Economy, Evolution, and Intelligent Design

Steven Dutch, Natural and Applied Sciences, University of Wisconsin - Green Bay
First-time Visitors: Please visit Site Map and Disclaimer. Use "Back" to return here.

Adaptive and Top-Down Systems

About the time biologists were grappling with the question of how different species developed, economists were facing a puzzling question of their own. Here they had this complex economy delivering goods and services at undreamed-of levels of output and variety, with remarkable efficiency, and nobody was in charge of it.

It turns out the two questions are connected in a deep way because both market economies and evolution involve the interplay of conflicting forces that interact in very complex ways. They are examples of complex adaptive systems in which objects in the system change their behavior in response to the forces around them. We can contrast complex adaptive systems with top-down systems in which the behavior of the system is dictated from above by some supreme planner. In economics, capitalism was more of an adaptive system, while communism strove to be a top-down system. Evolution, the adaptive model of species development, stands in contrast to Intelligent Design, a top-down model.

The terms adaptive and top-down apply only to the overall forces that direct the system. Even in adaptive systems there are top-down elements at the micro-level. Market forces may dictate prices, but customers don't generally come into a store and tell the cashier what they're willing to pay. The store sets the prices and that's that. In nature, herds and packs go where the leader decides. Prey organisms don't get to vote on whether there are predators. And in top-down systems there are adaptive elements as well, although that limitation has an ominous potential for change.

Human Systems

No human system is purely adaptive because humans are [semi-]intelligent and therefore you can (and Intelligent Design proponents do) argue that all human systems are Intelligently Designed, thereby denying any analogies between human systems and natural ones. However, intelligence within the system doesn't make it Intelligently Designed. Animals, after all, use their intelligence to attract mates, evade predators, and find food, and higher mammals go through a long period of dependency to hone this intelligence in the young. That doesn't make a wolf pack or a herd of antelope an example of Intelligent Design.

Human intelligence may apply to individual decisions, but individuals vary so much in their values, tastes, and desires that their individual decisions are statistically almost random. Adam Smith modeled his market economy on the notion that buyers and sellers had complete and accurate information and used it rationally. Even in that case, something like the price of vegetables will be dictated by individually unpredictable matters of taste like how many people like peas versus beans, and intangible and unpredictable factors like, "normally the market down the hill offers better goods, but that road is muddy today so I'll go uphill instead." Thus, even if the buyers and sellers have perfect information, there will still be a huge element of statistical randomness in the economy. And of course we now know that people don't strive for perfect decisions, but strive for decisions that are "good enough" instead, because the cost of achieving perfection isn't worth the extra effort. It's the old 80-20 rule: if the first 80% of a task takes only 20% of the time, why not just settle for 80%?

A system can only be said to be Intelligently Designed if it is directed by an intelligence that has enough information and expertise to run the system effectively. A flock of sheep is not an intelligently designed system. A flock of sheep maneuvered by a sheepdog is closer, although the intelligence is not the dog's but the farmer who trained her.

Some Successful Complex Adaptive Systems

A good example of a complex adaptive system is standardization agreements. You can buy a ream of copy paper made in Maine, mix it with paper made in Oregon, and the paper will be indistinguishable. You can buy a 1/4 inch nut in Florida and screw it onto a 1/4 inch bolt from Alaska and the threads will mesh. This is because there are literally tens of thousands of agreements among manufacturers so that one company doesn't make 8-1/2 x 11 paper while the next makes 8-7/16 x 11-1/8. In the 1970's, someone in the Government realized that, horror of horrors, all of this was going on with no Government oversight. So Congress held some hearings and came to the conclusion that trying to regulate the process would bring American industry to a standstill. Even Ralph Nader, who hardly ever met a top-down regulation system he didn't like, agreed. This might be as close to a pure adaptive system as we find in the economy.

