A Description of a Cooperative Venture in the Production of an Automatic Coding System

The exact origin of the idea of a co-operative undertaking to produce an automatic coding system is hard to determine as we look back at the history of the Project for the Advancement of Coding Techniques (PACT). For years, people in service organizations, like the computing groups in which many of us work, have been looking for ways to persuade fellow organizations to "co-operate". Achieving co-operation in this case consists of convincing a fellow worker that he should supply you, gratis, with a set of board wiring diagrams or some similar piece of know-how in the hope that some day you may have something the lender can use. West Coast groups, being younger and less experienced than many of our eastern neighbors, have tended to become experts at this kind of co-operation. The purpose of this paper is not to discuss the merits of this technique, but to point out that PACT is the product of a different sort of co-operative undertaking. The joint venture, PACT, got under way during November of 1954. That was the time when several organizations on the West Coast were taking a fairly critical look at their achievements and future needs in computing equipment. Several of them had decided to turn back the first large digital computer that they had rented in favor of a bigger, faster, and more flexible machine. Several facts emerged clearly during these self-examinations. Problem check-out was expensive and time-consuming. In some cases, check-out time was running as high as 40 % of production time. Estimates of the cost of writing and checking a program put the cost at something like $2 to $10 per instruction. Far too many of the problems that were being run were inefficiently and awkwardly put together, resulting in an excess of machine time for execution. Experienced programmers were hard to find. Nearly every computing organization was expanding. One of the suggested ways of relieving the programmer shortage, and at the same time improving the quality of the program that gets to the machine, was to employ more automatic coding; that is, to let the machine do a larger share of the work of preparing a problem for machine computation. However, the more sophisticated problem preparation programs were by no means an obvious clear-cut answer to the problem of what to do about coding. Every West Coas~ organization had already made considerable investment in utility programs. Everybody had an interpretive system of one kind or another that was aimed at making problems easier to code. These interpreters helped some by reducing : preparation time, but the cost in machine time required to execute the programs was high. Everybody had a stack of library programs for helping to solve the