NZSM Online

Get TurboNote+ desktop sticky notes

Interclue makes your browsing smarter, faster, more informative

SciTech Daily Review

Webcentre Ltd: Web solutions, Smart software, Quality graphics

Feature

Millennium-Bug Mayhem

Is it just hype or the end of the world?

Craig Webster

By now, most people have heard some of the things many computers will do when the year 2000 arrives. The Millennium Bug or Y2K problem is caused by computers storing the year as two digits, such as 12/06/99 and presuming that all years start with "19". When the year ticks over to 2000 in only a few months from now, many computers receiving "00" as the year will start making mistakes, often acting as if time has jumped backwards to 1900 instead of advancing to 2000. All kinds of computerised date-based calculations and decisions could be scrambled.

What this means for our now heavily technologically based society is a matter of great debate. Unlikely though it may seem, the truth is that no one really knows for certain if the lights will go out, the international economy will collapse, or if planes will fall from the sky. Will hospitals, the police and the government be able to function in the early days of the new millennium?

If the millennium bug is so potentially devastating how could such an obvious mistake be allowed to happen? Historians of technology will find many lessons in the events of the next year, because it will see the conclusion of a saga which began over 100 years ago and which can tell us much about the nature of technological development itself.

The "bug" started in the US in 1890, the year of the first mechanised census. Completing the previous census, undertaken in 1880, had required 1,495 clerks working full time for seven years using tally sheets. It was clear that something had to be done to speed up the process because the US population was growing at such a rate that the time required to compile a census would soon exceed the time available before the next one had to be undertaken.

Herman Hollerith entered a competition to find a faster way in 1889, and was awarded the 1890 census contract. Hollerith's winning system involved the first true data-processing punchcards and tabulating machines which could rapidly read the holes in the cards. One punchcard stored all the details for one person, and years were abbreviated as two digits to save space on the card. Hollerith's system was a great success -- the census was completed in record time and considerably under budget.

Fifty-five years later, the first commercially available electronic computer, the UNIVAC, was introduced and this too ran on Hollerith cards and stored years as two digits. Not only was punchcard space at a premium, but electronic memory space in the first computers was also expensive, so storing the year as two digits made sense from a programming point of view as well.

In the early 1960s, a turning point presented itself. Robert Bemer, a computer scientist, urged switching to a universal four-digit date format because of future Y2K problems. He was ridiculed and ignored -- the new millennium seemed so far away and, at a time of great technological optimism and advancement, no one believed that the same computer programs would be in use so far into the future.

A few years later, the hugely successful IBM System/360 was released which set in stone the two-digit year format for decades to come. It took IBM until 1995 to officially acknowledge the Y2K problem, but by then market giants like Microsoft were busy perpetuating the two digit year format in their DOS software for the personal computer market on a scale even IBM had not managed. Microsoft finally acknowledged Y2K problems in its Windows software in 1997. Despite this, components of Microsoft's recent Windows98 operating system are not compliant with a four-digit year format.

What's Coming?

The nature of impending computer snafus are likely to be similar to ones which have already happened. Most problems may not appear until 2000, but because many computers make decisions based on future dates, there have already been a series of mistakes and malfunctions because of the Millennium Bug. To the alarm of banks and other financial institutions, many automated backup systems have shown a disastrous propensity to discard tapes with expiry dates after 2000. When a new backup is made and dated with an expiry year of "00", many systems read this date as "1900" and immediately discard the tape as if it were 99 years old. A bank which lost all records of its customer accounts would be out of business overnight.

In the US, prison inmates have been released years early after having been issued with automatically generated release papers because their true release date fell after 2000. And in Europe, new goods for a continuous process factory were automatically rejected as expired, causing the entire plant to be shut down for lack of raw materials and costing millions of dollars in lost production time.

The problem doesn't just affect large computer systems either. Many smaller devices contain cheap computer chips which keep track of the time and date -- any of these "embedded" chips could start acting oddly if they store the year as two digits.

Are high-risk large-scale systems likely to do serious damage come 2000? In 1993, the US defence headquarters at NORAD rolled the clocks on their computers forward to 2000 to see what would happen to the system which monitors US airspace for incoming nuclear missiles. Among other problems, the alert system crashed.

British nuclear power regulators are also concerned about what the Millennium Bug will mean for nuclear power safety. The Health and Safety Executive reported last year that the bug has been found in four alarm and monitoring systems in British nuclear plants and that these could lead to serious operational problems.

Overall, the Y2K problem is likely to be most damaging in Third World countries or in countries that are strapped for cash, such as Russia. Although these countries generally have a less developed technological infrastructure, they often have a great deal of ageing Western computer equipment in operation in critical roles but without the money or the awareness of the problem to do anything effective about it.

The Gartner Group, a US technology consultancy, claims that the single biggest Y2K risk comes from embedded computer chips which are present in a huge variety of devices including traffic lights, sewage systems, cars and medical equipment. In many of these devices a faulty date will not matter -- most VCRs usually blink "0:00" most of the time anyway. However, the small percentage of devices which will behave differently and perhaps dangerously with a bad date are widespread enough to be a significant problem.

What's Being Done?

Industry and business have been bug fixing for a number of years. Some insurance companies had to start as early as 1988 when they realised their 7-year bonds would begin crashing their computer system from 1993 onwards. But the trouble is that there are too many occurrences of the bug out there and too little time before 2000 to fix them all.

