This is an archived page and is no longer updated.
Please visit our current pages at https://rvs-bi.de

The Year 2000 Problem / Das Jahr-2000-Problem

The Year 2000 Problem
Das Jahr-2000-Problem

Main essays/Hauptthemen:

For centuries, dates have been represented in shorthand in human endeavors. In particular, two decimal digits have often been used to represent the year in writing since the Renaissance. But in the latter half of the twentieth century, this habit has given rise to the single most costly engineering fault in world history. Estimates of the cost world-wide range from $300-600 billion (Gartner Group) for prophylactic operations to $3,6 trillion for both prophylactic operations and post-millennium recovery (software consultant Capers Jones).

There are about to be failures of complex systems due to date representation discontinuities. On 21 August 1999 at 23:59:47, the GPS time data will 'roll over' to zero; GPS systems without protection may think they are back at 5 January 1980 ('week 0'). On 1 Jan 2000, many software and hardware system components will 'roll over' to 1.1.1900 (the so-called 'millennium bug'). On 19 January 2038, current 32-bit Unix systems will 'roll over' to 1 January 1970. While the Unix and GPS problems are bounded in scope, it seems that millennium problems are not. In particular, there don't appear to be enough human resources available to investigate all millennium dependencies in existing systems. We can conclude that problems will occur and that most work will happen after the problems have shown themselves. Formal methods for analysis and 'debugging' of such problems will be needed, and we are adapting WBA (see The Why-Because Analysis Home Page) to this task.

Many people, including some reknowned computer scientists, believe that the collection of Year 2000 problems are scientifically trivial. We believe strongly that this is not the case. There are surprisingly few conditions required for adequate data representation. Each one of these five conditions is violated by some instance of a Year 2000 problem. We describe these conditions, and what happens when they are violated, in the article The Analysis of Data Discontinuities.

With thanks to Martyn Thomas, we have also made available some essays of his,

(© Martyn Thomas 1997, 1998, reproduced by permission).

Finally, some individual work by members of the group: