| | In the software development world, scope creep is notorious for driving costs up. Any project manager that isn't on top that doesn't have a chance of coming in on-budget and on schedule. Scope creep is usually driven by users who want more and more, after the project is under way. Or when they've received an early release and they come back with new ideas. If the client has the budget and the development team is willing, that can be treated as a great marketing opportunity. (I've helped customers realize needed scope creep - giving them a vision of what could be- in byte-sized chunks that took $300,000 projects to a million - and left them very happy). But more often is a source of customer annoyance - they don't want to pay for the extra costs, and delays, that are involved in what is often a rewrite, not just a little extra here or there.
Gold plating is also a problem. Having written code for many years I can tell you that there are times where one is in the Flow, just lost in that delightful bliss of watching your tweaks add better and better functionality or the same functionality with greater efficiency or cleaner looking, more elegant code. Writing code is an art form and some code is like a work of art to those able to "sight read" it. In this case it takes discipline and/or a good supervisor working with the code team to avoid Gold Plating damage to the bottom line and schedule.
Another peculiarity, closely related, was given the colorful name "polishing a turd" - that is where the application has already been released - its done, gone, shipped, but some programmers are still coding on this or that routine - getting it just right. I've seen that (and done it once or twice) even after a project is canceled, killed, dead and will never see the light of day.
Good programmers are idealistic and driven towards a sense of perfection. Good managers are tightly focused on schedule, budget, and checking off mile-stones of adequate functionality. Without both things you don't get good software in a business context.
|
|
