Two top US benchmarking specialist, Howard Rubin of Rubin Systems and Capers Jones of Software Productivity Research, have been looking into the costs of Year 2000 conversion. Rubin, a Meta Group research fellow, surveyed 100 Fortune 500 companies, together comprising about 14% of the US gross domestic product (GDP) as part of a study sponsored by CAP Gemini. The overall picture was a lack of readiness, with isolated outbreaks of over- optimism on the part of respondents. Two thirds of companies surveyed by Rubin do not yet have detailed plans to address Y2K problems. From that data, he extrapolates that up to 45% of the US GDP may be left exposed to unresolved Y2K glitches at the turn of the century. Furthermore, while only a third have put plans in place, only 20% are actually in the process of executing their Y2K plans. Therefore, says Rubin, 55% of the GDP is not (or not yet) protected. Not surprisingly, 82% pronounced their Y2K spending inadequate to date. Very surprisingly, 87% expected that more than half their systems will be Y2K compliant by January 1999, the point when, according to conventional wisdom, code should be ready for testing. (Evidently, there are a lot of optimists still out there.) How much do they expect to spend? Although most companies have not put plans in place, according to Rubin, three quarters of them expect to spend between 21%-30% of their IT budgets on Y2K work through the turn of the century. About 60% of overall effort will be devoted to testing. Yet, according to Rubin’s findings, most companies are likely to short-cut the testing process, with only 60% of the companies planning some end to end testing (from unit through application, system, and end user acceptance testing). The responses indicated that the full test life cycle would likely be followed for less than 25% of all renovated code. Most of the effort (60%) will be directed at the mainframe, the platform most associated with non- compliant code (the next largest chunk is 20%, apportioned to networks). Naturally, there will be impacts to other IT activities outside Y2K. According to Meta Group, the sucking sound that you hear in IT projects will be the diversion of developers to Y2K renovations.
Capers Jones’ latest work, The Year 2000 Problem: Quantifying the Costs and Assessing the Consequences (published by Addison Wesley Longman in conjunction with the Association for Computing Machinery), takes an overall look at the problem. It estimates software inventory, comprising 36 million applications, 1.7 billion function points. To repair it, there is an available population of 1.9 million developers. He then took an aggregate estimate of the amount of code thought to be date-sensitive, which was 8%. Jones then estimated code inventory and date sensitivity by industry, combining it with productivity estimates which he noted were consistent with Rubin’s benchmarks. Overall, he estimated that about 8% of all function points are date sensitive; he believes that wholesale and retail are the most exposed, with date problems impacting about 11% of function points. At the other end of the scale, manufacturing was least impacted, with only about 5% of code thought to be date sensitive. Jones then factors productivity, date exposure, and pay-scales for each industry to estimate overall costs of over $70bn. Per function point, costs will average $534 across all sectors. So what are the grand totals for the damages? For the US Jones estimates $277bn, of which only $70bn account for initial software repairs (e.g., the amount budgeted for Y2K repair projects). What’s significant are the amounts for database repairs (an area for which there are very poor metrics) and litigation. Besides the US, Japan, Germany, France, the UK, and Brazil are likely to be the most affected, when it comes to gross numbers But there is another factor that could skew the equation at least against western Europe. The anticipated conversion to a common currency in the European community, which is planned for a similar time-frame cou
ld make a bad labor shortage catastrophic. Jones is on the mark in summarizing the situation: Unfortunately, the glamour of the standard currency blinded political leaders to the painful fact that Europe does not have enough software personnel to do both the currency conversion and year 2000 repairs at the same time. There is no doubt as to what Jones contends is higher priority. He states that while the impact of delayed currency conversion will only result in financial and political problems, Y2K delays could shut down power, telecom, air traffic control, potential injury or death due to medical instrument malfunctions and so on. In fact, nothing is so simple, because currency control and year 2000 compliance are inexorably intertwined. The recent falling domino effect of Asian currencies, initially dismissed as a regional problem, is now being taken seriously throughout all financial capitals worldwide. Therefore, choosing which software problem to solve first is becoming akin to choosing your poison.