| The Millennium-Bug Muddle |
|---|
| From The Economist, October 4th - 10th 1997 |
In 1956 Howard Aiken, a computer pioneer from Harvard University, remarked, “If it should ever turn out that the basic logics of a machine designed for the numerical solution of differential equations coincide with the logics of a machine intended to make bills for a department store, I would regard this as the most amazing coincidence that I have ever encountered.” Don’t laugh. Mr Aiken was right: it was an amazing coincidence. The first generation of computer-programmers were also right to save precious space on their punch-cards by recording years as two digit-numbers: it freed costly storage capacity for better uses. Unfortunately, by another amazing chance, the result of that correct decision will be the most expensive industrial accident in history.
Unless they are reprogrammed, many computers will crash or start to blather incoherently on January 1st 2000, a date they will mistake for 1900. Specialist consulting firms, many as it happens selling advice about how to swat the “millennium bug”, are predicting a nightmare day when the clock literally is turned back—when banks, hospitals, power stations, factories, transport networks and governments suddenly lose the systems on which they have come to depend.
In fact, the cataclysm described in the consultants’ leaflets looks highly unlikely, and the scariest estimates of the cost of putting things right are wildly over the top. The scaremongers have served a purpose, though: because the world has taken fright, it is working to avert disaster. Over the next 800 days most, though not all, of the millennium bugs will be hunted down and exterminated. The eventual cost worldwide might come to $300 billion, rather than the ten-times-larger sums that have been mooted. But that is still a huge figure—and governments and companies may be interested to learn how such a mistake might be avoided in future.
One lesson is clear enough. As companies channel a good part of their planned spending on information technology into millennium-bug detection, the scale of that spending—on IT in general, that is—becomes apparent. Between now and 2000, companies will spend some $3 trillion on their computers. And as the millennium-bug drama also makes clear, many firms and agencies, public and private, now rely so heavily on these machines that they are not just hampered but entirely disabled when something goes wrong.
Despite the staggering cost, despite this near-total dependence, how many company managers know as much as they should about IT, or anything at all about it, in fact? Today’s more enlightened bosses may proudly display their keyboard skills, and some may even answer their own e-mail, but few can say with any confidence that the millions they have just spent on, say, client-server groupware was money well spent. Many are outright techno-illiterates, forced to depend on technology managers who typically know even less about the business than chief executives know about the boxes and wires that have invaded office and factory.
This is only now beginning to change, at least ten years too late. However, the millennium-bug scare teaches another, broader lesson—and that is not to expect too much even of technologically literate managers. Better-educated managers would have started to address this problem sooner and maybe more effectively, but they would not have prevented the bug from appearing in the first place. The point is that introducing the millennium error into programs seemed—and was—sensible at the time. This kind of “mistake” is intrinsic to many kinds of innovation, and it is as well to keep this in mind.
In many sorts of technology, the consequences of pragmatic decisions endure long after the circumstances that led to them—especially when an industry standard is involved. British railwaymen chose Stephenson’s standard gauge over Brunel’s wider one in the first half of the 19th century, as they did in America and most European countries. This standard, originally derived from horse-drawn wagons in British mines, has remained even as railway engineering has undergone 150 years of change. Not surprisingly, it is hopelessly inadequate. To exploit the power of modern high-speed locomotives, engineers resort to tilting trains and special track free of sharp bends: it’s cheaper than tearing up the existing rails and starting again.
Such decisions endure because basic technologies evolve slowly. Space missions are miracles of communications technology and precision engineering. However, 40 years after Russia launched the first Sputnik satellite, men and women are still being sent into space on top of a modified post-war ballistic missile. Cars are polluting, noisy and inefficient. Nearly a century after they first appeared, they are still powered by igniting a mixture of fuel and air.
Entirely new technologies are foundlings that come into the world unrecognised except by a fanatical coterie. After the invention of dynamos in the early 1880s, it took electricity 40 years to improve the productivity of factories. Scientists had to refine other technologies, such as generating plant and transmission; investors had to pay for new infrastructure; factory owners had to write off existing equipment; inventors had to find ways to exploit the technology, and entrepreneurs new markets for their inventions. And when a technology finally achieves prosperous middle age, inertia sets in.
All this makes technological change highly uncertain, to say the least. There are risks enough in the scientific underpinnings of technology. Intertwine these with the vagaries of competing ideas, with the accident-strewn adoption of a new technology and the inertia of the old, and Mr Aiken’s gloomy prophecy seems entirely forgivable. The record of technological optimists is no better. Predictions of robots, popular space travel and personal helicopters now seem as dated as Bakelite and brilliantine.
This is not a counsel of despair. True, the uncertain nature of technological change makes society liable to the ravages of millennium bugs and the irritations of QWERTY keyboards. Yet strange coincidences and happy accidents are part of the history of every successful technology. Every day companies have to guess where technology is going; many bankruptcies and some fortunes stem from those punts. The market guarantees a range of bets, most of which will fail. But that is not a waste, it is a process of discovery. Sometimes a “standard” will be chosen which later turns out to be “wrong” (two-digit dates, narrow-gauge railways), but market forces keep the waste in check, and eventually the standard becomes wrong enough to be replaced.
The correct response to the millennium bug is not dismay but humility. Millennium bugs, like other extraordinary occurrences, can only be predicted with hindsight.
| Please Panic Early | Index | Home Page |