After the publication of The Open Society in 1945 my wife pointed out to me that this book did not represent my central philosophical interests, for I was not primarily a political philosopher. I had in fact said so in the Introduction; but she was satisfied neither by this disclaimer, nor by my subsequent return to my old interests, to the theory of scientific knowledge. She pointed out to me that my Logik der Forschung had long been unobtainable and by then was very nearly forgotten; and that, since I was assuming its results in my new writings, it had become urgent that it should be translated into English. I quite agreed with her, but without her insistent reminders, through many years, I should have let it rest; even so it took another fourteen years for The Logic of Scientific Discovery to be published (in 1959) and another seven years for the second German edition of Logik der Forschung.
During these years I did more and more work which I intended to use in a companion volume to The Logic of Scientific Discovery; and in approximately 1952 I decided to call this volume Postscript: After Twenty Years, hoping that it would come out in 1954.
It was sent to the printers in 1956, together with the (English) manuscript of The Logic of Scientific Discovery, and I received the proofs of both volumes early in 1957. Proofreading turned into a nightmare. I could complete only the first volume, which was published in 1959, and I then had to have operations on both eyes. After this I could not start proofreading again for some time, and as a result the Postscript (edited by Professor W. W. Bartley) was not published until 1982–3, with the exception of one or two extracts.236 It was of course read by students since 1957.
In this Postscript I reviewed and developed the main problems and solutions discussed in Logik der Forschung. For example, I stressed that I had rejected all attempts at the justification of theories, and that I had replaced justification by criticism:237 we can never justify a theory. But we can sometimes “justify” (in a different sense) our preference for a theory, considering the state of the critical debate; for a theory may stand up to criticism better than its competitors. To this the objection may be made that a critic must always justify his own theoretical position. My answer is: he need not, for he may significantly criticize a theory if he can show an unexpected contradiction to exist either within a theory, or between it and some other interesting theory, though of course the latter criticism would not as a rule be decisive.238 Previously, most philosophers had thought that any claim to rationality meant rational justification (of one’s beliefs); my thesis was, at least since my Open Society, that rationality meant rational criticism (of one’s own theory and of competing theories). Thus traditional philosophy linked the ideal of rationality with final, demonstrable knowledge (either proreligious or anti-religious: religion was the main issue) while I linked it with the growth of conjectural knowledge. This itself I linked with the idea of a better and better approximation to truth, or of increasing truthlikeness or verisimilitude.239 According to this view, finding theories which are better approximations to truth is what the scientist aims at; the aim of science is knowing more and more. This involves the growth of the content of our theories, the growth of our knowledge of the world.
Apart from a restatement of my theory of knowledge, one of my aims in the Postscript was to show that the realism of my Logik der Forschung was a criticizable or arguable position. I stressed that Logik der Forschung was the book of a realist but that at that time I did not dare to say much about realism. The reason was that I had not then realized that a metaphysical position, though not testable, might be rationally criticizable or arguable. I had confessed to being a realist, but I had thought that this was no more than a confession of faith. Thus I had written about a realist argument of mine that it “expresses the metaphysical faith in the existence of regularities in our world (a faith which I share, and without which practical action is hardly conceivable)”.240
In 1958 I published two talks, partly based on the Postscript, under the title “On the Status of Science and of Metaphysics” (now in Conjectures and Refutations.241) In the second of these talks I tried to show that metaphysical theories may be susceptible to criticism and argument, because they may be attempts to solve problems—problems perhaps open to better or less good solutions. This idea I applied in the second talk to five metaphysical theories; determinism, idealism (and subjectivism), irrationalism, voluntarism (Schopenhauer’s), and nihilism (Heidegger’s philosophy of nothingness). And I gave reasons for rejecting these as unsuccessful attempts to solve their problems.
In the last chapter of the Postscript I argued in a similar way for indeterminism, realism, and objectivism. I tried to show that these three metaphysical theories are compatible and, in order to show the compatibility by a kind of model, I proposed that we conjecture the reality of dispositions (such as potentials or fields) and especially of propensities. (This is one way of arguing in favour of the propensity interpretation of probability. Another way will be mentioned in the next section.)
But one of the main points of that chapter was a description and appreciation of the role played by metaphysical research programmes;242 I showed, with the help of a brief historical sketch, that there had been changes down the ages in our ideas of what a satisfactory explanation ought to be like. These ideas changed under the pressure of criticism. Thus they were criticizable, though not testable. They were metaphysical ideas—in fact, metaphysical ideas of the greatest importance.
I illustrated this with some historical remarks on the different “metaphysical research programmes that have influenced the development of physics since the days of Pythagoras”; and I proposed a new metaphysical view of the world, and with it a new research programme, based on the idea of the reality of dispositions and on the propensity interpretation of probability. (This view, I now think, is also helpful in connection with evolution.)
I have reported here on these developments for two reasons.
(1) Because methaphysical realism—the view that there is a real world to be discovered—solves some of the problems which are left open by my solution of the problem of induction.
(2) Because I intend to argue that the theory of natural selection is not a testable scientific theory, but a metaphysical research programme; and although it is no doubt the best at present available, it can perhaps be slightly improved.
I will not say more about point (1) than that, when we think we have found an approximation to the truth in the form of a scientific theory which has stood up to criticism and to tests better than its competitors, we shall, as realists, accept it as a basis for practical action, simply because we have nothing better (or nearer to the truth). But we need not accept it as true: we need not believe in it (which would mean believing in its truth).243
About (2) I will say more when I come to discuss the theory of evolution in section 37.