|
Robert Blumen begins with a quotation from the WASHINGTON POST (June 3), which summarizes y2k Czar's Koskinen's response to Congressman Horn's grade of "F" to the U.S. government.
"The administration's point man on the date problem, John A. Koskinen, disputed Horn's estimate that 40 percent of critical systems won't be fixed on time, saying, 'The process is under control and agencies will be accelerating over the next several months their rate of progress.'"
Blumen then adds this:
* * * * * * *
This is really a lot of baloney. This contradicts my experience in every software project that I have ever worked for, and a large body of empirical research about software projects.
There is a well-known expression among programmers and software project managers termed "the 90% syndrome", meaning "projects spend 90% of their time in a state of being 90% complete." What this refers to is that many projects that are SEEMINGLY nearly done spend a long time being nearly done, but unable to complete. It seems like the project is nearly done because all the code is written, it just doesn't integrate properly and every time you fix a bug, another one appears.
The reason for this is that it is very easy to fool yourself about how much progress you have made, thinking that you have made more than you did. Until you integrate and test the software, you don't know how far along you are or aren't, usually less than you think. So anyone who says that the success of their project depends on accelerating the rate of progress is probably overly optimistic. Ed Yourdon says this quite well (and this corresponds to my experience in over eight years in this field) in one of his email newsletters: Nobody really knows whether the project is REALLY proceeding well, or whether it has been in deep trouble all along, until the system testing and integration begins. That's when the rubber meets the road that's when all of the pieces begin to come together -- or not.
|