Heuristic Principles

Some Eastern philosophies have a place for heuristics, some versions of Buddhism even emphasize them. In the West, we are forced to classify heuristics as anecdotal, social myths. Can we make a greater place in our thinking for heuristics, without subscribing to some form of Eastern mysticism?

I think so. Heuristics are practical rules used now and then, when they fit the situation. They might be "proto-principles," since, properly refined and qualified, they might be elevated to scientific principles. For example, "Murphy's Law" has something to do with probability and risk. "Gresham's Law" has something to do with the Tragedy of Commons, and a lot of economic, social and political behavior. The "Peter Principle" and "Parkinson's Law" encapsulate a lot of observations about behavior in hierarchical societies, human and otherwise. All of these folk-wisdom principles and laws are handy coveralls for a fairly wide range of behaviors and events. But, since they are not exact in nature, they are treated as mere gossip.

I want to elevate heuristic principles somewhat, but not too much. I don't want to make Murphy's Law into a founding idea in physics, but it is more useful than a joke. Similarly, Gresham's Law actually describes real motives in financial markets and business dealings, even if it is not quantified and allowing that "bad money" is probably a subjective value. The utility of heuristic principles is the starting point they provide in dealing with many situations that cannot be formally, exactly and immediately analyzed. They are also warnings against overly calculated expectations; most heuristics suggest caution.

Murphy's Law tells us not only 'what can go wrong, will,' but beware unexpected consequences. Murphy's Law isn't just about statistical odds; it is about what a prudent person should consider before acting. This difference becomes clear in cases like building a nuclear power plant. Nuclear power plants are very safe places. The odds of anything untoward happening are very, very low, at least as determined by traditional methods. But, as we discovered with Chernobyl, things can go wrong for unforeseen reasons and, when they do, the results are catastrophic. The very low odds of a catastrophe are multiplied by its extreme effects. The net insurable risk may still be very low over a very long run, but that is little comfort to the short-run casualties in an accident. Murphy's Law does not function as a method of assessing the extent of risk ("insurability") in such cases, but it does warn us to look at marginal cases more carefully than we would based on the technical calculation alone. It encourages us to "think outside the box," looking for new and different ways things might happen. After all, the technically correct calculation of risk only evaluates factors known to exist according to standard physical concepts and engineering methods. Thus, the Curies were unaware of the poisonous effects of radium and other radioactive elements, because radioactivity wasn't even a concept when they started their work. Similarly, the hazards of electricity or microwaves could not be evaluated until those fields were sufficiently developed.

Murphy's Law suggest we need to add a variable - judgement - to risk calculations. Judgement is a fudge factor, but it is very important in matters affecting human beings. Judgement should be based on the experience of knowledgeable practitioners. I think we need not consider "blue skies" estimates, or other proposals of risk based on fear, ignorance or other idiosyncratic factors. Judgement should introduce into risk calculations possible scenarios that may need to be hedged.

Parkinson's Law, introduced humorously in the 1950s, actually operates in most large organizations. The central notion is that people advance to responsible positions at least one step higher than their competence. This occurs in organizations that promote according to merit (however determined), but have no simple way to determine the "depth" of a person's abilities. Thus, a person's arrival in a demanding position is the only valid test of competence (sink or swim). Parkinson's Law only applies when organizations cannot or do not remove incompetent people; i.e., those who cannot swim are left to drown. In fact, this last is a widespread practice for several reasons. Promotion and demotion are highly emotional events for most people, and have ripple effects throughout the organization. It can be so disruptive to cull the ranks that it is better to leave somewhat incompetent members alone. Nonetheless, in fact, blatant incompetence is usually punished and excised, because of its cost to the organization. Thus, Parkinson's Law implies there is an "allowed" range of incompetence: "muddling through" is acceptable to most organizations. Parkinson's Law, taken together with the idea that intelligence is measured by the speed of learning, further implies that most organizations change slowly because most of its members are only marginally competent in their assigned functions. So, it is very rare to find an organization that has superior leadership, and truly capable members.

Are there other ways to explain the phenomena Parkinson exposed? It could be, for example, that large organizations are simply subject to laws of large numbers. Because large organizations have so many members, membership characteristics will approach the distribution of the same qualities in the underlying population. If you only need a few marbles, you can select all the red, white and blue ones you need from a giant marble pot. However, red, white and blue marbles may be quickly exhausted, so large draws from the marble pool may necessitate accepting "compromises." That forces the chooser to draw a population that looks more like the average in the pool. So, the need for specialists will be compromised, and everyone will only be partially fit for their assignment. This lack of complete fitness is inversely described as unfitness in Parkinson's Law.

While we might dispense with Parkinson's Law, if we could find more substantial, scientific reasons, for its observations, Parkinson still gives us a succinct insight into what is going on. Or, Parkinson's Law may encapsulate several "more rational" explanations that explain similar situations, so is an umbrella for a certain type of observation or explanation. I don't think there is sufficient reason to drop Parkinson's Law, as the more mechanical explanations do not bring into the mix the "human factor." Parkinson demonstrated a knowledge of human social behavior and psychology as it occurs in organizations, especially bureaucracies, which is usually ignored in more "scientific," Taylorist studies.

Gresham's Law - bad money drives out good - amounts to the Tragedy of the Commons. Enough of those who know how to abuse the system will do so, thus destroying the system. The availability of a Commons entices certain users (pirates) to take more than their fair share, and to avoid investing anything in it. The simple consequence is that all those who "play by the rules" (co-operators) are taxed by those who do not. Further, the pirates will be well rewarded for their efforts, if there are not too many of them. This last fact, taken together with the inevitable spread of piracy, reduces the non-pirates to poverty or even insufficiency and death. Piracy spreads whenever there is a "learning curve;" i.e., those who were not pirates are encouraged to change their status by the example of the rewarded pirates.

Gresham's Law is a specific instance of the Tragedy of the Commons, having to do with currency. When a small amount of counterfeit - bad - money is introduced into the system, not much will change until the true value is discovered. However, when an accounting is made, someone is going to come up short, because the counterfeiter will be long. Counterfeiting is a form of robbery. If there are enough robbers, non-robbers will be encouraged to rob as well. Why pay in solid gold, when plated copper will do? Thus, the public currency will be debased, while the players hoard what is perceived to have value. This scenario plays itself out everyday under various masks in financial markets. It is what happens during Depressions, and is involved in market booms and crashes.

Gresham's Law and the Tragedy of the Commons are themselves explainable by the Prisoner's Dilemma. In its various forms, the Prisoner's Dilemma is whether to co-operate or be selfish. In Game Theory, the prisoner is confronted with a reward for ratting on the other guy, which also punishes the other guy for keeping faith. On the other hand, mutual co-operation allows the players to achieve a much smaller reward, which is barely more than what is required to survive. Clearly, grabbing the reward and running works for whoever grabs first. A problem arises when the prisoner is required to play the game again and again: the other players won't trust grabbers, and will go for it themselves. Thus, those grabbers forced to endure many sessions of play may lose everything.

I understand the following have been shown (mathematically proven) to be true:

1. The best long-run strategy in the Prisoner's Dilemma is co-operate;
2. The best short-term strategy is Tat-for-Tat.