EWD249 - 79

We can phrase the observation in more technical terms. The primed version is messy because the image is explained away in terms of lines at too early a stage, thereby forcing us to explain "CLEARFIRST", "ISCANNER" and "COMPPOS" in terms of lines, while they could still be explained in terms of the image, i.e. independent of the representation to be chosen for it. Or, in other words, in the original version we have made a more judicious exploitation of our power of abstraction than in the primed one. The larger the number of pearls independent of the particular representation, the more adaptable one's program and the more easily understandable -because that set of pearls can be understood at a higher level of abstraction. The experience seems to indicate that the goals of adaptability and clarity have been given some substance and (what is more) go by their very nature hand in hand. This is very encouraging (although not surprising).

It also gives -me at least- a somewhat clearer picture of the scope of my present efforts. Whatever I shall develop, it will not be a General Problem Solver, not a mechanical one, not even one written for the benefit of the human problem solver. But it may give the human some appreciation for the various aspects of "elegance" of a solution when he succeeds in finding one. And as such it may give him a guide line.