The generally accepted view is that markets are always right - that is, market prices tend to discount future developments accurately even when it is unclear what those developments are. I start with the opposite point of view. I believe that market prices are always wrong in the sense that they present a biased view of the future.During the Second World War, Albert Einstein at the urging of Leo Szilard wrote a now famous letter to Franklin Roosevelt. This letter, expressing Einstein's view that nuclear fission could be used to build a weapon and that Germany might well be pursuing such a direction, set in motion a chain of events which led to the Manhattan Project and the first and only use of atomic weapons by the United States against Japan. The abstract subject of nuclear physics leapt to the foreground in people's thinking, influencing the creation of foreign policy and the international order among nations. Some scientists who possessed the knowledge to build the weapons were lifted out of the obscure world of academic research into the public eye. What scientists thought could and could not be done took on added significance. This scenario - the transformation of abstract knowledge into practical artifacts - is today repeating itself in an entirely new area.
-George Soros, 1987
The banking industry, long insulated from major technological change, has been hit by revolution that will alter forever the way it does its business. This revolution is a consequence of improvements in telecommunications, data processing, and, of course, the computer. A new class of people who have mastered this new technology has sprung into prominence and in several instances risen to leadership positions in major financial institutions.
Institutional survival in a highly competitive banking environment can depend on advances in computer modeling of markets and the economy, and software and algorithms, as well as telecommunications, that supply data input. Banks, which have long been hiring experts in data processing, are now hiring computer scientists, engineers, and mathematicians to help design their equipment and algorithms. They used to depend on their vendors, computer and software producers, for their internal needs. But they soon realized that to maintain a competitive advantage they had to take on the research and development responsibility themselves. Major financial services institutions now have their own research staffs examining how they can improve their data processing performance using new hardware and software. Abstract mathematics, sometimes developed to understand selective and adaptive systems, is now being applied to guide financial decisions. The sciences of complexity are impacting the business and financial world. And that impact is just beginning.
The real movers of the world economy today are the large international banks linked to each other electronically by a network that, seen as a whole, comprises the world's first global computer. In 1986 over $64 trillion was exchanged on this network, and that volume is still growing. (The other global computer, and second largest, is the U.S. military communications system.) The banking computer network is a parallel, not hierarchical, network, although it has hierarchical components. Within each financial institution the system is hierarchical, but globally no one is in charge and there is no central, executive authority. In this sense it is a ``free market.'' Some computer scientists are attempting to develop computer models of the world's first global computer to try and understand it better.
If we look back, we can see what events helped to bring about this global computer. A few decades ago the placement of the first satellites in orbit created a technological curiosity and a symbol of national accomplishment. Some people complained about the high cost of the satellites. However, satellites provided a highly reliable transcontinental and intercontinental communications link, and financial institutions quickly took advantage of them. Banks in London could release credit to banks in New York as the sun set in England and New Yorkers were still at work. Likewise, New York banks were able to communicate credits to the West Coast and thence to Asia. While people slept, their money worked. The satellite system enable a ``bulge'' of credit to rotate with the daylight zone around the planet. Some people estimate that satellites increased the world credit supply as much as 5 percent - hundreds of billions of dollars - much more than the entire cost of the satellite system.
When optic fibers are deployed across the Atlantic and Pacific by the end of this decade, many functions of the satellites will become obsolete. The increased bandwidth afforded by photonic systems will enable supercomputers on different continents to talk to each other. It is conceivable that European and Asian computers will be buying and selling on the US markets (and vice versa). Already, effective international telecommunications and computations have destroyed the arbitrage market that makes money on small differences in currency exchange rates. The only advantage of having a local computer near the market is the one hundred milliseconds or so it takes light to travel between continents. But that is a significant advantage if one has a fast algorithm. I recently spoke to a mathematician newly employed at a New York investment house who was developing sophisticated algorithms to determine buy-sell options. Why? So that his institution could get their orders in a millisecond ahead of their competitors.
It is well known that one of the most rapid forms of communications is a good joke. Businesspeople routinely leave their office in London with a fresh joke, fly to New York on the Concorde, only to find to find out at an evening cocktail party that everyone had already heard it. How is this possible? Banks and investment houses maintain open phone lines around the world in case there is a news break. The operators who maintain the lines often have no business information to transmit, so they trade new jokes. That's how jokes circle the globe so quickly. They are still one of the fastest forms of human communication.
The introduction of high-speed computing, data processing, and innovative software has transformed the financial services industry. Leaders in the financial services industry, while keenly aware that technology makes a difference today, were not always so aware. A decade ago the investment industry was hit by a technological revolution in the form of new electronic trading systems for stocks. Though suffering from an avalanche of paperwork (some called it the ``paperwork holocaust''), the New York Stock Exchange delayed the installation of this innovative technology. They were too busy making money and thought they would lose orders during the change-over period. The exchanges in Tokyo and London, which were not so concerned about short-term profits, became electronic markets. By their understanding of where the industry was headed, they got a bigger piece of the action. Even today a major problem is that as technology advances, systems quickly become obsolete and noncompetitive. How does one change over an entire network without bringing it down?
