IMPERFECT KNOWLEDGE-CLAIMS ARE QUALIFIED BY PUBLICLY ACCEPTED HYPOTHESES

15. According to the opinion presented here we have a test for imper­fection of knowledge-claims and a (partial) grading of them. A perfect knowledge claim is utterly unqualified.

Thus, the claim to know that Johnny is arriving in town on the evening train is less imperfect if one also thereby claims to know that the train will not be stopped by kidnappers than if one’s claim is not construed unjusti­fied under such strange circumstances. It is easy to observe that one and the same claim for knowledge may be sometimes unjustified, sometimes not. For instance, compare the claim made by a friend to the seemingly same claim made by a police inspector whose assignment is to guarantee Johnny’s safe passage. Although kidnapping exempts the friend from the responsibility even though the claim has turned out to be false, the very same kidnapping does not exempt the police inspector from responsibility; hence the inspector’s claim for knowledge is rendered unjust in the face of kidnapping. To put it differently, the friend but not the inspector may promise the arrival of Johnny without previously checking and excluding the possibility of a kidnap. The problem still remains, what is the criterion for the intervention of an act of God? Why is a kidnap an act of God for the friend?

16. We can put our theory formally first, without as yet offering a crite­rion for what counts as an act of God. When we speak of perfect knowl­edge, we either speak of a proposition whose probability is one, or of a proposition whose probability is one in the light of given evidence and the given evidence is known perfectly.

There may be a few objections to this formal presentation of the opinion advocated here. I shall not enumerate them. I am more interested in the question of empirical testability of my opinion than in the elegance of a formal variant of it. And, we can easily see, this leads us back to the problem of demarcating acts of God.

17. The opinion presented here, concerning the test for imperfect knowl­edge, may be empirically testable, provided that we do not shift the meaning of ‘an act of God’ in order to avoid refutation. It would be an improvement if we provide and utilize an empirically testable criterion of what people do usually consider as an act of God. Such a theory is provided with detail in the previous two chapters and elsewhere⁶ but it may be briefly sketched here. As a first approximation we may say, whatever kind of event is considered as impossible by all people concerned is, when it happens, an act of God.

When someone promises something which, our accepted views imply, is impossible, say a new invention, he thereby undertakes not to use our accepted views. This very instance indicates that whatever we all share with a claimant for knowledge may ordinarily be construed as a condition for his claim. (This case is not much different from the case of the inspector).

On the other hand, death cancels all appointments (though not all debts), earthquake may wipe out a wide range of business obligations - and these, then, must be construed as acts of God whose exclusions may be conditions for imperfect knowledge. Yet such events are considered all too possible. Hence acts of God are not only what our commonly shared views declare impossible.

18. A better approximation concerning acts of God is this: acts of God are events which we exclude by tacit but public agreement; partly on the ground of theories we all hold to be true, partly on theories which we agree to pretend to hold even though we do not believe them to be true. For the theories we all tacitly hold as true we offer no excuse; we do not even attempt to justify them. Not so for the theories we tacitly pretend to hold as true. These we not only do justify - we even insist that the justifi­cation be so simple and trivial that there is little room for controversy here. The reason for this is all too obvious: we may differ as to whether to adopt a fiction but hold the fiction that we are ready a adopt a fiction. This would make things quickly slip out of hand, especially since agree­ment is here tacit and so not amenable to subtle disquisition; and so every problematic case is to be excluded: whenever there is doubt as to whether a convention is accepted, we usually deny that it is tacitly accepted and demand its acceptance to rest on an explicit statement in a contract or some such.

19. This approximation, too, is probably false, and as yet it certainly is too sketchy and incomplete. It is more than sufficiently elaborate, however to enable the reader to construct with little effort a variety of instances against the equation of imperfect knowledge with opinions endorsed by scientists. That the equation is false is, by the way, though novel to philos­ophers, no news to lawyers. (It is no accident that the doctors of the law were called hypocrites, and by one who did not think much of social convention of any kind.) It is an explicit point of English law that acting on an error and causing damage or even death, may be considered an accident under a certain condition. The condition is that the error is shared by the actor’s community. Reference to the actor’s honest belief, on the one hand, is no support to his defense; and criticism of his commu­nity’s belief from current scientific belief, on the other hand, is no under­mining to his defense. This ruling, obviously, is only possible if some false claims for imperfect knowledge are not culpable. That this, indeed, some­times obtains has been stated earlier.

