Remarks on Simon's Comments

Herbert Simon has commented on my article, to which I have been asked to respond. This sequence of events is slightly misleading: It suggests the familiar article-critique-rebuttal pattern, whereas I believe that, in fact, Simon and I are in substantial agreement about the nature of causation and its formal model. A few points of disagreement do remain, though, and my purpose in these remarks will be to underscore these points and restate my opinions regarding them. Among the issues that are not in dispute are the elegance of causal ordering, and the utility of the notion in tasks such as diagnosis. The concept is as attractive now as it was when it was first introduced in the 1950s (I first encountered it in Lerner, 1965). Also, to the extent that my brief comments on it in my article suggested that the notion of causal ordering is inherently algebraic and precludes a sentential version, I am glad that the record is now set straight. I disagree, however, that causal ordering is “isomorphic” to my nonmonotonic account of causation. I wish it were: A major attraction of causal ordering is its simplicity, which contrasts with my almost embarassingly complex technical development. In fact, as alluded to in my article, one of the reasons that technical development is preceded by an intuitive exposition is the potential for recapturing the ideas in simpler form. Isomorphism between structures is defined only relative to specified operations, and it is not clear to me which operations are relevant here. However, such mathematical considerations seem beside the point, and Simon’s position seems clear enough. I will take it to be that (a) causal ordering possesses all the desirable properties of causation possessed by the nonmonotonic account; (b) causal ordering does not introduce any undesirable properties; (c) by my own admission, causal ordering is simpler; and, therefore (d) causal ordering is, to use a phrase due to the late Alan Perlis, an improvement on its successor. In my article I list eight properties that I wish to associate with causation. Of these, Simon agrees with all but two, and expresses partial agreement