Browse Prior Art Database

Anecdotes: Fast Predictors: Computers and the U.S. Presidential Elections

IP.com Disclosure Number: IPCOM000129571D
Original Publication Date: 1988-Jun-30
Included in the Prior Art Database: 2005-Oct-06
Document File: 5 page(s) / 25K

Publishing Venue

Software Patent Institute

Related People

A. F. Draper: AUTHOR [+2]

Abstract

One of the reasons why the name UNIVAC was interchangeable with ";computer"; during the bulk of the 1950s was because of its very public performance accurately predicting the winner of the 1952 U.S. presidential election with only three percent of the votes counted. This even surprised its programmers, as related in our first anecdote about UNIVAC and the Eisenhower-Stevenson battle of that year. By the time the 1960 election came around, all three major television networks had computer assistance on Election Day. The second anecdote consists of edited excerpts from an IBM report on its highly successful collaboration with CBS News in that hotly-contested struggle between Kennedy and Nixon. That night contained many surprises uncannily predicted by the combination IBM RAMAC 305 and 7090 processors. Both anecdotes give a good idea of the complexity of the calculations necessary, and the extreme attention to accuracy required for ";real time"; election reporting.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 19% of the total text.

Page 1 of 5

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

Copyright ©; 1988 by the American Federation of Information Processing Societies, Inc. Used with permission.

Anecdotes: Fast Predictors: Computers and the U.S. Presidential Elections

A. F. Draper

Remington Rand Laboratory of Advanced Research

Introduction

One of the reasons why the name UNIVAC was interchangeable with "computer" during the bulk of the 1950s was because of its very public performance accurately predicting the winner of the 1952 U.S. presidential election with only three percent of the votes counted. This even surprised its programmers, as related in our first anecdote about UNIVAC and the Eisenhower- Stevenson battle of that year. By the time the 1960 election came around, all three major television networks had computer assistance on Election Day. The second anecdote consists of edited excerpts from an IBM report on its highly successful collaboration with CBS News in that hotly-contested struggle between Kennedy and Nixon. That night contained many surprises uncannily predicted by the combination IBM RAMAC 305 and 7090 processors. Both anecdotes give a good idea of the complexity of the calculations necessary, and the extreme attention to accuracy required for "real time" election reporting.

UNIVAC on Election Night

It is my understanding that the general purpose of this session*1 is to discuss the use of computers. As this is the case, it is, therefore, quite characteristic that I should be selected to give you a discussion on the misuse of a computer. I am referring, of course, to the last election night debut of UNIVAC to the world's largest television audience that has ever existed. Various newspaper writers on this subject have expounded at great length on the injustice we did to poor old UNTVAC and how we undoubtedly hurt his feelings. It is, therefore, my job to stand up here and make a public apology to UNIVAC in the hope that he will not build up a neurosis that will interfere with his work for the Atomic Energy Commission.

I presume that most of you know that there are at present six UNIVACS in operation. No. 5, which now belongs to the AEC, was the one who appeared in television; but No. 4, also an AEC machine, was used to check the data for reasonableness, and No. 1 was all ready to take over at a moment's notice in case No. 5 misbehaved.

My real purpose in being here today is to try to give you some idea of just what was attempted on the UNIVAC election night and explain just what we think was accomplished. Essentially, there are two parts to the overall picture. The first of these was the mathematical development of a theory that would make an election prediction possible; and the second, of course, was the actual set-up and running of the problem on the machine as the returns started to come in.

1 *The AIEE Meeting of 22 January 1953.

IEEE Computer Society, Jun 30, 1988 Page 1 IEEE Annals of the History of Computing Volume 10 Number 3,...