Maybe the purest adaptive systems created by humans are the on-line futures markets in elections. Once touted as extremely accurate measures of opinion, such that the infamous upset of Harry Truman over Thomas Dewey could never happen again, polls turn out to have all sorts of flaws. Many people refuse to talk to pollsters, or tell pollsters what they think the pollster wants to hear, or deliberately lie. Bettors in on-line futures markets buy and sell shares of election outcomes, bidding prices up or down, and they have actually outperformed polls. Bettors have a financial stake in the actual outcome - even if the winnings are trivial there's a competitive stake - and buy or sell shares according to what they really think will happen, not what they want to happen. So far. We can anticipate people trying to control markets, stampede prices up or down, manipulate results in an effort to sway real-world voting, and so on. It's only a matter of time.

The corrective process in science is also a complex adaptive system. Yes, there are herds in science, and fields where leaders disproportionately influence results, but the Alpha Leader in stellar astronomy doesn't have any influence at all over the organic chemistry herd, and the Alpha Leader of organic chemistry has nothing to say to the particle physics herd. Science as a whole is extremely interconnected, but since it takes a lifetime to become proficient in even a small area, science is extremely decentralized. The reason it doesn't seem so to the average crank scientist is because the gulf between crank science and even the lowliest real scientist is huge. You might as well put on an elk suit and try to lead an elk herd.

The Problem of Subversion

Human systems have one failing that natural adaptive systems don't. Humans know there's a system, and that if they take it over, they can abuse it to their own advantage. Humans are constantly trying to replace complex adaptive systems with top-down systems because top-down systems seem to offer the advantage of allowing one All-Knowing Leader (generally the person or movement attempting to take it over) to make the system run better than that complex, fumbling, trial and error natural system. Elk may try to displace the Alpha Bull Elk, but they don't think: "Hey, after I take over this elk herd, I can use the same methods to take over the herd across the valley. And maybe I can even take over a buffalo herd. Once I unite all herbivores under my wise leadership, we can starve off all the predators and have better grazing."

As much as humans profess to hate top-down systems, what they generally want is not to replace them with adaptive systems, but to substitute their own top-down systems instead. Thus people who hate corporate power aren't willing to let market forces work to improve energy efficiency or impacts on the environment, but hope to impose their own system of top-down decrees and mandates to do the job. Atheists who hate the authority of religion want to impose restrictions from above on religion, and people who complain about restrictions on religious freedom are often really upset that religion has been stripped of its former ability to dictate the rules of society. To both groups, "freedom" really only means freedom to restrict the freedom of others, and inevitably it degenerates into using authority to gratify personal desires. If our hypothetical intelligent elk above did manage to unite all herbivores, the plan to starve out the predators would quickly take a back seat to "I can mate with all the females I like."

The fact that human systems can be taken over by members of the system means that no human system can ever be purely adaptive. Every human system includes top-down features designed to prevent the system from being taken over. These top down elements don't make the system as a whole a top-down system unless they are run by people who have enough information or power to command the entire system effectively. So we allow the free market to function but circumscribe it with regulations intended to prevent excessive control by small segments of the society. Of course, the people who want to control the system for their own purposes invariably try either to control the regulators, neutralize them, or evade them. Often they succeed to the extent that what was once a reform becomes an impediment if not an actual vehicle for abuse. Thus the legal reforms intended to give plaintiffs a fairer shake in court have devolved into tools of harassment and extortion, and reforms intended to protect defendants have devolved into a network of loopholes that more or less guarantee that defendants wealthy enough to hire skilled attorneys can escape serious punishment. The cat and mouse game between regulators and those who want to co-opt the system is, itself, an adaptive system.

Wikipedia is a good example of a system that was intended to be adaptive but ran into the problem of subversion because of a lack of checks and balances. The theory was that knowledgeable readers would correct errors and add better information, and a purely adaptive process would result in better content than any editorial board could. The reality was that once the wrong people discovered Wikipedia, they discovered they could inflate their egos by pretending to be experts on subjects they knew little about, could suppress unflattering or inconvenient information, slander people they disliked, present themselves more favorably, and generally do all the obnoxious things people do on line, except that instead of promoting themselves on an obscure personal Web site, they could do it on Wikipedia. Wikipedia has responded to these threats by imposing restrictions on the editing of sensitive pages and by installing a feature to enable readers to learn the identity of editors. The rapid cycle of editing pages by users who spot malicious or incompetent editing still gives Wikipedia a high level of adaptive behavior.

