|
(feel free to mail this page) (Links to documents appear after the summary.) Here is the problem in a nutshell. 1. There is no agreed-upon standard for the change: 20000101 or 01012000. Also, does 01 refer to January (American version) or the first day of the month (British version)? All computers must agree on this after the fix. They will not agree. 2. There is no coordinating agency with the power to impose sanctions for failing to comply. The market will impose the sanctions. The market is merciless. 3. A noncompliant computer will infect a compliant computer with bad data, rendering the compliant computer noncompliant. Why bother to pay for making the repairs if everyone else doesn't? It's Catch-22. 4. Locking out a noncompliant computer keeps that computer out of the system or network. How many must be locked out to destroy the system? Now, think one word: "banking." It boils down to this sentence: "If you can't fix every computer in a system, then you cannot be sure of the system." It's not good enough to make your computer individually compliant. You must make every computer that it interacts with systems-compliant. But you can't do this. No one can do this. Except God. And He's amused. "And thou say in thine heart, My power and the might of mine hand hath gotten me this wealth" (Deut. 8:17). All confident talk about a solution to the year 2000 problem must demonstrate, step by step, system by system, how this operational virus can be avoided: technically, managerially, and financially. It's coming down. The system is coming down. By "system," I mean the West. We can date it. Never before in man's history has anything this big been predictable with anything like this precision. What will amaze historians of the future is that anything this huge could have been ignored for so long and then, when discovered, been denied so universally. It's Cassandra's curse, multiplied across an entire civilization. I know of only one work of fiction that describes anything like this: Isaac Asimov's short story, "Nightfall." I'm not an expert on literature, however. Still, you might read "Nightfall" to get a vision of what has been programmed into the system. Are you really at risk? Here's a test. Send out a letter to any company or organization that you rely on: public utility, mutual fund, bank, railroad company, airline, or whatever. Ask if the organization is Year-2000 compliant? (It won't be.) Then add this sentence: "If you are not Year 2000-compliant, please answer the following three questions: (1) How many lines of code are in your system? (2) How many programmers are presently working full-time on repairing it? (3) When did they begin the repairs?" The fact is, the outfit is probably not to the assessment stage. It's probably not even to the inventory stage. Actually fixing code? Highly unlikely. Here is what will happen. 1. You will not get a written reply. This is the standard response. 2. A manager (Dilbert's boss's boss) will send you a letter saying that the company has been working on this, and plans to be compliant on December 31, 1998. (This is the standard date, since it leaves a year to run tests.) He will not give you the specifics you need to judge if the outfit has a prayer of getting compliant: number of lines of code, etc. 3. A manager will tell you his organization is already compliant. His organization uses only PC's. Unfortunately, every supplier it depends on is still non-compliant. What happens if he tells you that his company isn't going to make it? He gets fired. What happens if Dilbert tells his boss that the programming team won't make it? He gets fired. Who has any incentive to tell the truth? None. That's why every outfit gives you the late 1998 date: a full year for testing! (Right! On what? Where does the world get the spare mainframe capacity in 1999 to run all those tests? And if there is spare capacity, then hardly anyone got as far as testing.) (Other categories: "Too Late?", "Noncompliant Chips," and "Domino Effect.")
|