Many Y2K problems are so well hidden inside millions of lines of computer code or in non-compliant computer chips that even identifying their existence is difficult. This has lead to a triage-based approach, which involves concentrating efforts on bug-proofing the most essential and potentially dangerous services. Such an approach inevitably means that many bugs will survive to strike at midnight on the last day of the year. If the triage approach has worked, the remaining bugs that do strike will cause only minor problems.

Realistic estimates of the worldwide cost of fixing the Millennium Bug are as high as US$600 billion. In addition, many companies are upgrading their computer systems years early to make sure they are compliant. The amount that will have been spent on upgrading before 2000 has been estimated at US$3 trillion -- this pre-emptive expenditure may slow the growth of this sector of the economy for years after 2000. Many companies that are well into their bug fix now realise that if they had done nothing they would have been in serious trouble come 2000. Most planned bug fixes are progressing well and the costs so far seem to be about 5% to 25% of a company's information technology budget.

In New Zealand, Coopers and Lybrand have estimated the total national cost of 2000 compliance at $450 million. Almost all large New Zealand companies report having the problem in hand, although only 76% of small- and medium-sized New Zealand businesses are confident they can deal with it.

New Zealand banks are now issuing statements that their computer systems are compliant and that your money is safe in your account. They are also issuing credit cards with expiry dates after 2000, a sign that they are confident their systems are bug-free. New Zealand banks, like those overseas, have been running computer simulations of their banking systems with the clocks rolled forward for up to two years, completing millions of simulated post-millennium transactions.

Despite this, 53% of people asked in a Time/CNN poll conducted this year say that they expect banking systems to be disrupted by the bug and 47% said they will be taking extra money out of their bank accounts before 2000.

In hospitals, the most pressing compliance problems are whether vital life-giving equipment will keep working into the New Year. This is of particular concern in accident and emergency departments, which are guaranteed to be stretched to the limit by casualties of the once-in-a-lifetime millennium celebrations.

Modern hospitals are full of computerised equipment, such as devices which give the patient the right dose of a drug over time, monitors which alarm when the patient is in trouble, and vital emergency equipment such as defibrillators which restart a patient's heart. Most of these devices contain chips which track the date and are designed to behave in a time-sensitive way.

Hospital electronics departments have been busy over the last year or two testing everything which switches on or beeps. Often the most straightforward way of dealing with non-compliant equipment is to replace the entire device, since replacing only the embedded chips is often so time-consuming as to not be cost-effective. The hospital where I work seems well prepared -- green-tick compliancy stickers are everywhere.

Lessons

Like the Chinese blessing/curse, we indeed live in interesting times. The Millennium Bug will never happen again in our lifetimes (although others might) and so it is a genuinely fascinating event, much rarer than a total eclipse of the Sun or the passing of some comets. It won't be the end of the world, but it will likely cause numerous hiccups for reasons that the bug fixers have just not thought of.

Remember that the Auckland power blackout happened completely unexpectedly even in the absence of Y2K problems and so it is impossible to say exactly what will happen. Probably people somewhere will die because of it, but if you consider the bizarre and unlikely circumstances under which many people in the world die every day this is not such a surprising prediction.

Hopefully, we will learn from it and build more robust technology from here on. We certainly know much about how to do this -- the first step is to develop standards based on thought and foresight rather than the market production pressures which have led to so many de facto technological standards in the past. If the Millennium Bug were fixed in the 1950s or 1960s, when concern about it was first raised, it would have cost a tiny fraction of the price that is now being paid all over the world. It is an unprecendently large-scale example of our computers as error amplifiers.

Despite frequent "face lifts", most of our personal computers and banking systems still run 30-year-old COBOL code or DOS at their cores. The lesson is that technology has a surprisingly long functional half-life, and development happens not by scrapping the old and wheeling in the new, but by adding new layers on top of old layers like tree rings.

For example, if you believe that no-one uses punchcards anymore, that ancient source of the Millennium Bug, you would be very wrong. Although the US consumption of punchcards peaked in 1967, at 200 billion per year, they are still being used in 1999 at a rate of 60 million cards a year. Among the users are Ford Motor Company and departments in the US government who use punchcard technology in conjunction with their Pentium PCs.

Our technologies are now too complex and invasive not to make their safety a science.

Precautionary Principles

I am not planning to fly in the first few days of the new millennium. Taking some money out of the cash machine for New Year before midnight and trying to avoid emergency hospital care are also probably good ideas.

Watch out for the 9/9/99 bug in September. Four 9s has long been used by programmers to indicate "end of job" or "unlimited expiry date" and, while the arrival of this date isn’t expected to cause as many problems as Y2K, it should provide an interesting prelude to what may be to come.

The Millennium Bug is one of the best opportunities for Macintosh users to ridicule PC users -- Macs have been 2000-compliant since the mid-1980s. The best way to check if your PC is compliant is to back up all your files, roll the clock forward, and see if things keep working. Even if the version of your operating system is compliant this is no guarantee that individual applications will not start acting strangely (see your supplier).

Apart from that, enjoy the ride -- it might be something worth telling your grandkids about.

Craig Webster is currently a clinical researcher in the Anaesthetics Department at Auckland's Green Lane Hospital.