Cliff Shaw (1922–1991), formally John Clifford Shaw, was an American computer scientist and systems programmer at the RAND Corporation. He collaborated with Allen Newell and Herbert A. Simon on some of the earliest working programs in artificial intelligence, including the Logic Theorist and General Problem Solver. He also helped develop the Information Processing Language family and led the initial development of JOSS, an early interactive computing system.
Work at RAND
Shaw’s principal contributions emerged from RAND’s computing research during the 1950s and 1960s. At a time when computers were used primarily for numerical calculation, he helped develop techniques for manipulating symbols, logical expressions, and structured descriptions of problems.
His collaboration with Newell and Simon combined research on human problem solving with the practical construction of executable programs. Shaw contributed programming methods and system implementations that made their proposed reasoning procedures workable on the limited hardware of the period. The resulting work became important to both artificial intelligence and the development of symbolic programming.
Early artificial intelligence programs
Logic Theorist
The Logic Theorist, developed in the mid-1950s, was designed to prove theorems from the propositional-logic portion of Principia Mathematica. It is widely recognized as one of the earliest artificial intelligence programs.
Rather than exhaustively enumerating every possible proof, the program used heuristics to select promising steps. Its operation demonstrated that a computer could search through symbolic expressions and construct proofs using procedures intended to resemble aspects of human reasoning.
Shaw worked with Newell and Simon on the program’s implementation and the symbolic-processing facilities it required. The project helped establish automated theorem proving as a practical research activity, although its scope was restricted compared with later theorem-proving systems.
General Problem Solver
Shaw also collaborated on the General Problem Solver (GPS), developed beginning in the late 1950s. GPS sought to separate general problem-solving procedures from the descriptions of particular tasks.
Its characteristic technique was means–ends analysis: comparing a current situation with a desired goal, identifying a difference, and selecting an operation that could reduce that difference. If an operation’s prerequisites were not satisfied, the program could treat them as subsidiary goals.
GPS became influential in symbolic artificial intelligence and research on human cognition. Its name described an ambition rather than unrestricted competence: successful use depended on suitable formal representations, available operations, and manageable search spaces.
Information Processing Language
Newell, Shaw, and Simon developed Information Processing Language (IPL) to support their symbolic programs. Beginning in the 1950s, IPL provided facilities for organizing and manipulating lists and other structures that were difficult to express using the numerical programming tools then available.
IPL was an important early setting for the development of list-processing techniques. A linked list represents a sequence through connections between its elements rather than requiring every element to occupy a consecutive memory location. Such structures suited reasoning programs whose expressions and relationships changed during execution.
Shaw’s work helped demonstrate how flexible data structures could support symbolic computation. IPL preceded Lisp, which subsequently became much more widely used for artificial intelligence research. The two belonged to the early development of list processing but differed substantially in their notation and organization.
JOSS and interactive computing
In the early 1960s, Shaw led the initial development of JOSS, the JOHNNIAC Open-Shop System, at RAND. The original system ran on JOHNNIAC, RAND’s stored-program computer.
JOSS allowed users to enter expressions and instructions at remote typewriter terminals and receive responses interactively. It was designed to make numerical calculation accessible without requiring users to prepare conventional batch-processing jobs or master a complex programming language.
JOSS combined a relatively approachable language with an interactive computing environment. It was an early example of time-sharing and a significant development in human–computer interaction, emphasizing direct access and rapid feedback. Later implementations and related systems extended its influence beyond the original RAND installation.
Historical significance
Shaw’s contributions connected early theories of machine reasoning with the programming infrastructure needed to test them. Logic Theorist and GPS explored heuristic problem solving; IPL supplied tools for representing symbolic structures; and JOSS addressed the practical accessibility of computing.
These achievements were collaborative. Shaw’s particular historical importance lies in the design and implementation of working systems, rather than in a single independently attributable theory or invention.