Larry Towner is a y2k project manager and the creator of a widely disseminated bibliography on the Year 2000. (See "Introduction" category, 22 Jan 1997.) Here he offers his assessment of the IRS's problem: too much money to get the job done.
* * * * * * *
Date: Thu, 30 Oct 1997 14:09:03 -0600 From: Larry Towner To: firstname.lastname@example.org
Seeing the $850 million Y2k estimate that the new CIO of the IRS has come up with made me do a bit of arithmetic and the result is a bit amusing. If we assume that IRS will spend $350 million of that money replacing nearly every computer in their inventory and lots of other stuff with embedded chips, we are left with $500 million in labor costs.
I used $100k/yr as an average per person cost. The result is 5000 labor years to be expended between now and 2000. Assuming that, since the federal government fiscal year 2000 begins on October 1, 1999, there are two years left, that is an instant labor force of 2500 bodies.
1. No one could possibly ramp up that fast
2. There is a general shortage of experienced programmers
3. Managing that horde will, in itself, be improbable.
I can't see how IRS can spend $850 million (on Y2k) between now and the point where the bell tolls. Even if they divert ALL of their current staff, the odds against even partial success stagger the imagination.
We have reached the point in time where it is unlikely that most medium to large organizations can physically spend the money they estimate it will take to do the Y2k upgrade. Before 2000, that is.
Back in the 20s, Bell Telephone estimated that the rate of growth in the telephone system they were seeing would, by 1950, require every man, woman, and child in the US to be a telephone operator. The result was the dial telephone. Unfortunately, we don't have that kind of solution for Y2k. Current cost estimates truly will require the majority of the population of the US to become programmers in order to spend the money.
Just as unfortunate, this tossing of money (and bodies) at the problem is a personification of Brook's Law (the way to make a late project later is to add people). We have entered the stage of this problem where contingency planning and damage control should be one of the major parts of the assessment phase. But, so far, I don't see much evidence or much emphasis placed on those topics in our forum.
Usual disclaimers apply.