Paradoxes of Democracy

This interaction of top-down and adaptive components results in what I call the Paradox of Authority-in-Democracy. Overall, despite its failings, we live in a democratic society. (Ron Paul and Ralph Nader types disagree mostly because it doesn't do what they want to impose on society.) Nevertheless, most of the day to day decisions in our society are made by authoritarian means. Students don't elect their teachers or vote on grading standards, workers don't elect their bosses, we don't elect our police officers or set speed limits by market means. Yet there are enough adaptive mechanisms to keep authority in check. For one thing, we are so indoctrinated in democracy that lots of authority figures at least seek consensus before making decisions. Companies with bad bosses have high turnover. If police forces get out of line, they may find themselves dealing with less sympathetic legislators. The whole Constitutional checks and balances process is an adaptive system, and the worst problems in society result from failures or absence of checks and balances: corrupt religious organizations, inept workers with excessive job security, or companies that don't rein in bad bosses because they can always hire new workers.

Actually, the Constitution nowhere guarantees a democracy. The Constitution specifies (Article IV, Section 4): "The United States shall guarantee to every State in this Union a Republican Form of Government (emphasis added)." The difference is that in a republic, decisions are made by elected representatives. James Madison, in Federalist Paper 51, gave about as clear a statement of the problem of subversion of an adaptive system ever stated: "It is of great importance in a republic not only to guard the society against the oppression of its rulers but to guard one part of the society against the injustice of the other part. If a majority be united by a common interest, the rights of the minority will be insecure." Indeed, the Constitution contains a host of deliberately designed stumbling blocks to hobble brute force majority rule and require responses to opposing factions. The separation of powers, two chambers of Congress, electoral college, and super-majorities required for Constitutional amendments are examples.

The Third Option: Subversion From Below

In addition to top-down and complex adaptive systems, there's a third possibility. Adaptive human systems can be taken over from above, but they can also be subverted from below. Indeed, there's even a biological counterpart called mobbing, in which smaller organisms gang up in large numbers against a predator. Most people have seen flocks of small birds chasing hawks or crows. If the larger bird could think rationally, it would realize the mob of small birds can't possibly do it any harm, would counterattack and get an easy meal. The mobbing behavior, in turn, is only a response to a specific threat. It stops once the threat goes away. The small birds can't think rationally and decide to suppress predators completely.  Humans can. The result  is that animals that pose even slight threats to human interests are exterminated, herbivores proliferate and in turn are culled, and we get the sorts of ecological problems we are all too familiar with.

But overly democratic systems don't merely suppress threats. They suppress any feedback mechanisms that prove inconvenient or painful so that the lowest common denominator triumphs. Everyone's vote counts the same, everyone's money has the same value. In the case of early Wikipedia, everyone's input had the same value. The failings of Wikipedia illustrate a problem with democratic systems. They are great for finding solutions to complex problems that satisfy the majority, at least for the moment. They don't necessarily guarantee the best solutions. Market forces eventually gave us VCR's. They failed to deliver the triumph of Betamax, which by most accounts was technically superior to VHS. Indeed, it's not hard to compile a list of products that triumphed in the market over technically superior products. Wikipedia in theory would achieve the best possible information; in reality it often generates what people want to believe. And don't get me started on television ratings or movie box office results. It's often said that in a democracy, voters get the government they deserve. The problem is that people who know better get the government the voters deserve as well.

Both excessive authority and excessive democracy interfere with adaptive systems by stifling the feedback that makes adaptive systems function. Overly centralized authority stifles any feedback that threatens the rule of the authority, questions its wisdom, or interferes with what the authority wants to do. Excessively democratic systems stifle feedback that penalizes irresponsible behavior or rewards achievement.

