Software may be growing increasingly ubiquitous. But that just means the problems associated with poor design are growing as well. The Economist cites the Standish Group's estimates that "30% of all software projects are cancelled, nearly half come in over budget, 60% are considered failures by the organisations that initiated them, and nine out of ten come in late." The publication goes on to point to a 2002 study by the National Institute of Standards (NIST), which contended that software errors are costing the American economy $59.5 billion each year. "Worldwide, it would be safe to multiply this figure by a factor of two."
The Economist pins the blame, primarily, on weaknesses in the development process and the relative immaturity of development tools and applications. Fortunately, it sees promise in the areas of software life cycle management, testing, and the open source movement. One driver of improvement, it contends, is the vendor collaboration (particularly between Microsoft and IBM) necessary to successfully compete in the emerging market for Web services.
But all is not quite right: "One snag is that, so far, web services have turned out to be much harder to deliver than their champions had hoped." It goes on to cite the example of America's failed initiative to enable 6 million Americans abroad to vote on the Internet in recent elections. The pilot program, which involved 100,000 expats, was abandoned (after $22 million had already been spent) because the software was deemed unreliable.
On the postive side, the publication contends various approaches "to prevent the occurrence of such disasters...have more similarities than differences. IBM tends to favour Java as its native programming language, while Microsoft prefers C#, a language it developed itself. However, both firms' platforms support other languages. Borland claims that, being neutral, it does a better job, but marketing seems to be as important as technology when it comes to winning market share."