SUCCESS AND RATIONALITY

The problem of induction, many say, is, can we project past experiences into the future? Will the future be like the past? Alas, no; I am unable to expect rejuvenation. That is a witticism.

The problem of induction is an expression of the wish to be reassured not by a condescending parent or priest, not by blind faith, but by rational means. Can we justly predict and explain the success of science? This is the problem of induction. Popper says it is insoluble in a positive manner. All he can offer by way of a solution is his notorious modus tollens, or, to be precise, the rule of retransmissions of falsity from the conclusion to the premises. In the explanatory model we assume a hypothesis and initial conditions from which a prediction follows. If and when the initial condi­tions are confirmed by experience but the prediction is refuted by ex­perience then, we may say, experience refutes the hypothesis in question. Assume that all that experience can offer is such refutation of scientific hypotheses, and, in particular, no confirmation and no guarantee. If that is all to it, then all scientists and technologists are leading precarious lives with no insurance for the future. As a hard datum, this is not so. Banks are ready to offer long term credit to scientists at least as easily as to businessmen.

But all this is a mere dramatization of a very well known prosaic point. Given the principle of induction we can explain the success of science; and we can make the following conditional prediction: those who will employ the principle in their research will be - certainly or in most like­lihood - successful. This is the famous infinite regress, step number one.

But if science has been successful due to the employment of this prin­ciple and if it will be successful as long as scientists continue to employ it (and banks do give them credit on the assumption that they will), why then can we not ask scientists what principle they employ? Why can we not become scientists or employ scientists to find out this principle?

This, indeed, is so puzzling one might wonder if it is not a scandal. Hume’s subtitle to his Treatise may read, an attempt to make induction a part of empirical psychology. 200 years passed. I think that Popper’s new philosophy of science, that the studies in the theory of rationality, of Popper, of Bartley, and others, have given us something that we may in­deed call a method which is not an algorithm or sausage-machine and which those more sensitive to success than to the history of ideas and to our heritage in the philosophy of science may indeed call induction. It is a partial method with the partial guarantees - not the absolute one - that most civilized citizens of civilized communities have come to expect. Painters, for example.

A painter produces regularly and his art should not be produced by a sausage-machine. Here is the crux of our situation, in art just as much as in science: the gulf between them is part Baconian - science is a sausage machine - part romantic - art is not produced regularly. Both science and art have their routine side and their inventive side. I think that even the dichotomy between the routine and the inventive is here overworked; to do an utterly routine job we have to mechanize it so much with conveyor­belts and what-have-you, that you may just as well dispense with the worker altogether and fully automate production.

But the robot we employ and the bread we eat may betray us; or they may be useless in a total catastrophe. A total catastrophe of one kind and of another is declared possible by science, and so in the long run some total catastrophe is, we are told, more or less inevitable. When insuring ourselves against future catastrophes we ignore such eventualities. The grand axiom of insurance theory is, there is no insurance against epidem­ics; if one in a catastrophe is compensated, this is because others like him prosper and pay him through the insurance-system, be this an insurance company proper, a federal agency, or the village system of mothers baby­sitting for a sick mother. And here, we can see, the artist and the scientist integrate well in the system by various means: he teaches, he administers teaching and production - he produces almost routine works, he is one of one hundred who idle in the hope of producing a masterpiece and they all feed on the one masterpiece produced among them (for example, we all sell logic texts to the publisher, and he makes his profit from very few of them).

The artist and scientist, then, have bread-and-butter operations which signify within the system, and wind-falls to boot; even regularly though not algorithmically. An algorithm or a quasi-algorithm is possible within a system: given our present system, the number of chemical combinations in organic chemistry and molecular biology is finite and so, even in accord with the strictest Popperian canons, induction (probability or eliminative induction) is demonstrably possible.

Can we say the same about the framework itself? No. It cannot be justified, especially since we tend to improve it, i.e. admit it to be faulty. But, like democracy, its the best we have. Like democracy, also, we do not give it up when we find its defects until we can replace it.

Jeffreys and many in his wake have said, we do not give up our belief in a refuted hypothesis until it is replaced by a better and more viable one. This shocked and puzzled my younger self of a two decades ago or so. Does Jeffreys mean to advocate the belief in a refuted hypothesis? This is absurd! But I do think he is right on technology; we do not believe in the framework we use, but we go on using it until it be replaced. And when we try a replacement we try pilot plants. A pilot plant is designed for the achievement of positive evidence, though its test is performed as attempt­ed refutation: we want to be right at least to the extent of having a better system to implement. But if we are not, we prefer to learn about it in the pilot plant not in the main plant. Hence pilot plants want corroborations in Popper’s sense of the word. But the degree of corroboration required varies from country to country, from time to time, it may be improved by trial and error too.

Socially speaking, or economically, we allocate some of our intellectual resources to pure research which is problem oriented, critical, and seeks refutation; we allocate some of our intellectual resources for technology which is task oriented, critical, and seeks corroborations. We support the two groups from the common pool; we do not like our inventors, scientists, and artists, to depend on their really inventive imagination because then they starve in attics before they can perhaps burst into fame and riches.

This solves the problems of success and of induction, in an empirical fashion and sociologically. Psychologistic induction is now replaced with sociologistic one.

V.