The Futility of Top-Down Thinking

So far, no large scale human system has ever been exclusively top-down, simply because no ruling class, however ruthless, has had complete and perfect information, or the ability to enforce decisions microscopically. Generally even the most rigid authoritarian system leaves people free to make minute decisions that don't threaten the stability of the system, or to make decisions it isn't practical to make from higher levels. A much-cited example was the Soviet nail factory. If its output quota was set in terms of weight, it produced large nails. If it was set in terms of numbers of nails, it made small nails. What it didn't do was make the number and sizes of nails consumers needed, because there was no way for those needs to be communicated back to the nail factory. In theory, dissatisfied consumers could write to the Ministry of Production, which, in theory, could act on the complaints. In reality it didn't happen. If there had been numerous independent nail factories each deciding independently what kinds of nails to make, a manufacturer who failed to make the desired product would have found out quickly that customers were turning to competitors instead. That would have had far more effect than any number of memos from the bureaucracy.

As Ludwig von Mises pointed out in 1922, planned socialist economies failed the most fundamental economic task of all: setting prices. This was after the Bolshevik Revolution, but well before the post World War II Soviet expansion, and at a time when there was already more than enough observational evidence that the experiment was failing. Socialist economies essentially plagiarized their prices from capitalist societies.

There's abundant evidence that the progress of Western technology is directly related to concepts of individual freedom and autonomy, and laws to protect them. Imagine trying to invent anything in a world ruled by the mentality sketched by Bernard Lewis:

At a time and place and place where neither law nor custom restricted the powers of even local authorities, visible and mobile assets were a poor investment. The same fear of predatory authority [top-down rule] - or neighbors [bottom-up rule] - may be seen in the structure of traditional houses and quarters: the high, windowless walls, the almost hidden entrances in narrow alleyways, the careful avoidance of any visible sign of wealth.

The spread of computers makes it possible to imagine a day when rulers do have complete and perfect information, plus the ability to enforce decisions microscopically. It might even incorporate safeguards to prevent the ruling class from going slack and losing its grip-. That's scary. The one thing we can be sure of is it won't work better than an adaptive system except in terms of staying in power. In Bernard Lewis words, it will "have failed every test except survival."

Victor Gruen and Malls

Victor Gruen was an Austrian architect who fled the Nazis and came to the United States. Gruen conceived of shopping malls as a way of recreating the town square life of Europe. In classic top-down thinking mode, it never seems to have occurred to him to ask whether Americans wanted that sort of lifestyle, or, if they did, why it hadn't already evolved here.

If you spend a little time hanging around a town square in Europe, you'll discover quickly that all those quaint tables are for people who buy something. Yes, there are benches and fountains just like there are in shopping malls, but European shopkeepers don't put out tables for people to sit at for free, any more than shopkeepers in American malls do. Also, the sorts of people who have the time to hang around all day in malls aren't the sorts of people shopkeepers or shoppers want hanging around in malls. The people who sit around European plazas all day chat and don't bother anyone else; the people who hang around in malls all day aren't the same.

So the outcome was inevitable. Once Gruen introduced shopping malls, they evolved. Americans took them in typically American directions. Gruen moved back to Austria, bitterly disappointed and wondering what went wrong.

Mass Transit

For top-down thinking at its worst, it's hard to top mass transit. Transit companies raise fares, design convoluted and inconvenient routes, set infrequent schedules, decide that drivers will not make change, and then they wonder why nobody rides the train or takes the bus. It never seems to occur to mass transit advocates to ask people why they don't use mass transit. A little thought will supply the answers: cost and inconvenience.

For 150 years, we have seen the evolution of a complex adaptive system in action in the domain of personal mobility. It actually began with bicycles and interurban railroads, both of which created their own selective pressures. Bicycles inspired a demand for greater speed and range, less physical effort, more cargo capacity and weather protection, which were answered by early automobiles. Interurban railways also created demands for flexibility, which were answered not so much by automobiles as by buses. Contrary to popular mythology, the cartel of companies that bought up and closed most of America's interurban railroads were motivated by a desire to sell buses, not cars. Although railways have certain advantages, they are expensive and extremely disruptive to traffic to build and can't easily be rerouted to meet changing needs. And street railways had abysmal profit margins.

