network of information-processing entities. As the size of a Boolean network grows, the potential for novelty increases exponentially. A Boolean network with 10 nodes can be in 210 possible states a An organization the size of Boeing also has inherently more headroom for innovation in the future— the larger number of states in the Boeing organizational network means that there are more potential ways for Boeing to make a living than for my corner coffee shop.
b) Number of connections between nodes. Complexity Catastrophes when a network is big
The mathematics of Boolean networks leaves us with a quandary, however. If large organizations have more headroom for innovation than do small organizations, then why does the mythos of business hold that small organizations out-innovate large ones?
If a network has on average more than one connection per node, then as the number of nodes grows, the number of connections will scale exponentially with the number of nodes
This means that the number of interdependencies in the network grows faster than the network itself. This, then, is where the problems start to arise. As the number of interdependencies grows, changes in one part of the network are more likely to have ripple effects on other parts of the network. As the potential for these knock-on effects grows, the probability that a positive change in one part of the network will have a negative effect somewhere else also increases.
To illustrate, let’s imagine you are the cofounder of a small start-up company with only two departments: product development and marketing. You run product development and have an idea for a new product. So you have a meeting to discuss your plan, the marketing department agrees to it, and you are ready to go—
You grew. You now have an idea for your third-generation product, but something bizarre has happened. You have your usual meeting with marketing, but now before you get the department’s OK, the marketing managers say they have to check the impact on their budget, which was approved by finance. The finance folks say they can’t approve your project until they get an estimate from customer service on the cost of the additional support needed. And customer service has to check with marketing to make sure its plans are consistent with the company’s brand and pricing strategy. All of a sudden, you have gone from three meetings to ten (if all the permutations occur)
Degrees of Possibility Versus Degrees of Freedom The problem isn’t dumb people or evil intentions. Rather, network growth creates interdependencies, interdependencies create conflicting constraints, and conflicting constraints create slow decision making and, ultimately, bureaucratic gridlock.
We thus have two opposing forces at work in organizations: the informational economies of scale from node growth, and the diseconomies of scale from the buildup of conflicting constraints. Taken together, these opposing forces help us understand why big is both beautiful and bad: as an organization grows, its degrees of possibility increase exponentially while its degrees of freedom collapse exponentially.
This tension between interdependencies and adaptability is a deep feature of networks and profoundly affects many types of systems. This tension creates upper limits on the complexity of organisms.
Hierarchies alleviates the size problem. Organizing the network into hierarchies reduces the density of connections and thus reduces the interdependencies in the network. Hierarchies are critical in enabling networks to reach larger sizes before diseconomies of scale set in. This is why so many networks in the natural and computer worlds are structured as networks within networks.
counter intuitively, hierarchy can serve to increase adaptability by reducing interdependencies and enabling an organization to reach a larger size before gridlock sets in.
Who would have thought that hierarchy actually saves meetings? Hierarchy does, of course, have its problems; for example, information can degrade as it travels up the chain, the top may become out of touch with the front line, and a poor performer in a senior role can do a lot of damage.
A related move is to give the units within a hierarchical structure more autonomy.
c) Network bias: Increasing predictability allows for more connection per node
According to Kaufman, nonhierarchical networks exhibit spontaneous order with one or two average connections per node, but went into chaos (thus creating cascades of change and the potential for a complexity catastrophe) at four connections per node or more.
Bernard Derrida and Gerard Weisbuch, discovered a parameter that could change the point at which this phase transition takes place. They called the parameter bias.
Let’s say we pick a bulb and start feeding it Is and Os at random. Our input stream will thus be approximately 50 percent Is and 50 percent 0s. If the output stream was also fifty-fifty Is and 0s, then we could say that the output was unbiased, but if the output was say 90 percent Is, then we could say it was biased toward 1. A high-bias node is easier to predict.
Derrida and Weisbuch found that the higher the bias, the more densely connected a network can be before the transition to chaos occurs. If the average bias is fifty-fifty, then the transition to chaos happens in the range between two and four average connections per node, as it did in Kauffman’s study If the average bias is closer to 75 percent, then the transition happens above four connections per node. At higher bias levels, the network can go up to six connections per node before it trips into chaos. The key point is, the more regularity there is in the behavior of the nodes, the more density in connections the network can tolerate.
