Computer by Martin Campbell-Kelly; William Aspray; Nathan Ensmenger; Jeffrey R. Yost

Author:Martin Campbell-Kelly; William Aspray; Nathan Ensmenger; Jeffrey R. Yost [Yost, Martin Campbell-Kelly; William Aspray; Nathan Ensmenger; Jeffrey R.]
Language: eng
Format: epub
Tags: Software Development, Software Design, Business & Management, vl-nfcompvg, Science & Math, Programming, Graphic Design, Culture, Software Design; Testing & Engineering, Computers & Technology, History
ISBN: 0813345901
Amazon: B00BAH8HU2
Publisher: Westview Press
Published: 2013-07-09T00:00:00+00:00

THE SOFTWARE CRISIS

The emergence of a software contracting industry helped alleviate one of the most immediate challenges associated with programming computers—namely, finding enough experienced programmers to do the work. And the development of high-level programming languages and programming utilities helped eliminate much of the cost and tedium associated with finding and eradicating programming bugs. But by the end of the 1960s a new set of concerns had emerged about software and software development, to the point that many industry observers were talking openly about a looming “software crisis” that threatened the future of the entire computer industry. For the next several decades, the language of the “software crisis” would shape many of the key developments—technological, economic, and managerial—within electronic computing.

One obvious explanation for the burgeoning software crisis was that the power and size of computers were growing much faster than the capability of software designers to exploit them. In the five years between 1960 and the arrival of System/360 and other third-generation computers, the memory size and speed of computers both increased by a factor of ten—giving an effective performance improvement of a hundred. During the same period, software technology had advanced hardly at all. The programming technology that had been developed up to this time was capable of writing programs with 10,000 lines of code, but it had problems dealing with programs that were ten times longer. Projects involving a million lines of code frequently ended in disaster. By the end of the 1960s, such large software development projects had become “a scare item for management . . . an unprofitable morass, costly and unending.” The business literature from this period is replete with stories about software projects gone wrong and expensive investments in computer technology that had turned out to be unprofitable.

If the 1960s was the decade of the software debacle, the biggest debacle of all was IBM’s operating system OS/360. The “operating system” was the raft of supporting software that all the mainframe manufacturers were supplying with their computers by the late 1950s. It included all the software, apart from programming languages, that the programmer needed to develop applications programs and run them on the computer. One component of the operating system, for example, was the input/output subroutines, which enabled the user to organize files of data on magnetic tapes and disk drives. The programming of these peripheral devices at the primitive physical level was very complex—involving programmer-years of effort—but the operating system enabled the user to deal not with physical items, such as the specific location of bits of data on a magnetic disk, but with logical data, such as a file of employee records. Another operating system component was known as a “monitor” or “supervisor,” which organized the flow of work through the computer. For IBM’s second-generation computers, the operating software consisted of about 30,000 lines of code. By the early 1960s a powerful operating system was a competitive necessity for all computer manufacturers.

When IBM began planning System/360 in 1962, software had become an essential