Heuristic Strategies

Strategies which provide a definite and certain way to reach a goal are called algorithms. They consist of exact, step by step, instructions and are familiar to computer programmers because this is how most structured programs are designed.

Algorithmic strategies are used extensively throughout life and form the basis of most training and educational programs where learning consists of being taught a programmed set of fixed methods or responses.

Algorithms work best in situations where they are employed by tacticians operating in a predictable and limited environment and where there is full knowledge of all contingencies.

Where algorithmic strategies break down is in unpredictable environments which contain uncertainty, change or competition. Algorithms can deal with a certain amount of variability but their design quickly escalates to unacceptable levels of complexity if there is too much variability and usually they will fail completely when presented with unknowns or novel situations.

When people talk about strategies, they are usually thinking of situations where algorithms prove inadequate, that is in environments where there are many unknowns and where there is much change and competition. Such strategies have been a constant preoccupation for the human mind since time immemorial.

War concentrates the mind wonderfully, and the war strategists of the 1930's and 1940's brought powerful minds to bear upon this problem of dealing with competition and uncertainty. The most influential thinker of that period was Professor John von Neuman, one of the greatest mathematicians of all time. He had brought out papers in 1928 and 1937 which laid the foundation of a new way to make decisions in conditions of competition and uncertainty which became known as game theory.

Game theory offered a statistical approach to making decisions and was successfully used in world war 2 to deal with such topics as logisitics, submarine search and air defense.

Von Neuman's book. "Theory of Games and Economic Behavior" which he wrote with Oskar Morgenstern and published in 1944 (improved in the 1947 edition), brought these mathematical concepts of strategy to the attention of the rest of the world.

Game theory can be thought of as a conceptual framework in which to consider problems of decision making in conditions of uncertainty and competition. The main conceptual mechanism of game theory is the heuristic strategy which is a set of rules developed to specify the most likely way to win a game or succeed in a goal.

Unlike the clear steps of an algorithm, which have predictable consequences, the rules of an heuristic strategy are more like 'rule of thumb' guide lines for achieving a stated goal. They are usually based upon empirical results: experience or successful application in similar circumstances.

Following the guide lines, or rules, of such a strategy will not produce predictable outcomes or guarantee certain success; instead, the users of the strategy are only assured that they are going to have the best chance of success in a statistical sense. i.e., the people who abide by the rules are more likely to succeed than those who don't.

These kind of "woolly" instruction sets are completely foreign to computer programmers and it often takes a while for them to wrap their minds around the concept. The essence is one of trading off absolute control and predictability against reduced complexity of the programming

For example, an heuristic strategy designed to help you succeed in creating wealth in a non zero sum game environment, would not consist of a set of clear instructions which you have to follow in order to be sure of becoming rich. Clearly, an algorithmic strategy is not possible in competitive economic environments. Instead, the rules would consist of general rules of behavior which would be likely to give you a competitive edge against others in the competition to become rich over a period of time.

The kind of rules you would find in such a heuristic strategy to help you succeed to in business would be rules that ensured you were honest, reliable and truthful in any dealings. Behavior which involved cheating, stealing, deceitfulness or coercion would be declared illegal.

In the long term, such a way of conducting your business dealings would eventually pay off as you built up credibility and reputation, although it would be quite possible for somebody to make short term gains by ignoring such rules.

Heuristic strategies need not be restricted to benefit a single individual, they can also be designed for the benefit of a group or a society. If the goal is creating common wealth, then a heuristic strategy is likely to include rules which would enhance opportunities for cooperation in order to increase the general efficiency (i.e., rules of law and ethical codes of individual conduct would be conducive to forming a successful society). The most effective and well known heuristic strategy used in the western world is the Ten Commandments of the Christian religion.

For the design of intelligent agents in object oriented environments it is essential to understand the nature of heuristic strategies because they have to be built into objects in order for the objects to gain "experience". Such techniques are far more efficient than algorithmic methods as they can allow objects to learn and adapt without having to use complex programs or large data bases.

[Index]
[Next - Lists and objects]
[Back - Wealth creation]