How much will it cost to fix the year 2000 problem in the US? Some calculate $US55 billion; others say $US300 million. I say the debate is pointless.
By disguising and deferring costs, companies can declare and hit any year 2000 budget they estimate. What matters much more is having a consistent and defensible approach to avoiding the two-digit doomsday.
When I compared year 2000 costs - reported in recent US Securities and Exchange Commission filings - I found that they were unrelated to a firm's revenue or industry.
However, there was a straight-line pattern when year 2000 costs were plotted against the same company's information technology budget.
It revealed large variances in year 2000 costs for comparable retailers and banks:
Why would some firms report twice as much in year 2000 spending as the trend line?
How come some firms get away with spending up to 30 per cent less?
It's all a matter of technique and budgetary artistry.
The most plausible explanation is how a firm approaches its year 2000 problems.
There are nine basic ways of solving the problem, as defined by the IEEE standard for year 2000 terminology (see http://grouper.ieee.org/groups/2000/p2000-1.doc). Many of these methods are cheap to execute but unreliable and expensive to maintain after the year 2000 victory party.
Other techniques are expensive and hard to deliver because they call for altering business processes. They promise lower long-term maintenance costs but risk not completing the job in time.
The cost disparities can also be traced to the differences in how companies account for life-cycle costs.
The Federal Accounting Standards Board has ruled that year 2000 costs must be charged to current operating expenses rather than capitalised, which increases the reported expenditures.
Some corporations have found clever ways to reclassify year 2000 fixes as depreciable "business process improvements," which will lower what they will report as year 2000 costs.
At the other end of the budget legerdemain are plans that depend on now-you-see-it-now-you-don't techniques such as "year shifting."
That would involve replacing the offending dates with something workable, such as backdating everything to 1972. It's a quick and easily automated bargain-basement remedy; costs for it have been as low as 35 cents per line of code. But this approach disguises the ultimate price for restoring phony data and contrived applications to a condition where they can operate reliably for decades to come.
Another source of distortion comes from the reliance on the "line of code" metric for estimating year 2000 costs. Counting lines of code will always be misleading, regardless of the unit prices quoted by the popular IT advisory services.
Lines of code don't recognise that what really matters is only logical code, its complexity and whether that code is in microcode, assembly language, Cobol or Smalltalk.
Using average lines of code pricing is like using average dollars per pound to estimate the cost of an airplane or a motorcycle.
One thing is certain: Actual year 2000 costs will definitely not be what year 2000 program managers claim.
Rather than worry about the figures released to the public, corporate management, auditors, year 2000 project managers and the SEC should instead focus on documenting an intelligent and consistent approach to planning, scheduling and implementing the year 2000 readiness program.
Lawyers will use inconsistencies in preparation of year 2000 cost estimates as one proof of managerial neglect to provide adequate resources that would thwart year 2000 damages.
Cost estimation worksheets will be extracted from the project proposal documents during pretrial intrusions into the files of contractors, consultants and customers.
Only by showing consistency can management beat back these legal attacks. As an added precaution, managers will also have to demonstrate that their resource requirements have been independently validated.
Together, this will demonstrate that due diligence has taken place in coping with managing what are inherently flawed computer assets.
It isn't what you will spend on the year 2000 problem that has much meaning. What matters is that you have a consistent, thorough and defensible approach to solving it. It's the only way to gain a proper perspective on year 2000 budget estimates.
(Strassmann (firstname.lastname@example.org) finds that what we see as the reported year 2000 costs are only the "pay now" portion of a "pay later" bill.)