CHANGING THE ENGINE OF THE CAR - WHILE DRIVING 60 MILES AN HOUR

Most software development projects don't practice what is usually cons idered ''proper'' software engineering practices: well-documented, tra ceable requirements do not exist, formal inspections are non-existent, analysis and design models are incomplete or not even done at all, an d so forth. We know that many of these projects fail, and it is easy t o blame the failure of the project on the lack of good software engine ering practices (''If we only had well-documented requirements, a comp lete and thorough test plan, and formal schedules then we would have b een successful!'') But many projects that do not practice what many co nsider to be good software engineering practices succeed, and succeed famously. Examples include the startup introducing a revolutionary new product, the skunkworks project of an established firm that succeeds in changing the status quo, and selecting a mission critical applicati on as the first OO application of a group just beginning to make the t ransition to the new technology. Gaining a deeper understanding of suc cess and failure on software projects is absolutely essential for the continued maturation of our industry, and object technology in particu lar.