The Interface of Computer Science and Statistics: An Historical Perspective

Abstract

Current scientific and -technical research is characterized by a rapidly expanding diversity of subject matter, magnitude of effort, and quantitative orientation of analysis. The principles and techniques which have been applied so successfully to the natural sciences and engineering are being adapted to the behavioral and life sciences. A similar trend is manifesting itself in contemporary research on the management of large organizations, both private and public. The diversity of subject matter leads to the creation of interdisciplinary task forces, which amplifies the long-existing need for a common analytical language -to facilitate the effective communication across discipline boundaries. In addition, the magnitude of effort and quantitative orientation of analysis combine to generate a tremendous volume of both qualitative (non-numerical) and quantitative (numerical) information for quantitative analysis. Two relatively young professions, computer science and statistics, have become indispensable to this research. They serve both as essential modern dialects of the classical analytical language of mathematical science, and as necessary instruments in the intelligent generation and analysis of information. It is inevitable that computer science, statistics, and their interface share, with the more classical portions of matheinatical science, a unique position at the very heart of research. Historically, the interface of computer science and statistics evolved from the interweaving of two strands: (1) the development of the mechanical prerequisites to the computer and the mathematical prerequisites to computer science and (2) the development of probability and statistics. This interweaving is partially the result of the common origin of computer science and statistics in mathematical science, and past contribu-tions of many mathematicians in multiple areas of mathematical science. Nonetheless, it is interesting that so many high-ranking mathematicians have made significant contributions to both. A summary of this interweaving, which chronologically initroduces some of the connections between the two strands, provides an historical perspective fron which to view the interface meaningfully. It is intended to be comprehensive and definitive, but not exhaustive and complete. The summary depends heavily upon Chapter II of HI.M. Walker's Studies in the History of Statistical Method for its discussion of the major contributors to probability, and upon the remnaining publications cited in the Bibliography for their complemen-tary discussions of ina-thematicians and mathematical discoveries. The interweaving began in ancient China. It -then continued from 16th and 17th Century Europe, with Pacioli, Cardano, Pascal, Fermat, Huygens, and Leibniz, to 18th and 19th Century Europe, wi-th Jacques Bernoulli, De Moivre, Legendre, Gauss, Laplace, and Boole. Finally, it entered the United States a-t the end of the 19th Century with Hollerith and has remained here with Von Neumann.