The Rules of the Game:


Comments on Professor Jarvie's Chapter

Joseph Agassi

Tel-Aviv University and York University, Toronto

All rights reserved by the author. This file may be downloaded and/or printed only for the purposes of individual private study. The making of multiple copies, either directly or indirectly, is forbidden.

Department of Philosophy
Tel-Aviv University
joseph agassi
agass@post.tau.ac.il


Professor Jarvie always manages to open a simple project that to begin with seems quite proper but not too exciting and to turn it at once into an exciting and thought-provoking study. This time he began by stringing together the different passages in Karl Popper's classical book on method, so as to view it as a sociology of science proper. The result is most intriguing. The exercise is presented as a study of Popper's break away from tradition, his replacement of methodology-as-psychological with methodology-as-sociological. This is expressed by the shift of questions from the traditional, how do I learn/know? to, how do WE learn/know? Popper's shift was shared by Michael Polanyi, with less challenging results, however, as in the name of the scientific tradition Polanyi replaced the traditional republic of science with the scientific community and appealed to authority. Worse replacements followed. One famous philosopher, Larry Laudan by name, has claimed that the "real" question is, why should I believe my colleague the physicist when he tells me who is the leading physicist today? This is obscurantism.
Professor Jarvie approaches the social component of science in accord with his study of rationality. He and I rebelled against the division of rationality into rational thought and rational action. Instead, we took thought to be in part opinions, as parts of the conditions in which rational action occurs, and at times as thinking, which is rational action. (This move requires the admission of degrees of rationality, but I will not discuss this now.) Rather, let me show how all this can help solve an old problem. It is, does methodology prescribe or describe? Authors may say whether they describe or prescribe. This does not help, since matters are complex: prescriptions are of the possible, and since the proceduresto be described are in fact prescribed in the republic of science. How then should we go about this problem?
Professor Jarvie's solution is quite unexpected: method emerges from the mere fact that scientific research is social and rational (it is a partly institutionalized rational action). For two reasons the problem is particularly fascinating for students of Popper's views. First, he demanded least and promised least: he promises no success and saw success in the development and testing of (possibly true) explanations. He observed that the game of science is fruitful, and, not wishing to guarantee that it will remain so, he did not try to explain this fruitfulness; as long as it is found fruitful, he said, people will play it. The second reason is more intriguing: Popper said, the rules of logic should suffice for scientific research. In what sense can this be true? in the sense of the logic of the situation. This makes Professor Jarvie's study exciting.
Is methodology descriptive or prescriptive, then? Professor Jarvie says, the logic of the situation of research mediates between the two options; it also allows for some explanation of the divergence between the institutionally prescribed and the institutionally practiced. Popper's theory is then both prescriptive and descriptive of the better practices. Also, the logic that Popper said suffices for science is the logic of the situation, not logic in the narrow sense of the word. Any activity obeys some rules, and they are known as the logic of that activity. Similarly one speaks of the rules of the game, where the game is the institutionalized activity in question. Methodology is such a game. What are its rules and how binding are they and how much of the procedure do they prescribe?
This general question has no general answer, though for special cases special answers are available. Take rituals; in some systems ritual is more stringent than in others. In most religions there is a standard formula for prayers and even for the choice of which prayer to perform on which occasion -- sometimes details may be added to fit the occasion and sometimes private improvised prayers are allowed or encouraged. In the arts the rules are more liberal, especially in the West. But everywhere rules limit variation, and at times they are broken. In novel writing permitting the choice of commoners for heroes was revolutionary. Traditionally, some areas were closed even to research, such as theology and sexology, perhaps even the sociology of science. This, to the best of my knowledge, is ignored in the republic of science, especially by methodologists. I mention this not because I think that it is significant, I really do not know, but because as an anthropologist Professor Jarvie takes it for granted that the rules described and the rules followed need not be the same. Researchers want such variances explained and reformers want them changed, and he is both.
This is contrary to the latest fashion in the sociology of science, which is fashionable because it repeatedly presents researchers as having no intention to follow the canons of proper conduct, that to the extent that they follow the rules, they do so only out of fear of discredit. In other words, it presents as the chief interest of researchers not the satisfaction of curiosity but the sale of their wares. Such people do exist; they are pretenders, though. Pretenders appear in different field, and their conduct is of little interest. Professor Jarvie's point is much more interesting: his interest is in the variance between rules described and rules followed by honest researchers. What are the rules regarding research and how closely are they followed by the better and more honest researchers and why? Honest discussion of this may reduce pretence such as that of the fashionable sociologists of science.
Rules of research first appeared around 1600, during the scientific revolution. Galileo said then, researchers must know mathematics, as the Book of Nature is written in mathematical symbols. Bacon said then, researchers must believe only their own eyes. Today these rules sound strange, because of the great influence of William Whewell of the mid-nineteenth century. He said, researchers develop explanatory ideas, test them empirically, and accept the results of the tests: refutations lead to their rejection and confirmations render them scientific. He viewed mathematics as essential for science for reasons different from those of Galileo. Popper closed this chapter when he said, mathematics helps furnish testable theories. Bacon's rules are still popular though Whewell proved them impossible when taken strictly and useless otherwise.
