The Software Crisis

 home -> software -> software_crisis
The complexity of software is growing exponentially. According to Remi H. Bourgonjon, director of SW technology at Philips Research Labs, Eindhoven, (cited in ref. (Gibbs, 1994)), the amount of code in most consumer products is doubling every two years. TV sets contained up to 500 kilobytes of SW in 1994, electric shavers two kilobytes and power trains in the (then, 1994) new GM cars ran 30,000 lines of computer code. This general trend also applies to larger systems.

According to a NIST Study NIST Study on the impact of software bugs, conducted in 2002, "Software bugs, or errors, are so prevalent and so detrimental that they cost the U.S. economy an estimated $59.5 billion annually, or about 0.6 percent of the gross domestic product. More than half of the costs are borne by software users, and the remainder by software developers/vendors."

In the cited article (Gibbs, 1994), the author wrote in 1994 that For every six new large-scale SW systems put into operation, two others are cancelled, the average software development project overshoots schedule by half, larger projects generally doing worse, and three quarters of all large systems are "operating failures" that either do not function as intended or are not used at all.

The new Denver Airport baggage handling system is a software project whose troubles were well publicised at the time (ref. Sci-Am94). On 21 miles of steel track 4000 independent telecars were to route and deliver luggage between counters, gates and claim areas for 20 different airlines. The system was to comprise 100 computers, 5000 cameras, 400 radio receivers and 56 bar code scanners. At delivery time, BAE Automated Systems´ bill was $193 million. Take-off was scheduled for Halloween 93, but due to software bugs in the system, operation of the airport had to be postponed. In June of 94 no prediction as to full operation was possible yet. At that time, daily losses due to capital and operation costs of the non-functional airport were $1.1 million. This is cited as a typical case, and several similar debacles are described in the same reference.

Within conventional computer science, several remedies are discussed: Algebraic methods of code development, better software methods better (human) management methods, and rigorous testing schedules. However, due to intensive human involvement in the process, progress on all of these fronts is very slow and costly (for instance, new software methods are said to take on average 18 years before becoming standard practise), calling for a radically new approach. There is an interview with Jaron Lanier, which shows more apects of this problem.

Last Update 2007-02-26 by <webmaster@organic-computing.org>