So when people call for mass transit to replace automobiles, they set up a head to head contest between a highly evolved complex adaptive system and a top-down design, and it's no contest. Making matters worse is the fact that mass transit advocates display top-down thinking at its most rigid and unimaginative. Most mass transit advocates think only in coercive terms: higher taxes on private vehicles, tolls, and so on. They never seem to think of, say, lower fares, denser and more direct routes, or more frequent schedules. (One exception are the no-fare zones in some central cities.)

A Present-Day Top-Down System in Action

Actually, for top-down thinking at its worst, it is - amazingly - possible to top mass transit. If there's one area where American capitalists have completely forgotten everything they ever knew about market forces, it's in computer software. Nothing illustrates the futility of top-down thinking better than computer software.

Until mass-market software began to emerge in the 1980's, computer software development was much like an adaptive system. A lot of it was written by end users and modified to suit their needs. Bugs that interfered with getting results were fixed. Essential features that were lacking were added. Often the working nucleus of the program would be written first, with the interface built around it after experience had shown what was useful and desirable. The early evolution of computers was nicely adaptive. Once it became apparent how powerful computers were, hard-wired circuits (First Generation) gave way to punch-card entry of data and programs (Second Generation). Second Generation programming was extremely unforgiving - even a misplaced entry on a card would cause a crash, but in response to the need for better and easier programming and more efficient use of computer time, by the Third Generation, programming languages like BASIC and time sharing made programming about as adaptive as it has ever been.

The peak of adaptive behavior was reached in the early Fourth Generation with the appearance of PC's that allowed individuals to own a computer and program on it. For a short time there was even a commerce in individually written programs (as there still is in shareware). Computer purists sniffed that programming had sped up dramatically only because improved computer speed and access made it possible to "waste" more computer time more cheaply. This mentality ignored the fact that computers were made to serve people, not the other way round, and flew directly in the face of the fact that the essence of an adaptive system is experimentation and weeding out of unsuccessful experiments. But the purists would have the last laugh.

The purists began to win in the 80's. First of all, Nicklaus Wirth decreed that the GOTO statement was bad because people used it to create spaghetti code. It seems never to have occurred to him the problem was instead, first, that text editing of programs in those days was hopelessly primitive, so that often the only alternative to a total rewrite was a GOTO, and second, stupid people should not be allowed near computers. (I argued then, and still do, that all college computer programming courses should require two years of calculus as a prerequisite simply to guarantee the programmers will be mathematically literate.) The IF...THEN...ELSE construction eliminated a good 90% of GOTO's. The stupid people, on the other hand, were allowed to breed and take over computing. Instead of allowing inept programmers to adaptively learn good programming or be naturally selected against on the job, programming gurus opted to create a system that would eliminate adaptiveness. The paradigm that arose in computing to allow stupid people to survive was called top-down programming. The idea was that programs would be properly structured from the beginning and that stupid people would have to declare variables and create an interface before being allowed to do anything else. (As one wag put it, "Strong typing is for weak minds." It is a hallmark of top-down thinkers that they blame problems on externalities rather than the humans who actually create the problems.) On the other hand, even the most primitive version of BASIC can do any operations whatsoever because it has been shown that all possible computer programs can be written if the language includes stepping, branching, and iteration. Old-fashioned line-number BASIC would be perfectly functional today - and vastly preferable to much of what replaced it - if it were updated to allow interfaces with modern peripherals, make use of modern high resolution displays, and be compiled so the programs could run independently.

One problem with top-down programming was that it replaced simple languages that could yield immediate results with complex ones that required lengthy preliminaries and a long learning curve. Consider the simple "Hello, World" program. In old-fashioned BASIC, it looked like this:

10 REM Hello World in BASIC
20 PRINT "Hello World!"

Wirth's "improved" Pascal looked like this:

{Hello World in Pascal}
program HelloWorld(output);
  WriteLn('Hello World!');

But, oooh, not only is there no GOTO, but you can indent lines in Pascal. Seriously, I saw a text once that touted that as one of its biggest advantages. What, there is a GOTO in Pascal after all? So what was the point?

