|
|
 |
 |
 |
 |
   |
COMMENTARY: THEN THERE'S THE COST OF FIXING THE FIXES...If you think the Year 2000 problem will be mostly solved when the millennium begins, think again. A cardinal rule of software development states that new bugs emerge whenever programmers touch software code. So get set for Y2K: The Sequel. The aftermath--the cost to fix Y2K fixes--could add as much as 40% to the billions spent preparing for 2000--or more if companies start testing Y2K remedies too late, injecting even more errors, according to Capers Jones, chief scientist at software services company Artemis Management Systems. Businesses and consumers should brace for unexpected costs and disruptions spanning several years, warns Daniel D. Galorath of Galorath Inc., which sells programs that estimate software-development costs. ''There's been a naive presumption, particularly among key economists, that a certain amount of money will be spent and then it will all be over,'' says Edward Yourdon, author of Time Bomb 2000: What the Year 2000 Computer Crisis Means to You. ''It won't be.'' In case anybody had missed it, the Y2K problem proves once again how dependent on software we are--and how critical it is to get smarter about using the latest tools for fixing and testing bugs and for managing software projects. Historical data show that when companies try to fix programs, some 15% of newly introduced bugs aren't detected before release, according to Artemis' Jones. And when bugs are fixed, 7% of the repairs are faulty, with nearly half the new bugs capable of crippling an application or causing major errors. As the millennium nears, companies may be tempted to throw more programmers at the Y2K problem. Even now, about 90% of U.S. companies are behind Y2K targets by two to four months, according to Howard A. Rubin, chairman of the computer science department at Hunter College in New York. But Brooks's Law, a rule of thumb coined by Frederick P. Brooks Jr., the author of The Mythical Man-Month, says that adding programmers to a late software project only adds delays. Most companies consistently underestimate the time and cost of software development projects. But the Y2K problem ought to teach them to make smarter predictions. Defense contractors, aerospace companies, and about 12% of the most tech-savvy corporations now can make accurate forecasts of the cost and time for software development projects using new techniques called parametric modeling. It's time for more companies to adopt those methods. Both customers and vendors of software need to become more realistic about problems in the offing. That, at least, would be a positive aftermath to the Year 2000 problem.
By Steven V. Brull
|
Updated Dec. 3, 1998 by bwwebmaster
Copyright 1998, by The McGraw-Hill Companies Inc. All rights reserved.
Terms of Use