New skills are needed in order to manage the modern financial services industries - not just computer programmers, but high-level mathematicians who know how to design fast algorithms. In September of 1986 there was a ``computer assisted'' slide of the market. One of the reasons for this slide was that the buy-sell programs for many investment and brokerage houses differed in such a way as to produce an instability in the system. While the first step in most houses' programs for buying and selling was the same, the second, third, and fourth steps differed. This can produce a positive feedback loop; when the market becomes unstable and before human beings can intervene, the market can drop precipitously, costing many people a lot of money. When I asked many stock analysts about ``instabilities'' or ``singularities'' in market behavior, they never heard of them. Most are not trained in even rudimentary modern mathematics.
What are the chances that we will ever understand economic systems? They are clearly examples of extremely complex systems, but there is lots of quantitative data to check one's ideas on. Professional economists who bother to concern themselves with practical matters don't have an especially good batting average when it comes to predicting the future of the economy. They are smart, but they just don't have the right intellectual tools in their hands.
When I was in school learning about supply-and-demand curves, I asked my professor, ``Where did those curves come from? Were they made up, based on data, or did they represent a theory?'' The best answer I got, at least the one I remember, was that they represented the theory of economic equilibrium. The market, it was asserted, establishes an equilibrium, and the point at which the supply-and-demand waves intersect determines the price. This, reasonable as it seems, is of course nonsense.
The economic system, if it is anything, is a system far from equilibrium like the evolutionary system or the immune response. It is continually making adjustments to keep itself far from equilibrium (although there may be local equilibria). Next to nothing is understood about dynamical systems far from equilibrium. Probably the various kinds of attractors - fixed points, limit cycles, and strange attractors - play a role in coming to grips with how a complex system like the economy functions. Some mathematical economists such as Stanford's Kenneth Arrow have expressed cautious excitement about the application of the new ideas about chaos to economic dynamics. Mathematicians and others are endeavoring to apply insights gleaned from the sciences of complexity to the seemingly intractable problem of understanding the world economy. I have a guess, however, that if this problem can be solved (and that is unlikely in the near future), then it will not be possible to use this knowledge to make money on financial markets. One can make money only if there is real risk based on actual uncertainty, and without uncertainty there is no risk.
As emphasized by the investment manager George Soros in his book The Alchemy of Finance (1987), human biases profoundly influence markets in a reflexive fashion. Because such biases are influenced by political developments and cultural factors it is probably impossible to make a reliable model of the international economy. Like the weather, the international economy is an unsimulatable system. Yet short-term prediction and seeing long-range global trends may be possible using mathematical models.
I remember that back in the 1960s popular intellectuals spoke about ``the information age'' and ``the global village.'' Well, it has arrived, but not exactly in the form that these intellectuals anticipated. Felix Rohatyn, a New York investment banker and public-spirited citizen, recently remarked that we are now living in the ``money culture,'' a development brought about by the new data processing technologies. By this he meant that the dominant form of commercial exchange between people is not goods and services, but money. Money is, of course, a form of information, and it can move at the speed of light. People can easily invest it, transfer it, and lend it. And lots of people are doing this, some accumulating great wealth.
Only a few decades ago, if one picked up American business magazines, the articles were about new products, industries that produced goods and services, and the people who made that happen. Today the big news stories are about deals, financing transactions, buying, selling, conglomerating, integrating, divesting, and destroying companies. Smart young people who want to enrich themselves are attracted by all these deals and want a piece of the action. Nothing is being produced, but wealth is seemingly created. This bubble burst with the collapse of the market on October 19, 1987.
One could even imagine a satire on the theme of the ``money culture.'' People invest in the financial services industry, which, in turn, services their investment. Nothing but information is ever exchanged; no one produces anything; money, however, is always changing hands. The whole system bootstraps itself into existence - just money being exchanged and making more money based on the human confidence that it will continue to be exchanged. The image on gets is of an immense chain letter with promises of payment to all at a cost to none. Of course, it cannot work forever. At some point human confidence gets shaken, and a lot of people are hurt.
The real money culture, of course, invests in products and services. What has changed is the speed with which this is done. Speed, while a quantitative parameter, can, if dramatically increased, lead to qualitative changes - the changes we see in the global economy and, in particular, the large multinational corporations that play such an important role in maintaining it. In several such large corporations there has been a shift in both leadership and emphasis. The companies used to be run by traditional executives who understood the product and how it was produced and sold, whether it was automobiles or oil. But with the rise of the money culture many corporations, especially the oil companies, discovered that they could make more money investing and trading their surplus capital than doing what the company traditionally did - look for oil. Engineers and salesmen were replaced by international money market analysts and accountants. These new people began to run the companies. Which, of course, causes one to wonder who's minding the shop.
