Until the 1990s many programs abbreviated four-digit years to to save memory, so a machine reading 98 as 1998 would read 00 as 1900. The predicted consequences ran from banking and utilities to elevators and medical equipment. Then the date arrived and almost nothing broke, which is now often told as proof the whole thing was invented. What came in between?
More than a year of work, at roughly $300 billion. After international alarm, feverish preparations and programming corrections, few major failures occurred in the transition from December 31, 1999 to January 1, 2000. An estimated $300 billion was spent upgrading computers and application programs to be , almost half of it in the United States. The quiet night followed the spending rather than replacing the need for it.
Until the 1990s many computer programs, especially the older ones, abbreviated four-digit years as two digits to save memory space. Such a system recognizes a two-digit 98 as 1998 and cannot recognize a two-digit 00 as the year 2000, perhaps reading it as 1900. The fear was that affected computers would be using an incorrect date from midnight at the start of 2000, and would fail to operate properly unless repaired first.
The exposure was wide. Banking, utilities and government records were expected to suffer most, with mainframes running insurance companies and banks thought most at risk, though networks of desktop computers were considered vulnerable too. Programs projecting budgets or debts forward could begin failing during 1999, and some software did not treat 2000 as a . There was concern about September 9, 1999 too, because early programmers often used a series of nines to mark the end of a program.
Nor was it only conventional software. Devices containing computer chips, from elevators to building temperature controls to medical equipment, were believed to be at risk, so these had to be checked for date sensitivity as well.
What followed was a remediation effort on a national and international scale. Business and government technology teams worked to check and fix systems before the end of December 1999. The United States passed an act in 1998 to encourage companies to share Y2K information. The European Commission warned that efforts in member countries were insufficient, and the UN convened its first international conference on the problem in December 1998. An estimated $300 billion was spent, almost half in the United States. Few major failures occurred on the night. The inference that the work prevented them cannot be proved, and it is the reading the sequence supports.
wat. Nothing broke in January 2000 because roughly three hundred billion dollars had already been spent making sure it would not.
wat wat
It is the clearest example of a problem whose successful management makes it look imaginary. A averted failure leaves no evidence except the invoices, which is a hard shape for public memory to hold.
How do we know?
Britannica’s account sets out the two-digit date problem, the systems thought to be at risk, the national and international remediation efforts, and the estimated cost, against the outcome on the night itself.
What is still unknown
The $300 billion figure is an estimate, and no account can prove what would have happened without the remediation. The reasoning that the work prevented the failures is an inference, however reasonable.
Learn something?