Whewell and Popper differed significantly: the former recommended to stop testing confirmed theories and the latter recommended the opposite. Both recommendations are problematic, since there is no rule as to when to stop tests, yet they must be stopped to avoid stagnation. Also Whewell recommended and Popper rejected Bacon's rule of "the ladder of Axioms": Bacon assumed the existence of degrees of abstractness of theories, and he demanded that moves be made from one degree to the next, without skipping a step. Popper disagreed: the more abstract is more testable, he said, and he recommended to go for the highest degree of testability. For my part I am wary of all this: I do not know what the steps in the ladder of axioms are; I do not know if and how degrees of abstraction can be assessed. The same goes for degrees of testability. All we know is, in any valid inference the conclusion is not more testable than the premise. Does this holds for abstraction too?
As I moved from Whewell to Popper, I skipped the conventionalist school of Duhem and Poincare. They suggested completely new rules: do not consider any theory as final, but no scientific hypothesis is ever to be removed, even when a more sophisticated alternative to it is available, since for some purpose the less sophisticated one is the better means, namely, the simpler. Also, when devising an alternative to a theory, deviate from it minimally. I think the notions of small deviation and of simplicity are mere metaphors here.
This is the literature on the rules of the game. I did not mention Polanyi and Kuhn, as they say, reject all explicit rules; worse still, they say, the scientific tradition recommends the appointment of scientific leaders to prescribe scientific public opinion. This is true to some extent, especially after World War II, but it is not science, and I have no good word for it. Perhaps we better not speak of science at all but of the search for the truth instead. This will at once render conventionalism a non-option, as the chief aim that its fans ascribe to science is practical, not theoretical.
The overturning of Newton's mechanics refuted the views of Bacon and of Whewell. The great gap between it and its heirs refuted conventionalism too. Some conventionalist historians, notably Whittaker, have declared the gap small. This is incredible. The situation then looks like a blind alley. In response great historian of science A.N. Meldrum suggested to replace the logic of science with a psychology of science. Kuhn endorsed this, even though it clashes with Polanyi's sociologism that he endorses too. Popper stuck to his sociologism and presented methodology as a part of logic. Even were he utterly mistaken, this would make him one of the greatest. Philosophers who ignore him, for good reasons or bad, are plainly unphilosophical and irresponsible.
There are all sorts of games, and they have their rules, their logics; to abide by logic, to be logical proper, the moves allowed and/or prescribed by any game must be allowed by logic. It is very important to notice that these are very limited to two items. First, contradictions must be deemed false. Hence their negations must be deemed true. This follows from the basic demand of logic: in order to be valid, an inference must comply with the rules of transmission of truth. This rule is very easy to follow as it does not hold for surmise: when an inference is found invalid, it may be still upheld, but as a mere surmise. Alternatively, it may be rescued by the claim that some premises were left unstated but understood in context.
Consider the game of the axiomatization of a given system, then. It allows neither surmise nor the omission of premise: it demands of all axioms to be stated explicitly, without omission or repetition, and without contradiction, and to entail all the theorems of the system in strictly valid inferences. The logic of axiomatization is more strict than logic proper, but is not in any way opposed to it, of course. Is the logic of science of the same ilk?
Logic permits inductive inferences on the condition that they are viewed as mere surmises. This was stressed by Bertrand Russell, for example. He wanted more: he looked for some further rules to give some inductive inferences the status of more than mere surmise. This extra bit required by Russell, said Popper, is neither possible nor necessary. Suffice it to require, he said, that surmises should be put to empirical test. Can all surmises be tested? No. Only the more promising ones should. What do tests tell? Logic says that successful tests attest that the surmise tested are false and unsuccessful test attest that if the surmise is an error this test does not evince this (though another test might). Russell said, this is not enough. Popper said, it is enough for research. What should researchers do when an error is found? Popper said, admit the reports of the empirical information as true so that thereby the surmise is declared false so that dogmatism is avoided. Now clearly dogmatism does often clash with the aim of the search for the truth, and then -- but only then -- it is not allowed. Is the requirement to endorse information a necessary and sufficient condition for the search for the truth? The methodical study of the logic of the situation invites the study of this question.
The reason Russell was dissatisfied with Popper's suggestion is that the two were not sufficiently clear about the logic of the situation. Russell considered the aim of science both the search for the truth and the guidance for scientific technology. Popper did not discuss guidance; his social and political philosophy allows for such a discussion, however. This too is required by the logic of research: research as an activity may be motivated by different kinds of aims. So there is no need to insist that science is the search for the truth; what Popper presented as the logic of science may be taken to be a part of it, namely, the logic of the search for the truth. What conventionalism-instrumentalism presented as the logic of science is likewise incomplete, and may be taken as the logic of the search for utility, i.e., the logic of technology. This will show that conventionalism-instrumentalism is in error even about utility, since, as Bacon said, the search for the truth may be more useful than the search for usefulness. Bacon's claim is not always true: engineers may find it more useful to center on technology rather than wait for some handy scientific progress; socially, however, Bacon's claim is admitted by all. So in a sense even Bacon, the father of modern methodology and the father of the psychologistic trend in it, he too had in mind research as social not as psychological. Except that he was not so clear about matters as we ought to be.