In an organizational context, we can think of bias as being a measure of predictability. If there is predictability in the decision making of an organization (the equivalent of the light bulbs’ rules), then the organization can function effectively with a more densely connected network. If, however, decision making is less predictable, then less-dense connections, more hierarchy, and smaller spans of control are needed. Thus, for example, in an army, where regular, predictable behavior of troops is highly valued, it might be possible to get away with larger unit sizes than, say, in a creative advertising agency. Factors that make behavior less predictable, such as office politics and emotions, can limit the size an organization can grow to before being overwhelmed by complexity
If we combine Kauffman’s original result with the later results on hierarchy and bias, however, the phase transition shifts to the range of six to nine nodes. Interestingly, the numbers that come out of the analysis of Boolean networks are quite close to what we typically see for the size of effective working groups in human organizations
Depressions, recessions, and inflation are not exclusively modern phenomena; they are patterns that have recurred since the beginning of recorded history. Time series have a not-quite-regular, not-quite-random character to them. Economists have had very little success in trying to use the irregular historical patterns in economic
The ultimate accomplishment of Complexity Economics would be to develop a theory that takes us from theories of agents, networks, and evolution, all the way up to the macro patterns we see in real-world economies. Such a comprehensive theory does not yet exist, but we can begin to see glimmers of what it might look like. Such a theory would view macroeconomic patterns as emergent phenomena, that is, characteristics of the system as a whole that arise endogenously out of interactions of agents and their environment.
(2) Lesson from a Beer Industry simulation: business fluctuations does not flatten out, they goes through storm and calm
d) The game setup: 4 players are retailer, distributors and manufactures of beer. Compete to lower their cost
Four volunteers are asked to play a game simulating the manufacture and distribution of beer. The game is played as follows. Each participant has an inventory of cases of beer (represented by chips on the game board). At the beginning of each turn, the retailer turns over one card from the deck to get the order from the consumers (e.g., four cases) and then submits an order to the wholesaler. The wholesaler looks at his or her orders from the retailer and submits an order to the distributor,
1) Players incur costs of $0.50 per case for holding inventory (e.g., the cost of storing and securing the beer), and costs of $1.00 per case for running out of beer (e.g., angry customers and lost sales). Therefore,
2) The winner of the game is the one who incurs the lowest cost.
3) Likewise, there is a small delay between when orders are submitted and when they are processed
4) Finally, no communications are allowed between the players other than through the orders. Thus, the brewer doesn’t know what the customer demand is down at the retailer’s end;
The game starts out in equilibrium, with each player getting an order for four cases of beer and shipping exactly that many.
Unbeknownst to the participants, the first several cards in the consumer deck remain at four. Some players may order a bit more or a bit less than four, depending on how risk averse they are. Otherwise, not much happens. Then, suddenly, on one turn, the consumer-order card jumps from four to eight. The players do not know it, but the customer-order level will stay at eight for the rest of the game.
e) The results: a shock in the environment induces a shock in inventories that is amplified over time
According to Traditional Economics, this exogenous shock in demand should simply cause the players to move to a new equilibrium after a few turns of adjustment, with everyone ordering eight, and everyone’s inventories staying constant
In experiments with real people, however, the players inevitably overreact to the jump in demand by over-ordering as their inventory falls. As this wave of over-ordering travels up the supply chain, it is amplified.
Just what kind of behavior leads to such wild oscillations in a relatively simple environment? Sterman has been able to statistically derive the decision rule used by the participants This rule is based on a behavior known in the psychology literature as anchor and adjust. Rather than deductively calculate their future beer needs by looking at all the inventory on the board (which they can see) and incorporating the effects of the time delays and so on, the participants simply look at the past pattern of orders and inventory levels, and inductively anchor on a pattern that seems normal. Their IF-THEN rules consequently try to steer them to maintain that normal pattern. Thus, a participant might anchor on four cases as the normal pattern of orders and then struggle to adjust when things are not normal, for example, “My inventory is dropping, order more!” In an environment with time delays, the anchor-and-adjust rule causes individuals to both overshoot and undershoot, which in turn leads to the emergent pattern of cyclical behavior.
Humans don’t do well when there is a time delay between their actions and the response to those actions
There are two ways to dampen the cycles in the Beer Game: one is to reduce the time delays, and the other is to give the participants more information (e.g., giving the brewer direct visibility into what is happening at the retail level).24
a) What is punctuated equilibrium: phases of calm and storm
Over the century that followed the publication of Darwin’s Origin of Species, biologists assumed that evolution proceeded in a stately and relatively linear fashion, leading to a smooth pattern of speciation and extinction. Then, in a landmark paper in 1972, the paleontologists Niles Eldredge and Stephen Jay Gould overturned this conventional wisdom and argued that the fossil record shows that biological evolution has not followed a smooth path.2 8
For example, during the Cambrian period about 550 million years ago, a burst of evolutionary innovation saw the takeover of the earth by multicellular life and the creation of most of the major phyla on earth today. Then, about 245 million years ago, during the late Permian period, there was what Gould called “the granddaddy of all extinctions”; when 96 percent of all marine species on earth disappeared
Patterns of punctuated equilibrium show up not just in biological evolution, but in other complex systems ranging from the slides of avalanches to the crashes of stock markets.
