Crisis sw

Th e SoftwareResearch C risis
ROBERT

L.

GLASS,

Computing Trends

*

With the advantage of

more than 2 X years’ hindsight, this tzventy-first centu y author looks askance at the “crisis” in sof2ware practice and expresses deep concern for a Crisis in software research.

I

t is the year 2020. We in computing and software research are certainly glad we’ been able to put ve the“research crisis” behind us. What was this crisis? It was the realization that occurred, right around the turn of the century, that research in computing and software - as it was then focused - was all too often both arrogant and narrow. It was arrogant because many computing researchers of that era were doing research in a topic they thought they understood, but didn’ It is t. amazing that, inretrospect, those computer scientists simply didn’ t know what they didn’ know. t It was narrow because, of all the possible research models twentieth century computer scientists might

have used, most of them were stuck using only one of them. Even though we now understand how terribly limiting that approach was, it is easy to see, in retrospect, why computer scientists could not see the error intheir ways. Before I continue, I think it is worth spending a moment or two elaborating on the term researchcrisis. There is a fine irony to it, and the story makes good telling. To understand this research crisis, it is important that we confront and deal with a prior “software crisis.” It has been about 50 years now since the software community coined the term sojiware Crisis. It was invented ata conference predominantly attended by theorists rather than practitioners, but still, at the outset, almost

42

07407459/94/$04

W 0 1994 IEEE

NOVEMBER

1994

everyone agreed there was some validity to the term. What did “software crisis” mean? It is hard to believe, now, but what it meant way back then was that the practice of software was in crisis - that it was characterized byprojects that were “always over budget, behind schedule, and unreliable.” With the advantage of hindsight, we can look back on the last century as a time of wonder. Of all the dramatic changes that took place in the world, computing and software gave rise to the name “computing era” that we now recognize as the proper characterization of the last half of the twentieth century. Oh, there weresoftware project failures, of course. Some of them were even catastrophic - runaway described the worst of them. But their existence provided only anecdotal evidence, certainly not solid data, to support the claims of crisis. In fact, the issue of solid data became the shoal on which the claims of crisis foundered, as we will see later in this story. Software crisis indeed! We may look back on thatearlier time as archaic and ignorant, perhaps, but certainly not as a crisis. It was the beginning, in fact, of the Golden Age of Computing Practice that persists today. Nevertheless, the notion of software crisis reached a fever pitch of in the 1980s and 1990s. intensity Researchers began nearly every paper on software engineering by invoking the software crisis as a reason for listening to whatevernew theoretical notion they were advocating. Researchers at the time didn’ realize that t what many of them were doing would later be characterized, derisively, as advocacy research. But the software crisis was in fact the platform on which most of this advocacy research was founded.

SOFTWARE CRISIS DISCREDITED
At the time, researchers frequently cited both the anecdotal evidence referred toearlier and a government study from the US General Account-

ing Office on the problems of building software, as support for their claims of crisis. In most cases, this GAO study was the only real, nonanecdotal data cited, and almost every researcher cited it. It seemed to say that most software projects failed, and that most money spent on software was wasted. Fortunately, cooler heads...