20. When we refute the identification of claims for imperfect knowledge with claims for scientific knowledge we refute the theory identifying im­perfect knowledge with rational degree of belief with scientific opinion. This is so regardless of whether scientific opinion is declared to be the one highly probable, the one best stood up to severe tests, or the one which best fits our general scientific outlook at present.

What I insist on is that what is the presupposition for a claim for knowl­edge is one which we all take for true; or, rather, one which we all pretend to be true; still more precisely, we do so in a rather unproblematic manner. Abner Shimony suggests that I mention here the fact that there may be different orders of pretending that a proposition is true. The reason may be a matter of technicality, from the simplification of a calculation to the simplification of the application of an idea to a complex and/or pro­blematic situation, and it may be a matter of social attitudes, such as a conservative or an aesthetic one.

21. Finally, the view presented here raises a few new problems. Here are two. First, how do we change rationally the institional framework within which we can claim imperfect knowledge? We have to base our proposed reforms on some imperfect knowledge; and so we seem to be trapped. The answer to this problem must rest on the claim that our institutional framework is built so as to be imperfect and allow inconsistencies so as to allow us to use one part against the other. This is particularly the case when we make an institutional room for science: we neither endorse its conclusions automatically nor allow them to be declared utterly irrelevant (our hypocrisy is tempered by reformism). This leads us to our second question. How do they determine institutional knowledge in sicentific society, whether an inventor’s world, its patent office, clubs, journals, and all; or the U.S. Space Center. I shall not go now into this fascinating discussion.

Let me conclude with one observation which brings back a large chunk of the old view which has been rejected early in this chapter. When examin­ing the content of the presupposition to a claim for knowledge carefully, one always finds some elaborate opinion about the significance of tests and some reports about past tests, namely that they were performed honestly yet without leading to a refutation. And so, a theory of what hypothesis under what test may be ‘projected’ (to use Nelson Goodman’s terminology), is a part of the publicly and tacitly accepted presupposition. But such a hypothesis cannot be generalized as it varies from one commu­nity to another, both concerning the ‘law-likeness’ (to use Popper’s term) or ‘projectability’ of some observations, and as to the severity of the required tests.

Finally, in his last work John Austin has elaborated on Bacon’s corre­lation of truth and promises, attempting to reduce the concept of truth to the concept of promise. The opposite direction seems much more promising - which is why truthful people are also honest people who do not break promises, even when they later do not make the relevant clauses in their promises come true.

NOTES

¹ I remember a number of spoken remarks from Popper which clearly indicate to me that in Popper’s opinion today’s science equals todays well tested but unrefuted hypo­theses. I have severely criticized this opinion in my review of T. S. Kuhn’s ‘The Struc­ture of Scientific Revolutions’, J. Hist. Philos., 1966.1 do not think Popper will expli­citly endorse this opinion.

² We may be still ignorant of the refutation. We may dogmatically deny the truth of some refuting evidence. We may hope it is false and be on the way to checking it. I shall ignore these and similar cases.The reader interested in them may consult Chapter 8 above.

³ As I have learned from Imre Lakatos, the concept of perfect knowledge even in mathematics has undergone deep alterations. Consequently we have to narrow this. Only those who share our concept of a proof will admit they were in error when offering a proof of a false theorem though they should have known better. But even this may go too far. Therefore, though I do think we always tend to frown about erroneous proofs, I do not like this.

⁴ All theories of probability assume the likelihood of an outcome to be the measure of our imperfect knowledge of the outcome; this leads to a well known paradox: for every possible outcome we have a high degree of imperfect knowledge that it will not turn up, and so we have a high degree of imperfect knowledge that though one outcome will certainly show up, none will; which is obviously absurd. The absurd is to declare a perfect knowledge of a likelihood and any likelihood a measure of imperfect knowledge.

⁵ Also, Russell’s analysis is not contradicted by anyone, and in a sense deeper than the claim that p cannot possibly be known: the other analyses of the claim for the certainty of p in the literature on the foundations of mathematics are equivalent to Russell’s in the sense that they can be extended in the same manner both within mathematics and in adjacent fields.

⁶ See also my ‘Conventions of Knowledge in Talmudic Law’, in Bernard S. Jackson (ed.), Studies in Jewish Legal History; In Honour of David Daube, Jewish Chronicle Publications, London 1974 (also published as Special Issue of J. Jewish Studies 25 (1974), 16-34). See also next two chapters.