A general game-playing program must know the rules of the particular playing game. These rules are:(1) an algorithm indicating the winning state;(2) an algorithm enumerating legal moves. A move gives a set of changes from the present situation.There are two means of giving these rules:(1) We can write a subroutine which recognizes if we have won and another which enumerates legal moves. Such a subroutine is a black box giving to the calling program the answer: 'you win' or 'you do not win', or the list of legal moves. But it cannot know what is in that subroutine.(2) We can also define a language in which we describe the rules of a game. The program investigates the rules written with this language and finds some indications to improve its play. Artificial Intelligence and Heuristic Programming Edinburgh University Press

To every 3-dimensional scene there correspond as many 2-dimensional pictures as there are possible vantage points for the camera. It is, however, possible to construct pictures for which there is no corresponding scene containing physically -realizable objects. Pictures of such'impossible objects' can be useful in giving insight into the constraints or grammatical rules associated with the'language' of pictures, just as nonsense sentences can be useful in illustrating the rules of other languages. Impossible objects have been used by psychologists (Penrose and Penrose 1958) to create visual illusions which successfully challenge the ability of our perceptual systems to synthesize a 3-dimensional world from 2-dimensional information. The incompatibilities among the various portions of pictures of these objects are a novel way of testing our picture analysis procedures. The purpose of this paper is to demonstrate some possible decision procedures and to test them on pictures of both possible and impossible objects.

In this paper, we develop the algorithm, given in Plotkin (1970), for findingthe least generalization of two clauses, into a theory of inductive generalization.The types of hypothesis which can be formed are very simple. They allhave the form: (x)Px --> Qx.We have been guided by ideas from the philosophy of science, followingBuchanan (1966). There is no search for infallible methods of generatingtrue hypotheses. Instead we define (in terms of first-order predicate calculus)the notions of data and evidence for the data. Next, some formal criteria areset up for a sentence to be a descriptive hypothesis which is a good explanationof the data, given the evidence. We can then look for the best such hypothesis.Machine Intelligence 6

One of the central principles upon which intelligent devices seem to operate is that of maintaining internal models of their external environments. How difficult this is, depends upon both the complexity of the model and its method of representation. In particular, it is usually easy when the problem is posed in the classical heuristic search paradigm, and the data structures used to represent static configurations of the puzzle are relatively unproblematic (arrays, lists, and so on). The lack of side-effects reflects the simplicity of the physics which such models embody. This limitation to elementary forms of interaction is not, of course, intrinsic to the heuristic search method; but when more complex models are constructed it becomes less trivial to pursue the consequences of performing an action. This approach is more general than the heuristic search method (but the latter -- when it has sufficient expressive power -- wins at present by its computational advantage). Assertions mentioning several different situations can then be used to describe dynamical laws which move us from one situation to another. But in some ways the resulting sharp separations between states of affairs are an embarrassment. For if we distinguish two situations s1 and s2, then from the fact, if such it be, that a predicate p is true of Si, nothing whatever follows concerning s2. And this is true even when s2 is directly associated with sl. Say s2 results from s1 by the performance of some action: s2 do (a, si) then no matter how remote -- speaking intuitively -- the connection between the property p and the action a, it still does not follow that p is true of s2. If we want it to so follow we must state this explicitly. Now, unfortunately, there are innumerable facts which might remain unchanged when actions are performed. So instead of writing a law of motion' in the form A(s) B(do(a, s)) where A and B are fairly short expressions, we are apparently obliged to list systematically all conceivable facts which are not changed. So that the law looks more like (Ci(s)& Ci(do(a, s))& & C„(do(a, s))&B(do(a, s)) for some very large n. This works for small problems (such as the familiar hungry anthropoid), but these are usually better formalized in the heuristic search paradigm anyway.

Amsterdam: North-Holland.

New York: American Elsevier.

New York: McGraw-Hill.

JOINDER OF CLAIMS, COUNTERCLAIMS, AND CROSS-COMPLAINTS: SUGGESTED REVISION OF THE CALIFORNIA PROVISIONS. Research in artificial intelligence, a branch of computer science, has illuminated our capacity to use computers to model human thought processes. In this Article we will argue that the time has come for serious interdisciplinary work between lawyers and computer scientists to explore the computer's potential in law. Interdisciplinary work between the lawyer and the computer scientist has floundered on the misconceptions that each has of the other's discipline. As a result, no one has yet attempted computer programs incorporating complex techniques of legal reasoning. Even efforts in legal information retrieval have been hampered by these misconceptions. In retrieval, lawyers have viewed the computer as, at most, a storehouse from which cases and statutes might be retrieved by skillfully designed indexing systems. But the lawyer rarely looks for, or even expects, clear answers. So far, the efforts in legal retrieval have given little consideration to the possibility that computers might operate on the legal data base the way a lawyer does. Yet the work in both fields law and computer science -,suggests that the computer modeling of legal reasoning would be a fruitful area for research. In this Article we speculate about the dimensions and possible directions of this research. Under the most promising of outcomes, interdisciplinary research could lead both to a greater understanding of the legal reasoning process and to the design of machine methods for performing parts of it. The prospect of using computers to model legal reasoning processes is likely to prompt a typically lawyer-like response: So what if we understand legal reasoning or legal argument formation better? Knowing more about the ways in which lawyers search and manipulate the legal data base might lead to improving the lawyer's skill at his work. We recognize the possibility that the work of many lawyers might actually involve little use of the legal data base for argument construction or dispute resolution.

Ph.D. dissertation "Bi-directional and heuristic search in path problems" (Stanford, Computer Science, 1970) summarized in this article in Machine Intelligence 6 (1971).In the uni-directional algorithms, the search proceeds from an initial nodeforward until the goal node is encountered. Problems for which the goal nodeis explicitly known can be searched backward from the goal node. Analgorithm combining both search directions is bi-directional.This method has not seen much use because book-keeping problems werethought to outweigh the possible search reduction. The use of hashingfunctions to partition the search space provides a solution to some of theseimplementation problems. However, a more serious difficulty is involved.To realize significant savings in bi-directional search, the forward andbackward search trees must meet in the 'middle' of the space. The potentialbenefits from this technique motivates this paper's examination of thetheoretical and practical problems in using bi-directional search.

This paper explores new approaches to the symmetric traveling-salesman problem in which 1-trees, which are a slight variant of spanning trees, play an essential role. A 1-tree is a tree together with an additional vertex connected to the tree by two edges. We observe that (i) a tour is precisely a 1-tree in which each vertex has degree 2, (ii) a minimum 1-tree is easy to compute, and (iii) the transformation on “intercity distances” cij → Cij + πi + πj leaves the traveling-salesman problem invariant but changes the minimum 1-tree. Operations Research, 18, 1138–1162.