You drive on a highway, and the traffic is dense, nearly a traffic jam. You notice that the cars on the next lane are overtaking you while your own lane is slow. You see a gap, take the opportunity and change lane. But a few moments later, the cars on the lane you are on now are slowing down and the lane you have just left becomes faster.
Obviously, you were not the only one changing lane. Several people did so and as a result, traffic on that lane became denser and was slowed down. Many people made the same observation: the other lane is faster. The problem here is that this observation is part of the system. An external observer standing on a bridge above the highway can make the same observation but it has no effect. But if the observers are inside the system, driving the cars, and if their actions are influencing what is happening in that traffic-jam-system, the observation becomes an active part of the system. Acting upon it changes the rules of the system and as a result, the observation becomes invalid. As a result, there is probably no such thing as an optimal strategy for this kind of situation.
Another example of such a system where observations destroy their own validity if they become part of the system is the stock market. If you are a trader at such a market and you have some information that gives you an edge, e.g. you know something about a certain stock or bond that others don’t know yet, you can act based on that information and buy or sell. Others might react to your buying or selling. E.g., if you buy a larger quantity of stocks and prices go up, others might follow, until the information you had is “priced in”. As a result, you lose your informational advantage. Whatever predictive information one of the participants has is lost this way by being included in the price.
Now consider you are a stock trader and you have access to some prediction software that others don’t have. The software contains some algorithm or theory to predict how the prices will develop. As soon as you start using this to inform your trade decisions, its predictions will be priced in. This means that the system will start behaving in ways that are not predicted by your software as soon as this software becomes part of the system and you base your trading decisions on it. The market, this means, will maximize its own unpredictability. Any model of how it works will become invalid the moment people start using it to base their trading decisions on it.
What holds for traffic jams and stock markets is probably true for other phenomena involving people as well. Cultural or social systems consist of people interacting with each other. These people are at the same time observers of what is going on in the system. But at observers, they are not just bystanders on a bridge observing what is going on. They use their observations to decide what they should do. The observations or theories, hypotheses, models, opinions based on these observations become active information inside the same societies, cultures or economies they were derived from by observation and analysis. They become, so to speak, part of the software that runs in the society. As a result, the societies change, often in ways making the theories we have formed about them invalid.
A system that can make observations about themselves and make these observations part of its active information is reflexive. Reflexivity makes such systems creative and unstable in the sense that they might change their behavior and rules at any time. so that there are no fixed laws describing what is going on. A complete theory of such a system is impossible because theories formed about what is going on become part of the system, part of its active information, its running software. As a result, the formation of a theory changes the rules and nudges the system out of the scope of validity of the theory. As a result, such systems don’t have fixed laws but they have a history.
Human beings are reflexive systems in this sense themselves. We can observe out own actions and thought processes and the observations we make can be turned into parts of the active information that comprises the “programs” of our thought processes. If we look at it in terms of software, we are able to reprogram ourselves. This means that we are creative.
(The picture is from http://commons.wikimedia.org/wiki/File:Traffic_Congestion_Brasilia_2.jpg)