Many types of networks self-organize into a structure that has a mixture of very dense connections and very sparse connections. It is just such a network structure that underlies the emergence of punctuated equilibrium in biological ecosystems
The researchers found that if they randomly removed “species” from their simulated ecosystem, typically not much happened. Yet, once in a while, removing a random species would set off a cascade of events leading to a mass extinction. Certain species are very densely connected to other species in the web of food relationships and niche competition. Biologists call these “keystone species”.
b) Technology evolves following evolution patterns: i.e. with emergent properties
Technological innovation proceeds in similar patterns of calm and storm.
Technologies are inherently modular: a car, for example, is made up of an engine, a transmission, a body, and so on. Modules are then assembled into “architectures,” in this case, the design of the car itself
It is innovations in architectures (e.g., the PC revolution itself) that tend to have the big catalyzing ripple effects on innovation. We thus have two of the key features that led to the punctuated equilibrium pattern in Jain and Krishna’s model—sparse-dense networks of interaction, and catalyzing effects from individual nodes.
Technology webs might be subject to cascades of change leading to the emergent pattern of punctuated equilibrium, and that certain technologies could play the role of keystones in those webs.
Several researchers have shown statistically that stocks do not follow a random walk. The clumpy pattern for IBM stock price shows that the volatility of price movements is correlated in time. This is the stormy-quiet-stormy pattern of punctuated equilibrium A few points skyrocket above, or plunge far below, the rest of the sample. What could lead to such dramatic movements in prices?
The surfacing of news does not explain much of the swing in prices. We have a mystery: why is there so much news-less volatility in the market? The answer to this mystery lies in an interesting observation: while stock price movements don t look much like a random walk, they do look like another phenomenon: earthquakes.
The following straight line on a log-log scale meant that, with earthquakes, there is no “typical” size in the middle of the distribution as there is in body heights. Rather, earthquakes occur across all size scales, but the bigger the quake, the rarer it is—specifically, with each doubling in earthquake energy, the probability of a quake of that size occurring drops by a factor of four. It is thus a slippery slope down
Physicists call this kind of relationship a power law, because the distribution is described by an equation with an exponent, or power.42 Power laws have been discovered in a wide variety of phenomena, including the sizes of biological extinction events, the intensity of solar flares, the ranking of cities by size, traffic jams, cotton prices, the number of fatalities in warfare, and even the distribution of sex partners in social networks.43 Power laws, along with oscillations and punctuated equilibrium, are another signature characteristic of complex adaptive systems.
Pareto’s study of income found a lot of poor, a middle class that stretched over a wide range, and a very few superrich. He found that for every increase of income by 1 percent, there was a corresponding decline in the number of households by 1.5 percent—graphed on log-log paper, this produces a straight line—a power law. Pareto
Power laws reemerged briefly in economics in the 1960s, when Benoit Mandelbrot became interested in the fluctuations of cotton prices on the Chicago Mercantile The fluctuations seemed to have no natural timescale. If he took one section of the graph, say, one hour, and stretched it out to the length of a day, one could not tell which graph was the hourly data and which was the daily data. He then looked at data from other commodities, including gold and wheat, and saw the same pattern—power laws
They found that the fluctuations in stock prices follow clear power laws in the tails of the distribution.
One of the consequences of this result is that financial markets are far more volatile than Traditional Economics leads us to believe.
The size of companies as measured by employees also scales according to a power law. Company sales growth, as well as the GDP growth of nations, likewise scales according to a power law.
The key is that there are two types of trades one can make on most stock exchanges. The first is a market order, in which a trader says buy (or sell) stock X right now for the best available price. The second is a limit order, in which a trader says buy stock X if the price falls to $100 (or conversely, sell stock X if the price rises to $100)
The cause of large price fluctuations was the structure of the order book itself—large fluctuations occurred when there were large gaps between the price levels in the book.
The regularity of the order pattern implied that there was also some regularity in the behavior of the traders placing the orders—a result at odds with the Traditional theory that all trading is driven by unpredictable news events.
(6) Conclusion: pattern behavior emerges as consequence of: individual behavior regularities; institutional traits; exogenous inputs
Complex emergent phenomena such as business cycles and stock price movements are likely to have three root causes:
- The first is the behavior of the participants in the system. As we have seen, real human beings have real behavioral regularities, whether it is the anchor and adjust rule of the Beer Game participants, or the yet to be understood regularity that leads to student distributions in stock ordering.
