Improving computer program readability to aid modification

Abstract

The modification of computer programs is a costly and constant job. An informal survey conducted at General Motors and reported on by Elshoff [6] concluded that about 75 percent of all programmer/analysts' time in a commercial data processing installation is spent on program modification. This conclusion agrees with independent assessments made by Liu [16], Boehm [2], and Lientz and Swanson [15]. Moreover, the reasons for modifying programs will not disappear. As pointed out by Lehman [14], all programs are models of some part of the real world and, as the world changes, programs must be modified to keep pace with these changes or they become progressively less relevant, less useful, and less cost-effective. As new software is developed, the inventory of programs to be maintained grows, and thus this high level of modification work is not expected to decrease.