1 The problem of moral philosophy is that of explaining in some way the workings of those parts of language which are concerned with morals. It is worthwhile considering first of all what sort of thing an explanation is, for if we take too narrow a view of explanation then we may find that there is no explanation of these things, and if we take too broad a view of it then we admit explanations which though not false do not seem at all illuminating. The progress of science has been very much dependent upon changes in what is acceptable as an explanation, the abandonment of teleological explanations (or at least their decline), the demand for predictions or falsifiability, the preference for quantification, and more recently, the acceptance of abstract mathematical predictive 'explanations' without any supporting analogy of the billiard ball variety, are some of these changes. In philosophy too shifts in the notion of explanation have had their importance, though these are less easy to pinpoint, perhaps because they are too pervasive. In contemporary moral philosophy every theory seems to offer an entirely different kind of explanation, linked only by the desire to say something about ethical language which is neither glaringly obvious nor false. N1
2 I would like first of all to outline a sort of explanation which would in some sense constitute a complete, ultimate explanation of our ethical language (in fact, of everything), and then to consider some ways in which this sort of explanation might nevertheless be unsatisfactory, and how it relates to more satisfactory explanations. What we are trying to discover is a pattern. There are two distinct criteria which we can apply in evaluating such patterns as we discover. The first is completeness and the second is simplicity. If we are using our pattern for making predictions then it is complete if it enables us to predict everything, it is more or less complete as it enables us to predict more or fewer characteristics of our subject matter. It is simple or not, depending on how easy it is to arrive at the conclusions. The sort of complete knowledge I am speaking of here is not quite the same as knowing every true proposition, for it is possible, even though not knowing every true proposition about some system, to know a subset of the true propositions from which every other proposition can be deduced. If I have before me a list giving the age of every philosophy student at Keele University, then even though I do not know the average age of Keele philosophy students, I am in a position to deduce the age from propositions all of which I do know (assuming that there are among these a few mathematical truths).
3 It is suggested sometimes that the fundamental laws of physics are, at least potentially, complete in the sort of sense I am speaking of. If true, this would mean that the set of law-like statements which are true contains only the fundmental laws of physics and those law-like statements which are deducible from the fundamental laws of physics. What I want to consider, is whether there could be a set of basic propositions sufficient to generate by logical deduction, a larger set of propositions containing all true propositions about the use of moral language and no false propositions.