- Second, the institutional structure of the system makes a big difference. In the case of the Beer Game, the structure of the supply chain between the manufacturer and retailer created dynamics, that when combined with participant behavior, led to oscillations. In the case of the stock market, the structure of the limit order system, when combined with trader behavior, led to power law volatility.
- Third and last, are exogenous inputs into the system. In the case of the Beer Game it was the onetime jump in customer orders, and in the case of the stock market it is news.
Business Plans are instructions for creating businesses that can be implemented by qualified Business Plan readers. These instructions bind Physical Technologies and Social Technologies together into modules under a strategy.
Business Plans are differentiated through the deductive-tinkering of agents as they search for potentially profitable plans. While the distribution of experiments created by this process differs from the purely random differentiation of biological evolution, it nonetheless feeds the evolutionary algorithm with a superfecundity of Business Plans for selection to act on.
At some point the plans are implemented and the market renders its judgment. Finally, successful modules are rewarded by gaining influence over more resources.
Following the framework I have just outlined, we can reinterpret markets as an evolutionary search mechanism. Markets provide incentives for the deductive-tinkering process of differentiation. They then critically provide a fitness function and selection process that represents the broad needs of the population (and not just the needs of a few Big Men). Finally, they provide a means of shifting resources toward fit modules and away from unfit ones, thus amplifying the fit modules’ influence.
Markets win over command and control, not because of their efficiency at resource allocation in equilibrium, but because of their effectiveness at innovation in disequilibrium.
the reason that markets work so well comes down to what evolutionary theorists refer to as Orgel’s Second Rule (named after biochemist Leslie Orgel), which says, “Evolution is cleverer than you are.” Even a highly rational, intelligent, benevolent Big Man would not be able to beat an evolutionary algorithm in finding peaks in the economic fitness landscape.
The reason that markets are good at allocation has more to do with their computational efficiency as a distributed processing system (i.e., they get the right signals to the right people), than with their ability to reach a mythical global equilibrium.
In 1971 Georgescu-Roegen’, published his magnum opus The Entropy Law and the Economic Process, where he stated that that economic activity is fundamentally about order creation, and that evolution is the mechanism by which that order is created.
The Second Law thus provides a basic constraint on all life: over time, energy inputs must be greater than energy expenditures. All organisms must make a thermodynamic “profit” to survive and reproduce. The design for an organism can be thought of as a strategy for making thermodynamic profits long enough to reproduce, before the Second Law eventually catches up.
Competition for the energy and materials needed for order creation is, of course, intense; plants compete for ground, water, and sunlight, and many species have the strategy of stealing energy and materials from other species by eating them.
Just as in biological systems, the economic process materially consists of a transformation of high entropy into low entropy.
A pattern of matter, energy, and or information has economic value if the following three conditions are jointly met:
1. IRREVERSIBILITY. All value-creating economic transformations and transactions are thermodynamically irreversible.
2. ENTROPY. All value-creating economic transformations and transactions reduce entropy locally within the economic system, while increasing entropy globally.
3. FITNESS. All value-creating economic transformations and transactions produce artifacts and or actions that are fit for human purposes.
Consequently, low entropy might indeed be necessary for something to have economic value, but defining what kinds of order are valuable and what kinds are not seems rather subjective—order is in the eye of the beholder.
Taken together, the three G-R Conditions say that economic activity is fundamentally about order creation. Faced with the disorder and randomness of the world, humans spend most of their waking hours ordering their environment in various ways to make it a more hospitable and enjoyable place. We order our world by transforming energy, matter, and information into the goods and services we want, and we have discovered the evolutionary Good Trick that by cooperating, specializing, and trading, we can create even more order than we otherwise could on our own.
They select forms of order that meet their needs, fulfilling drives and preferences
In physics, order is the same thing as information, and thus we can also think of wealth as fit information; in other words, knowledge.
Information on its own can be worthless. Knowledge on the other hand is information that is useful, that we can do something with, that is fit for some purpose.
Evolution is a knowledge-creation machine—a learning algorithm.53 Think of all the knowledge embedded in the ingenious designs of the biological world. A grasshopper is an engineering marvel, a storehouse of knowledge of physics, chemistry, and biomechanics—knowledge that is beyond the bounds of current human ability to replicate
A grasshopper is also a snapshot of knowledge about the environment it evolved in, the foods that were good to eat, the predators that needed to be defended against, and the strategies that worked well for attracting mates and ensuring the survival of progeny. There are terabytes of knowledge embedded in a single grasshopper.
We have found the answer to our quest. Wealth is knowledge and its origin is evolution.
 Sterman, J. D. 1985. A Behavioral Model of the Economic Long Wave. Journal of Economic Behavior and Organization 6:17-53.