This is from Paul Strassmann (May 15). He offers it as good news. I see it as final proof that there is no way to avoid a y2k disaster.
Only 85 companies have submitted estimates that total $10 million or more. The other 4,000+ firms are either not saying or think they have small tasks to pay for.
The total they plan to spend is $9 billion. At $1 per line -- a cost so preposterously low that it is hardly worth discussing -- this means nine billion lines of code are presently budgeted for repair.
Let's consider some specifics. General Motors has
two billion. There is little doubt that Ford and Chrysler are comparable. AT&T has
500 million. Citicorp has
400 million. Chase Manhattan Bank has
So, it looks like there just isn't going to be anything like a fix of all of the code in the U.S.
And, even more frightening, most of the spending has yet to take place. In fact, under one-third of the money has been spent. But prices of programmers will rise in 1999. This means fewer lines of code will be corrected with the funds budgeted. Also, if a firm has spent a third of its money, it is nowhere near the final phase, testing, which absorbs 40% to 70% of any y2k repair project's resources, including (most important) time.
What this means is that large U.S. firms cannot possibly meet the form letter deadline for testing of December 31, 1998.
What this means is that y2k cannot possibly be fixed.
Meanwhile, the vast majority of U.S. large businesses have not even bothered to budget for the repair.
How Stassmann can put a happy face on this information is beyond me. He does for a few paragraphs, but then he backs off:
"The shattering revelation in the reported details is that only four companies so far claim they've progressed with complete systems testing. Until test results are independently validated and verified, we can't place much credence on easy prospects of fixing the year 2000 disease. The old advice 'Trust, but verify,' is more applicable to software than to any other human activity."
The web address is too long to fit in my Web software's box:
http://www.idg.com.au:8008/cwt1997.nsf/ 09e1552169f2a5dcca2564610027fd24/ 91dfbd0f62d6e3ca4a256603007e9394?OpenDocument
* * * * * * * * *
Until now, all year 2000 estimates came from conjecture, opinion polls, guesswork and questionable extrapolations by often self-anointed experts. We have, for the first time, information about year 2000 costs and progress from an authoritative source: first-quarter 1998 reports filed by US companies listed with the US Securities and Exchange Commission (SEC). The information was compiled by Triaxsys Research.
The reports conform to regulations issued last year. The data has greater credibility than anything published so far because it is likely to be used in any litigation against company officers and directors.
To understand what this wealth of year 2000 information means, relate it to the characteristics of companies that file reports with the SEC. The total revenue of the 4260 listed companies that must report is about $US6.6 trillion. For the first quarter of 1998 (covering financial data through 1997), the SEC identifies only 85 major companies with year 2000 budgets greater than $10 million. That includes the nation's largest companies.
Although the number of companies with reported year 2000 budgets is very small, together they account for 19 per cent of all corporate revenue and 24 per cent of all corporate profits. Clearly, the companies that are ready to talk about their year 2000 progress are the premier US organisations. Those organisations plan to spend a total of $9.4 billion to fix their date-change problems. Because they represent a mix of industries, I was able to come up with a cost estimate for the entire US corporate population on the basis of revenue. Such an estimate suggests that the total budget for fixing any year 2000 headaches for corporate America is only about $50 billion - a number much smaller than any prognostications so far, by a wide margin!
Now comes the interesting part. Although the 85 big companies plan to spend a total of $9.4 billion, they've spent only 29 per cent of that amount to date, which suggests that corporate America is ready to pay only about $36 billion for fixing the year 2000 situation in the time that remains. Can that be accomplished? I have data for IT budgets for 29 of the companies that reported their year 2000 budgets to the SEC. With high reliability, I can now predict the expected year 2000 spending if a company follows known patterns. That analysis suggests that in the time remaining, year 2000 spending will siphon off only 8.3 per cent of annual IT budgets. That calls for a material shift in the budgets toward software services, but is quite manageable despite moans and groans that failure is imminent because the job can't be completed. In fact, if this money is used prudently to correct the inherent flimsiness of the existing software, it may be a good investment.
Before we dismiss all the alarmists of the coming doom, note that I don't think we should depend entirely on the data that has been filed with the SEC. The shattering revelation in the reported details is that only four companies so far claim they've progressed with complete systems testing. Until test results are independently validated and verified, we can't place much credence on easy prospects of fixing the year 2000 disease. The old advice "Trust, but verify," is more applicable to software than to any other human activity.
My advice is to proceed immediately with testing a representative sample of critical systems. The experiences gained from such a pilot would confirm whether the estimates filed with the SEC are fiction or reality.