Here's "Hello World" in C++

// Hello World in ISO C++
#include <iostream>
#include <ostream>
int main()
    std::cout << "Hello World!" << std::endl;

The length of the programs is the least of the problem. The BASIC program requires you to learn only two commands before you can write the program, and the syntax is intuitively obvious. I count 13 commands and symbols that have to be learned in C++ just to print a line of text on screen. Intuitive? What does "std::cout" mean?

I think it's quite obvious that most of the abysmal quality of present-day software is traceable directly to top-down programming and thinking.

Even though, in theory, software should be subject to market forces, in practice the design of software is almost entirely top-down. The bureaucracy of the typical software company, and the fact that development is carried out by "teams" (a euphemism for "committee") pretty much guarantees that no individual innovation from within will be implemented. The concentration of market share in a relatively few hands eliminates competition and channels it into mostly trivial matters like appearance of the interface, or relieving the uninformed user of the need to make decisions or become technically informed, rather than fixing serious operational flaws. Thus, Windows has a master folder called "My Documents" for the benefit of people who can't master the idea of directories, but to this day lacks a feature to generate a text listing of files in a directory. It's adaptive all right, like those island ecosystems where the birds all evolve flightlessness because it's so much easier than flying, and it works great until somebody brings cats ashore. Design decisions are made almost entirely by marketers, and by programmers and executives who have a preconceived notion of how users should behave and know what users want better than the users do. The programmers may be technically proficient at writing code but they don't seem ever to use their programs for their ultimate purpose. At best they do some superficial tests to see if their routines work. And the proof is the abundance of irritating bugs that would be a trivial matter to fix except there is no mechanism for effectively suggesting it. Finally, copyright restrictions make it impossible for users who could make improvements from doing so. Pretty much every possible adaptive process has been stifled.

The flightless bird model of software design leads to too many absurdities to count. These are shortcomings that are so amateurish that any programmer who was actually using his own work to get real results would fix them immediately.


The record of top-down systems is clear and wretched. In politics, in society, in economics, in religion, in computing, top-down systems are abject failures.

Even the military, the archetypical top-down management system, is far more adaptive than it once was. Overall objectives come from above, but once given, the decisions about specifically how to achieve them are often left to lower echelons to decide. In history, whenever disciplined and adaptive forces (as opposed to mobs) went up against rigidly organized forces, the results were usually catastrophic for the top-down force. Ask Publius Quinctilius Varus who lost three Roman legions (and his life) after being ambushed by German tribes at the Teutoberg Forest in 9 AD, or Major General Edward Braddock who pretty much re-enacted the Teutoberg Forest in Pennsylvania 1746 years later. Or George Armstrong Custer. Or the U.S. in Vietnam, where an officer and NCO corps as corrupt, ossified and rigid as we ever had tried to deal with unconventional warfare.

A real Intelligent Designer would not create a top-down system because it's simply incompetent design. Top-down systems are not Intelligent Design. A real Intelligent Designer, even a perfectly informed and omnipotent one, would build an adaptive system because they work better.

If it's reasonable to infer from human experience that complexity and order imply an Intelligent Designer, it is equally reasonable to infer that an Intelligent Designer would use complex adaptive systems rather than top-down methods. Top-down design is not intelligent design.

Yes, but what if the designer is so intelligent that it could design a perfect system a priori? If Intelligent Design is supposed to be scientific, and not supposed to be making any theological assumptions, then where would you get the assumption that the Designer has perfect intelligence? Why couldn't an Intelligent Designer make mistakes or use trial and error? But even if we assume the Designer is perfectly intelligent, why wouldn't the designer still use complex adaptive design? The Designer might very well see that adaptive methods would yield results as good as top-down design, or might be better for complex, intertwined objectives. You might be able to design a perfect hummingbird suited to a particular flower, but to design an ecosystem of thousands of species all optimized to function together, complex adaptive methods might work more efficiently by anybody's criteria. And if the Intelligent Designer chooses to use complex adaptive methods, who exactly are you to say that's not the best choice?

"Intelligent Design" is not intelligent design. Evolution is intelligent design.

Return to Pseudoscience Index
Return to Professor Dutch's Home Page

Created 12 March 2007;  Last Update 27 January, 2012

Not an official UW Green Bay site