To: Hawkmoon who wrote (2052 ) 6/27/1998 2:45:00 PM From: John Mansfield Respond to of 9818
'Mayday, Mayday <(FAA)> When a system becomes so complex that no one manager can comprehend the entirety, traditional development processes break down. The Federal Aviation Administration (FAA) has faced this problem throughout its decade-old attempt to replace the nation's increasingly obsolete air-traffic control system [see "Aging Airways," by Gary Stix; SCIENTIFIC AMERICAN, May]. The replacement, called the Advanced Automation System (AAS), combines all the challenges of computing in the 1990s. A program that is more than a million lines in size is destributed across hundreds of computers and embedded into new and sophisticated hardware, all of which must respond around the clock to unpredictable real-time events. Even a small glitch potentially threatens public safety. To realize its technological dream, the FAA chose IBM's Federal Systems Company, a well-respected leader in software development that has since been purchased by Loral. FAA managers expected (but did not demand) that IBM would use state-of-the-art techniques to estimate the cost and length of the project. They assumed that IBM would screen the requirements and design drawn up for the system in order to catch mistakes early, when they can be fixed in hours rather than days. And the FAA conservatively expected to pay about $500 per line of computer code, five times the industry average for well-managed development processes. According to a report on the AAS project released in May the the Center for Naval Analysis, IBM's "cost estimation and development process tracking used inappropriate data, were performed inconsistently and were routinely ignored" by project managers. As a result, the FAA has been paying $700 to $900 per line for the AAS software. One reason for the exorbitant price is that "on average every line of code developed needs to be rewritten once," bemoaned an internal FAA report. Alarmed by skyrocketing costs and tests that showed the half-completed system to be unreliable, FAA administrator David R. Hinson decided in June to cancel two of the four major parts of the AAS and to scale back a third. The $144 million spent on these failed programs is but a drop next to the $1.4 billion invested in the fourth and central piece: new workstation software for air-traffic controllers. That project is also spiralling down the drain. Now running about five years late and more than $1 billion over budget, the bug-infested program is being scoured by software experts at Carnegie Mellon and the Massachusetts Institute of Technology to determine whether it can be salvaged or must be canceled outright. The reviewers are scheduled to make their report in September. Disaster will become an increasingly common and disruptive part of software development unless programming takes on more of the characteristics of an engineering discipline rooted firmly in science and mathematics. Fortunately, that trend has already begun. Over the past decade industry leaders have made significant progress toward understanding how to measure, consistently and quantitatively, the chaos of their development processes, the density of errors in their products and the stagnation of their programmers' productivity. Researchers are already taking the next step: finding practical, repeatable solutions to these problems. [Continued in Part 4] ------------------------------- Richard Church webmaster@lucidimages.com It's the year 2000. Do you know where your government is?lucidimages.com -----== Posted via Deja News, The Leader in Internet Discussion ==-----dejanews.com Now offering spam-free web-based newsreading ___ 'Subject: Software's Chronic Crisis - Part 3 Date: Fri, 26 Jun 1998 04:46:47 GMT From: webmaster@lucidimages.com Organization: Deja News - The Leader in Internet Discussion Newsgroups: comp.software.year-2000