by John Rovito
Step back for a moment and take a walk in time. The Year is 1889 and the Paris World’s Fair is drawing record crowds. Not just with the Eiffel Tower and its homage to technology, but as a harbinger of one of the supreme transformations in the history of the world. In less than 25 years, an astonished public will witness the invention of the machine gun, the diesel engine, and synthetic fiber; the pneumatic tire, the steam turbine and the gramophone disc. Kodak will market the box camera; Rotegen will discover x-rays. Ford will mass-produce the Model-T; Marconi will invent the radio. The Wright Brothers will enact the first powered air flight. Einstein will formulate the Theory of Relativity. Catalytic change will become the benchmark for everything new and exciting. And the romance of technology will begin.
Today we are engaged in a love affair of equal passion. If the great metaphor for the machine age was the Eiffel Tower, then the universal symbol of the moment is the Computer with its underlying technologies of chip density and spreading bandwidth, processor speed and storage capacity. Each with trajectories plotted toward the vertical. All radically transforming both institutions and relationships with an optimistic paean to the Information Age.
But things have a way of biting back. Especially information technology which demands more, not less, of our time and energies. Computers may deliver greater efficiencies but often with an insidious aftershock as simple solutions conceal underlying problems that ultimately become far more difficult to treat. Think about the Year 2000 crisis and you quickly get the point. For decades, storage was scarce and expensive. So we shaved the date to cut the space. And for years it worked.
But by satisfying the moment, we mortgaged our future with an interest rate that blindly escalates toward century’s end. So when ’00 rolls around, computers will follow a logic that assumes 1900. Programs that store less than four digits for the year will suddenly be confronted with a number smaller than its predecessor. Computations based on value will silently change direction. Subtractions to discover time intervals will yield negative numbers.
The result will be algorithmic anarchy with everything at risk. And we’re not just talking about bank accounts or home equity loans. What we’re talking about is the switches in train tracks and instrumentation in hospitals. Control panels in airplanes and timing devices in elevators. The chips in the power grids and the locks on the cells. What we’re talking about is a potential crisis of mythical proportion and except for a few lonely Cassandras, no one seems to be listening.
A WORLD AT RISK
Most business systems are time sensitive, which means that an estimated 85% of all programs may be affected. Currently, there are an estimated 500 computer languages in use throughout the world with the cost of identifying and repairing the problem specific to each language. COBOL, which represents the largest installed base, is supported by a range of automated Year 2000 tools. Similar solutions exist for RPG, Fortran, PL/1, Focus and Natural. At present, the most difficult common language to repair is “C ” which is terse, allows registers, encourages the use of pointers to pointers, and permits very complex nesting constructs – all of which limit automated remedies to, at best, a 45% to 50% level of accuracy.
But that’s just the tip of the iceberg..
Because we live in an increasingly networked world, most systems exchange data with a variety of external entities such as state and federal agencies as well as trading partners using EDI to transfer invoices and purchase orders. If these organizations are not compliant, their data can act like a virus spreading faulty information to interfacing files and databases. As for hardware, the problem spans not only the mainframe and midrange worlds but also the servers, hubs and routers in networks, the BIOS in PCs, even the date-driven chips that comprise embedded systems.
An equally critical problem is third-party vendor packages. For some companies, these packages comprise the majority of the systems portfolio. Because most vendor contracts prevent the licensee from manipulating the code, an IT shop has no control over when and how the Year 2000 upgrades will be made. That doesn’t mean you can’t be proactive. Written communication is essential to ensure an audit trail in the event of future legal action. A determination must also be made of risks and contingencies, not only for switching to alternative packages but for testing all third-party software, even those packages that the vendors claim to be compliant.
A PROJECT MANAGEMENT NIGHTMARE
Changing two digit years to four digit years is not a complex technical challenge. Because the problem spans the entire enterprise, however, project management becomes critical. And there’s the rub. Recent studies by the Standish Group reveal that 84% of IT projects are either cancelled, significantly off schedule, come in with far less functionality that originally planned, or are grossly over budget, many by as much as 300%. When you’re dealing with a collapsing timeline, this tendency toward failure is a major cause for concern.
