The Realness of Theories

Chapter 1, Section 7: The Realness of Theories

Since facts cannot explain facts, theories used to explain phenomena must be abstract to some extent. Abstract concepts are not facts. This made some people feel theories were no longer hinged to reality, they were just big talks, empty and useless. Realism became a big debate. Nowadays, the dispute has calmed down, but it still requires a clarification.

Realness has many meanings, so a debate would be endless if its indication is not first defined. Abstract concepts are of course not facts, so to say a theory unreal is definitely OK. But the final motivation of any theory is to derive fact-related validations or predictions, so in this sense we could say useful theories are indeed real. As for the theories that we cannot derive any testable implications (like many ones in development economics in the 50's and 60's), they are merely games and has nothing to do with the real world.

When a theory with explanation power is said unreal, there are at least four understandings, of which three are superficial. First, a theory must have abstraction. To say it's unreal is of course correct. But concluding accordingly it explains nothing is wrong. Since facts cannot explains facts, without abstraction, nothing can ever be explained. Second, all the descriptions on facts or observations are simplified, which makes facts unreal. This is just quibbling. Let's take apple as an example. If we really need detail what an apple is,  we can't succeed even with all paper of the world used up. To fully describe the color and shape of an apple alone, not to mention its taste or vitamin, would be unattainable. By the same standard, no description in this world ever made to a phenomenon or a fact is real. Criticizing science in such a quibbling way, though plenty of people do it, is therefore unscientific.

The third understanding is about simplification as well. The world is complex; simplifying assumptions (different from those in abstraction) are necessary. Such simplifications are just for the ease of handling; omitting them doesn't change the main intention. For instance, we can ask questions like if there are two countries in this world (there are more, which makes current assumption unreal) how would it be when they trade with each other, and so on. Changing the number to three or four doesn't make any difference at all. Of course, there are occasions when changing from two to three makes great difference. When dealing with those, we can't ignore the distinction; though, other simplifications other than this one are still necessary.

The last type of unreality is not as shallow. Many people take the aforementioned test conditions as just assumptions. Of course these assumptions carries the unreality of simplifications, but not like the ones that are detached from reality in thought abstraction, they shall not be considered as castles in the air. No matter how simplified, test conditions must be traceable and not differ too much from reality. For example, when performing a chemistry experiment, we must use a clean tube (cleanness here is a test condition) instead of a soiled one, the latter of which can never be assumed clean.

In economics, test conditions are usually called constraints. No theory without a constraint exists in economics, as is the case in other fields, there must always be constraints, or else explanation power vanishes. We can assert, when there is no transaction cost (a constraint and barely an assumption), the happening of A leads to the happening of B. To test this implication, we must start when the transaction cost is really negligible. In other words, assumptions in constraints shall never be detached from the real world. That's to say, except for necessary simplification, test conditions must be real.

In conclusion, for scientific theories that originate from abstract thoughts or concepts, certain unreality is necessary, as facts cannot explain facts. "Not too detailed" and "simplification" are allowed types of unreality. But having constraints detached from the real world is a very big mistake. In economics, the research and simplification of constraints (test conditions) is the most difficult job along the economic explanation path. Worldly affairs are like chess games, every one is new, so it may take decades to get just a glimpse on certain constraints, which is normal. Time only advances one's age, so an economist of empirical researches never devotes all his heart to a problem only until the importance of it is assured.

Regarding the reality problem of theory, during the great debate of methodology in economics happened last 50's and 60's, there once was an embarrassing fallacy. That is, if we assert "A's happening leads to B's", then we have "if no B happens no A happens", and cannot have "if no A happens no B happens". The latter has been talked over previously. In that debate, many economists neglected this first lessen of logic and forgot we have no idea about B when A doesn't happen. The colleague that worked on Boston transportation companies thought the assumption A was not right and then made quite a fuss about what would B be going on. Such silly analyses never deserve a response, but sometimes the advance of science can be this baffling, that the responses of other scholars ignited that beneficial debate.