Lane Core Analyzes FAA Figures, Mission-Critical Statistics, and Testing
Link:
y2ktimebomb.com.
Comment:
<< Lane Core uses the FAA as a model of how the U.S. government is handling y2k reporting.
Of the 72,000 systems, "mission critical" have been identified as about 6,400. The figure used to be 9,000. But if Pareto's 80-20 rule has any validity here, 20% or about 14,000 are mission critical.
The FAA's readiness statistic has moved from 99% last September to 65% on March 31. These figures do not make much sense except as self-reported public relations devices.
This is from Westergaard's site (April 5). >>
* * * * * * * * * * *
September 30, 1998: FAA Ninety-Nine Percent Complete. . . .
December 14, 1998: FAA Ninety-Five Percent Complete. . . .
March 31, 1999: FAA Sixty-Five Percent Complete. . . .
All these terms -- "renovation," "implementation," "compliance" -- are tossed around with abandon. So are the numbers -- 99%, 95%, 65% -- they're tossed around like candy for the children lining a parade route.
Garvey threw the 99% figure at a congressional committee on September 29, 1998. Do you think she might have done that just because the Office of Management and Budget had established September 30 as a "target date...for agencies to have all their mission-critical systems renovated"? (See the Federal Computer Week article Coming up short on Y2K, October 5, 1998.) Well, maybe she threw the 99% figure at the congressional committee just because the "target date" had been reached. Maybe not. It sounds good, though, doesn't it? "Ninety-nine percent of its computer systems renovated": yes, it does sound good.
Why The Figures Aren't Always As Good As They Sound
It doesn't sound as good, however, if you know that "renovation" is only part of the Y2K solution. The rest is testing and implementation. The CA 2000 White Paper, from California's Department of Information Technology, indicates in general terms that testing and implementation should take at least as long as renovation (including design, planning, development and modification).
As Frederick P. Brooks, Jr., remarks in his software-project management classic, The Mythical Man-Month:
In examining conventionally scheduled projects, I have found that few allowed one-half of the projected schedule for testing, but that most did indeed spend half of the actual schedule for that purpose. Many of these were on schedule until and except in system testing. That is, most projects don't allocate enough time for testing, most of them end up having spent at least half of the time in testing -- and many projects don't get behind schedule until they're in testing. . . .
Why might that be? Sometimes, because the requirements of sufficient testing might be more complicated, and therefore more time-consuming, than originally thought. Sadly, though, sufficient testing might uncover problems that hadn't been fixed, and it might even discover problems that were introduced during the software modifications. As Brooks says: The fundamental problem with program maintenance is that fixing a defect has a substantial (20-50 percent) chance of introducing another. So the whole process is two steps forward and one step back. (Some recent articles demonstrate the accuracy of Brooks' observation: Chicago Tribune, New Zealand InfoTech Weekly, and ComputerWorld.). . . .
Another Reason The Figures Are Even Less Good. . . .
Rounding the numbers, about 1 in 12 U.S. federal government systems (6,400 out of 72,000) are now deemed to be mission-critical, you see, but that number used to be much higher (9,000). Additionally, from Yourdon's observations and from Willemssen's testimony, you can see that the assertions in the three articles quoted at the top must actually refer only to FAA's mission-critical systems, not to all of FAA's systems. That's an important fact that cannot be gleaned from the news stories, and which may very well have been blown right past the reporters and, maybe, blown right past congressmen and everybody else, too.
Remember: when the government and the press tell us that the federal government has achieved 90% compliance by March 31, 1999, what they are really telling us is that only eight percent of the federal government's 72,000 systems have been fixed. (That doesn't mean, by the way, that 66,000 systems will still need to be fixed: by no means do all systems require Y2K fixing, and some can be replaced or "retired".) Moreover -- and not to be forgotten -- the state of the project and the rate of progress is self-reported, as Willemssen notes in passing in his FAA testimony. . . .
|