The best way to manage that concern is to have a documented, repeatable methodology to drive the project. While current Year 2000 methodologies vary by phase and terminology, almost all include four core steps. The first is to conduct a comprehensive Inventory of the entire portfolio. After all, you can’t know what you need to fix until you understand what you have. The Inventory should be detailed and documented and include operating systems and applications, databases and files, internal interfaces and external dependencies, and, of course, vendor packages.
After you’ve completed your Inventory, the next step is to carry out an Impact Analysis to determine the size and scope of your exposure. For applications software, this means assessing all critical date occurrences, either manually or through the use of automated tools. Once you’ve identified dates and density by application, you can then prioritize either the repair strategy based on a range of factors including business criticality and time to failure.
Having made this determination, you then need to move on to the Conversion of all affected code. Again, this can be done either manually or with automated tools. During this phase, the method of repair becomes critical. Companies that choose to physically expand the date field from two to four digits will need to build bridges that enable repaired applications to interface with those still waiting to be converted.
Because it can be done far more quickly and costs nearly 50% less than expansion, windowing is the selected repair method for most companies.
It’s important to note, however, that wherever dates are found in keys, the windowing technique is not applicable and the dates must be expanded, in some cases adding significantly to the time and cost of a project.
Whenever you convert code, you run the risk of missing critical dates, introducing new errors or altering the functionality. To ensure full systems integrity, the final step is to implement a comprehensive Testing program. Such a program must include date testing at the unit level to ensure no defects in each module of code, plus testing at the string level to ensure no defects in the interfacing of dates between modules of code. Verification of file and data conversions must also be performed, as well as regression testing to ensure that the existing system functionality has not been compromised. Integration testing to ensure that critical data can be passed successfully between systems must then be performed prior to implementing the code back into production.
A TRILLION DOLLAR PRICE TAG
What’s it all going to cost? Gartner Group estimates $600 billion worldwide to repair applications, networks, embedded systems, etc. While the cost of repairs will be large, perhaps the most alarming threat is the cost of litigation. Public corporations, including directors and officers, that have a fiduciary responsibility to act in the best interests of their shareholders are particularly at risk, especially if it can be proven that they failed to take appropriate action to address the problem.
Recently, the American Bar Association projected an additional $700 billion for litigation associated with the Year 2000. The four potential classes of litigation include: litigation filed by clients whose finances or investment have been damaged; litigation by shareholders of companies that suffer financially from not making the transition; litigation associated with collateral damage; and class action suits by purchasers of nonfunctioning computers and software.
But perhaps even more troublesome is our current state of readiness. Depending on which report you read, anywhere from forty to sixty percent of all U.S. companies have yet to begin addressing the problem. That’s a startling statistic but one dwarfed by surveys coming out of Europe and Asia where the century date conversion has barely registered on the spreadsheets of countries dealing with challenges ranging from monetary union to economic collapse. And then there’s the Federal Government. Aside from the Social Security and Small Business Administration, most Federal agencies appear to be in denial with over 75% of program officials not affording the Year 2000 a high priority.
THE “MINIMAX” SOLUTION
With a crisis looming, why the indifference? From biblical prophets to latter-day activists, someone has always been shouting jeremiads. Perhaps we’ve become too accustomed to the rhetoric, taken the fears and turned them into white noise. Whatever the logic, one thing is certain: this is not fiction. Depending on who’s doing the shouting, the century date conversion is either a trillion-dollar problem or the tip of a computerized iceberg with enough force to sink the entire business world.
What’s to be done? Software guru Capers Jones employed game theory to arrive at what is know as the “minimax” solution – the strategy that will give you the best return at the lowest cost. His conclusion: time is at a premium, so avoid further delays and start your Year 2000 initiative as soon as possible. Unfortunately, he arrived at this conclusion in 1996.
John Rovito is a director with Computer Generated Solutions, Inc., a systems integration and consulting company. He can be reached at 213/625-2655.
