FIRST introduction: reliability is not a matter FOR PURE SCIENCE
The reliability of scientific theories has been taken by most philosophers to be a major contribution of science to humanity. It seems advisable to consider reliability or whatever should replace it, as we should consider all extra intellectual benefits of science, within the realm of scientific technology, and restrict the domain of science as much as possible to enlightenment only.
One may object to the view that reliability belongs to technology by saying that there are two senses of reliability, technological reliability of a theory in its practical application and intellectual reliability which constitutes the observation of the intellectual merit of a theory. I agree that observing the intellectual merit of a theory, as well as its weaknesses, is part and parcel of pure science. If this is what one means by the intellectual kind of reliability, then I accept the objection after observing that the two types of reliability have little in common. Those who wish to link closely the two types of reliability will have to say a little more about the nature of the second kind of reliability. This was viewed as the same as credibility, which is different from recognizing merit, as we can easily compare the merit of two views which we reject: the two views have equal credibility, namely zero credibility, but perhaps unequal merit.
Due to the religious ancestry of modern science, it was supposed that when the Copernicans expelled Aristotelianism from the body of established belief, some other doctrine should have replaced it; it was even alleged by some that the only way to replace Aristotelianism as an established belief is to replace it by the establishment of another belief; and further, that either the alternative belief is scientifically established, or else we merely replace one dogma by another. There is much sense in this view: such things as established doctrines do exist, now as then; and establishing a doctrine is part and parcel of instituting a number of social practices.
It turns out, however, that there is no need for anybody to believe in an established doctrine; nobody asks how many Englishmen believe in the Church of England, for example, before being ready to admit it as a state religion. The field of established belief is a matter of social technology; and so it is at worst a matter of tradition, or caprice, at best of rational or scientific technology; but not of pure science.One may still press the objection to my view that reliability is a technological matter. Even if scientists do not endorse the established scientific doctrine, they do have their own private beliefs, and these can be enlightened or not. Surely, to that extent the reliability of a scientific doctrine in the eyes of a scientist is a matter of his own enlightenment! Again, I will gladly endorse this view if someone offered a reasonable theory concerning enlightenment and belief. Historically, the philosophers Sir Francis Bacon and Rene Descartes linked the two most strongly. The link became a part of the scientific establishment with the foundation of the Royal Society in 1660-3, and the recognition of Sir Francis Bacon as its ideologue. Bacon assumed - we know now that he was in error - that a scientist acts on his beliefs at least to the extent that they direct his researchers and teachings. There is really little need to indicate that all respectable schools teach doctrines not endorsed by their members. Only doctrinaire Freudians for example ignore all non-Freudians and refrain from teaching extensively all non-Freudian psychology. In physics it is simply unthinkable not to teach the ideas of Newton, Hamilton, Maxwell, and other quite outdated masters. A philosopher not familiar with Greek thought is incompetent even though no one today endorses any Greek ideas as they stand. Nor do people conduct research entirely along their own beliefs; much is directed to examine other people’s beliefs. History easily illustrates this. Euler was a Cartesian who contributed to Newtonian mechanics in one way or another.
Boscovich was an Aristotelian who saught to compromise Newton and Leibniz, Hamilton was not a Newtonian but a Boscovichian. Maxwell did his main work while trying to reconcile Faraday with Descartes and both with Newton. It is hard to say what each of these people exactly believed, what he believed he was doing, and what he believed was the relation between what he believed in and what he did. Nor are historians of science agreed, either about the facts of the matter, or about how the facts appeared to the researchers involved.To conclude the criticisms, it is not denied here that reliability matters for pure science. It is only claimed that the two chief roles traditionally ascribed to reliability, are matters for scientific technology. In particular, the reliability of a theory in practical application is clearly not in itself of concern for pure science. Further, the idea that erroneous belief vitiates research is at least doubtful: we do not know what is the right belief, and how erroneous beliefs influence research.
It seems as if the point just made is in conflict with the conclusion of the last paragraph. This is not so. Any adequate theory of belief should take cognisance of the fact that thought influences action and thus research activities. Yet, no doubt, the views here discarded fail to do it. Bacon wanted scientists not to be led astray by false conjecture and so forbade all conjecture until enough facts were collected. Hence Bacon was driven to the view that facts have to be discovered by accident unaided by conjecture. He was in error; any discovery aided by thought shows this quite clearly.
There are three possible relations between a theory and a description of a general fact, deducibility, contradiction, or independence. To insure that true descriptions of facts to be deducible from a theory correctly, the theory should be true; and conjecture has little chance to be true. If a statement of fact contradicts a theory, said Bacon, then accepted theory blinds us to that fact.
Hence, false theories are evil and true theories have to be based on facts or come after the facts. The remaining case, then, is of independence: a new fact is independent of any existing theory though it follows from some future theory which will be based on it. The history of science forces us to reject Bacon’s views: in historical fact new facts constantly appear contrary to old established beliefs. Let us, then, examine Popper’s theory. The body of science was viewed traditionally as the body of established theories. According to Popper the body of science constitutes refutable conjectures. The progress of science Bacon and his followers view as that of the increase of body of established belief, whereas Popper views scientific progress as that of the making of refutable conjectures, and the occasional attempts at refutations of some of them, successful or unsuccessful as the case may be.At once we can see that thought and action may relate in a manner most obvious yet ingored by those who wish science to provide us with reliability. Namely, science can eliminate unreliability. And, in particular, when the action in question is scientific research, the thought may be a suspicious thought, an ingenious attempt to overthrow current theories, and discovery may be the successful refutation, the finding of the refuting facts.
This view of Popper encounters two difficulties, one classical, often attributed to David Hume, one modern, attributed to Duhem and Quine. Hume’s difficulty is, how is a theory ever established, or can a theory ever be established? Duhem’s difficulty is, how is a specific theory refuted, or can any specific theory ever be refuted? Let us take Duhem’s difficulty first.