Nonfact and Non-restriction

Chapter 1, Section 6: Nonfact and Non-restriction

I expressed again and again the importance that a theory must be refutable by facts. I also pointed out tautologies, theories with ambiguity, and theories with antinomy are impossible to be refuted. There are still two other types of theories that cannot be refuted and thus have no explanation power. One is those for which phenomena used for confirmation are nonfacts, and the other is those by which phenomena predicted to happen are non-restricted.

If I say "when it rains there must be clouds in the sky", then in this statement the rain and the clouds are facts and must be observable. But if the rain or the clouds are just castles in the air, or nonfacts, then the rain-cloud theory can never be confirmed. In this example there contains a principle of empirical sciences that's not shallow. For all predictions with explanation power, their confirmations must conform to the following implication: if A happens, then B happens, of which both must be observable facts. No matter how much the cost and time might actually be spent, A and B must be at least in-principle confirmable. Both A. Einstein's relativity theory and the genetic theory in heredities had untestable implications in their early times, but afterwards, all got confirmed.

The key is, as stated before, a fact cannot explain another. The happening of A cannot explain that of B. The regularity for A and B can only be used to confirm the implications of a theory. Even if facts are so plenty that we can pick up at will, and the regularities are very obvious, they still cannot explain each other. On the contrary, theories with explanation power usually originate from abstract thoughts; at first there are just some nonfactual hypothesis, but with logic reasonings applied, they can generate testable implications. This is exactly how the rain-cloud theory comes.

Such a job is never easy, though. A testable implication must be able to be refuted; facts don't explain themselves, while abstract theory itself is not testable. So it's all in the minute transitions from abstraction to validation that how the superior and the mediocrity are divided.

Let me have an example. In economics the well-known law of demand says: when the price of a good decreases, consumers' demand for it increases. Price and its changes are observable, but demand is not. Demand stands for consumers' desire or intended need, something abstract. So the law of demand itself cannot be testified by facts. However, this law is important and indispensable for economics. The mediocre usually take the trade volume as demand. This is calling a stag horse, of course mistaken. The correct treatment is totally different. We should say: if the law of demand is correct, then by logic reasoning, under certain observable circumstance the happening of A leads to the happening of B, while both A and B are observable facts (this is how the untestable law of demand derives a testable implication). If we see an occurrence of A without B's happening, then the law of demand is flawed, we need either add other circumstance clauses to it or deem it refuted. If nonoccurence of B implies nonoccurence of A, i.e. we cannot see A happen when there is no B, then the law is not refuted and we can regard the phenomenon about A and B as explained.

You are correct, such implications and their verifications can be made very wise and crafty, which shows the beauty of science. In this book I will demonstrate tirelessly the amazing explanation power of the law of demand. What needs to be stated right here, is that aforementioned additional conditions can vary greatly in both amount and content. In terms of scientific methodologies, the additional circumstances are called test conditions, or constraints in economics. Sometimes we say if A AND B happen, or A OR B happens, it leads to the happening of C. We can also say if A happens, it leads to the happening of B AND C, or the happenings of B OR C. The variables (A, B, C, etc.) can vary in amount; either they can appear in a single observation, or part of them, maybe one, two, or three, appear in different possible observations. All of them suits the theories that have explanation power. But no matter how many phenomena get involved in a validation (i.e. implication), there must be a restriction.

If we state, the happening of A leads to B's happening, or C's, or D's, or E's... and so on, which go endless, then this implication becomes impossible to be refuted. Rigorously speaking, this is the so-called disequilibrium situation in economic theories. Conversely, when an implication becomes confirmed and refutable because there is restriction on involved phenomena, it's called equilibrium.

Above definitions on equilibrium and disequilibrium are not the same as used by traditional economists. I think they are wrong from the basis. The traditional equilibrium concept in economics was borrowed from physics. In physics equilibrium means a pendulum stays in the middle when it stops moving, or an egg reaches a fixed point after rolling for a while, or a moving object enters an orbit and becomes predictable. These equilibriums are phenomena, or observable facts.

In economics equilibrium means totally different. For example, economists take the intersection of the demand curve and the supply curve as an equilibrium. But there is no such demand curve or supply curve in real world, they are just conceptual tools made up by economists. Without economists, these tools wouldn't ever exist. In like manner, equilibrium and disequilibrium in economics are just concepts as well and don't exist in real world. Being not phenomena or facts, they can't be seen.

In the spring of 1969, Coase and I drove from Vancouver to Seattle. During the two hour trip, Coase debated with me about the equilibrium concept in economics. He thought equilibrium and disequilibrium were just castles in the air, totally waste, and thus should be removed from economics. I agreed the castles-in-the-air opinion, but since the concepts were so popular, I could save them by revisions.

I proposed to Coase that disequilibrium can be interpreted as that a theory lacks refutable implications due to non-restriction on its predicted phenomena, while equilibrium is for the case that a theory has the restriction and thus is testable. This is the difference between the aforementioned "non-restricted" and "restricted". Coase agreed at the time such an interpretation can save the useless concepts of equilibrium and equilibrium in economics. A story more than forty years ago. Today, economists that understand and agree this notion is no greater than ten.

Abstract theories themselves cannot be validated. To explain phenomena, they must have one or more refutable implications. Such implications must have the possibility to be overthrown by facts; listed constraints and predicted phenomena can be paraphrased with the certain word AND or the uncertain word OR, but they shall not be unlimited. Of course, the certain AND is superior to the uncertain OR in explanation power, and the simpler the abstract reasoning and the testable implications of a theory are, the more powerful it is. It's why scientists who reduce complex phenomena to a great extent are called genius.