In 1986 I met with a group of bankers and businessmen. I told them that I knew of a ``computer nest'' operating in Luxembourg or Switzerland that was using a new ``massively parallel computer'' built by hackers in collaboration with a group of bright young traders for the express purpose of recognizing patterns in the commodities market. It had a learning capability similar to the Boltzmann machine. The traders were pulling in between two and three million dollars a day and wreaking havoc on the European commodities market.
My audience was stunned. ``Who are they? What are they doing?'' they asked, now on the edge of their seats. I told them that the story wasn't true, but could easily become true in the near future. This kind of ``technical breakout'' by an opponent, which is so often feared by military strategists, could also happen in the financial world.
Not only will advances in pattern-recognition systems influence financial decision making; so will the advent of detailed models of the global economy. There is an enormous amount of data generated by the world economy, so much that one human being or even a team cannot digest it. But computers can use that data in detailed models of various national and international economies and analyze it. Far-ranging supercomputer models will become a powerful asset in the hands of their creators - crystal balls that may make economic forecasting more realistic. One can foresee the characterization of economic systems in terms of different limit cycles and strange attractors. The international economy is a nonlinear system and can be understood as such.
There are dangers in the operation of the global computer system. A major instability could result in an international economic crash far worse than that in 1987. Many people predict that this can happen - that the markets will not stabilize after the October 1987 crash. Since no one person or group understands what is going on in the world economy and there is no central executive control, the entire system could end up in the basin of attraction of a fixed point representing very low economic activity. National governments would have to intervene to get the system started again, and new international institutions would have to be established, at a sacrifice of some national sovereignty, in order to prevent the recurrence of a crash.
In spite of all the advantages in computer technology it is not possible to abrogate human judgement in decision making. Much of this implementation of the new computer equipment, however, is designed to do just that. I find that distressing. Elementary decisions, lots of them, can and are made by computers/ Perhaps in the future more complex economic decisions will be made by computers as well. But people, with their innate desire to control their destinies, would be foolish to abrogate such high-level judgements to computers.
The diffusion of responsibility incurred by computers is a major danger, too. Once, waiting for breakfast to be served at a fancy new hotel, and after a long delay, I asked the wait what was wrong - where was my breakfast? ``The computer is down, sir,'' came the reply. I commented to my colleagues at the table that one will be hearing that excuse far more often in the future. My delayed breakfast was not the fault of the waiter or the cook. Not even the manager could be blamed. Only the computer manufacturer, programmer, or installer, all long since gone, could be responsible for the fact that my breakfast was delayed. The diffusion of responsibility serves certain interests, and it is important in each instance to identify them carefully. We are in deep trouble if we can't identify a human agent for these kinds of problems and hold them immediately responsible. But there are still other dangers.
Some intellectual prophets have declared the end of the age of knowledge and the beginning of the age of information. Information tends to drive out knowledge. Information is just signs and numbers, while knowledge has semantic value. What we want is knowledge, but what we often get is information. It is a sign of the times that many people cannot tell the difference between information and knowledge, not to mention wisdom, which even knowledge tends sometimes to drive out.
I've examined just one of the many impacts that the new sciences of complexity will have on the world - that in the financial services industry. There are other impacts - on education, medicine, and the legal profession. The computer, a new mode of production, has come into existence and created new classes of people, new jobs, and new forms of wealth. What I find especially interesting about this development is that abstract mathematics, sophisticated algorithms, and vanguard technology are going to determine the future of industries and professions long immune to such changes.
Someday, sooner than many people think, the sciences of complexity will impact on the legal system, not just in data processing but in actual decision making. Could an expert system replace an attorney or at least assist one? Probably a lot of mundane legal work can be done by computers, and lawyers will discover that they can serve their clients better by using computers. The use of content addressable memories, for example, would be a great aid in case work. Right now the impact of the new sciences of complexity on the legal profession is still minimal; but this will soon change.
A new salient of knowledge is being created, and our generation is privileged to see it unfold. Like all great changes throughout the course of human history, it provides challenge, opportunity, hope, and danger. We stand on the threshold of the human mastery of complexity - an agenda for science that may show us, for the first time, who and what we truly are.
Information, be it embodied in organisms, the mind, or the culture, is part of a larger selective system that determines through successful competition or cooperation what information survives. Information can be encoded in genes, nerve nets, or institutions, but the selective system that promotes survival remains similar. This insight is hardly original. Yet it remains a mystery to me why philosophers, psychologists, and social and cultural scientists have rarely grasped the import of the Darwin-Wallace notion of selection for their own work (this has recently been changing). A selective system is a pattern producing and recognizing system, be it the pattern of life on earth, the symbolic order of the mind, or the pattern of culture. A selective system manages complexity. ...