Personal tools
News from ICTP 87 - Features - Millennium Bug
As we cross into the next millennium, many of the world's computers may confuse the year 2000 with the year 1900, causing untold problems for our networked world. The Centre will likely avoid the anguish that this electronic bug may create for others. Here's why.
Bugged Out
... But not at the ICTP
There will be no millennium bug at the ICTP. Through a combination of good luck and wise investments, the Centre's computer systems and databases should be able to welcome the new millennium without experiencing any debilitating ailments.
Other computer-dependent institutions may not be so fortunate. Many likely will spend hundreds of thousands--even millions--of dollars to avoid problems ranging from malfunctioning automatic bank tellers to failed missile systems.
"The roots of the millennium bug," notes Alvise Nobile, Head of ICTP's Scientific Computer Section, "date back to the mid 1950s, the dawn of the computer age."
"At the time, computer memory was very expensive, and early programmers continually searched for ways to avoid consuming it. That's why they designated two digits instead of four to electronically record each year. Although memory became much cheaper in the 1970s, by then two-digit dating had become an industry convention, which both manufacturers and programmers continued to follow."
When the computer industry finally realized that much of the hardware and software it had created would fail to recognize the year 2000, confusing it with the year 1900, it was too late: millions of chips and billions of codes had already been embedded with this potentially troublesome flaw.
For computer-dependent private companies and public institutions alike, the millennium bug creates more than a mild headache. It could, in fact, strike damaging body blows to their core responsibilities and, as a consequence, generate colossal social disruptions. Computerized street lights, for example, may fail to turn on. Life insurance policies may be cancelled. Student grades and graduation records erased. Yearly pensions invalidated. Computerized defense systems may even be disabled.
Imagine if you woke up one day and thought it was 1900 instead of 2000, and then behaved as if you had stepped 100 years back in time. In effect, a lifetime of memories would vanish. That's exactly what could happen to many of the world's individual computers and inter-networked computer systems. No wonder experts are so alarmed.
Yet, how is it that the Centre has been able to inoculate itself from the millennium bug? As Nobile explains, "Computerized information about the Centre's visitors--for example, how many times they have come to Trieste and how long they have stayed--will have to be reprogrammed to account for the new millennium. Otherwise, this important data may be lost. Beyond that, however, the millennium bug poses minimal risks to our computer and data systems."
"The Centre's good fortune," says Nobile, "has a lot to do with the kind of research that is conducted here. Because much of our work in theoretical physics and mathematics deals with timeless truths and immutable physical laws, Centre researchers have had little need to systematically date their data as an integral part of their studies."
The one exception is ICTP's Physics of Weather and Climate Group, where baseline information is essential to determine, for instance, the buildup of carbon dioxide in the atmosphere or potential shifts in precipitation patterns due to alternations in cloud formations and wind currents.
Nevertheless, as Nobile notes, "The Weather and Climate Group is a relatively new group at the Centre and receives most of its data from other places like the National Center for Environmental Predictions and the Center for Ocean Land Atmosphere Studies in the United States. As a result, the millennium bug is not as much a threat to this group as it might have been under other circumstances."
The truth is that, across the board, coded dates have not been an integral part of the Centre's research software programs, and since they aren't embedded into the databases, the millennium bug has nothing to attack.
Another factor that has helped the Centre ward off the millennium bug is that the ICTP did not hitch on to the computer bandwagon until relatively late in the game. Abdus Salam, who founded the Centre and then served as its director for the first 30 years of its existence, initially doubted that computers would become an essential tool for theoretical physicists and mathematicians, especially those in Third World countries. In fact, the Centre's earliest machines were computer hand-me-downs shipped to Trieste in the late 1970s by ICTP's main administrative body, the International Atomic Energy Agency (IAEA) in Vienna, Austria, or by the European Laboratory for Particle Physics (CERN) in Geneva, Switzerland.
"Virtually all of the Centre's current equipment," Nobile notes, "was purchased after the computer industry acknowledged the millennium bug problem and had replaced two-digit calendar chips with four-digit chips."
In an ironic twist, the Centre's experience with the rapidly changing world of computers carries lessons comparable to those drawn from the fabled race between the turtle and the hare. Starting slow but remaining steadily on course once it embraced the computer age has allowed the ICTP to approach the year-2000 finish line--or should we say, precipice--in relatively good shape.
Nobile, however, urges the ICTP not to rest on its good fortune. "The Centre's servers linking each of our computers to other computers and networks both on-campus and throughout the world," he observes, "operate on the UNIX system. The experts who developed the UNIX may have avoided the two-digit dating problems that now threaten many of the world's computers and networks. But they did not eliminate all of the bugs."
"For example, as a matter of convention, most UNIX systems begin counting time as of 1 January 1970. An internal software program then translates the UNIX 1970 false counting system into real-time. The problem is that the system can count only 2 billion seconds and no more. As a result, with UNIX's world beginning on 1 January 1970 and clicking away ever since, the system's capacity will be reached sometime before 2030. When the computerized counter sweeps past that moment--much like the problems caused by the millennium bug--the system will fail unless its capacity is expanded or the counters are modified."
Nobile urges UNIX users not to worry. "Thirty years in computer time is like a millennium in real time. Just like it would have been useless to try to improve stagecoach transportation at the turn of the nineteenth century (why improve the efficiency of horse-drawn carriages when combustion-engine automobiles loomed on the horizon), it's likely that the UNIX system will be relegated to the dustbin of history well before the system's counting problem will have to be confronted."
And, so for now, Centre scientists and staff can rest easy. The turn-of-the century millennium bug won't be inflicting much damage here.