Sorry about the cliche, but, y2k is an unprecedented event. If you take even a well documented and large worldwide sample like Gartner Group does (15000 companies in 87 countries I believe these numbers are close), you still have a problem with analyzing the evidence. What does it really tell you.

Example:

If 100 people based on a survey have contracted HIV over a given period of time, you might conclude that from statistical analysis it will grow at a predictable rate. If you only find 10 people infected, can you say that the future future spread of the virus will be less than if you found 100 cases? Not at all. It depends on the sexual activity of the people, intravenous drug use, the population density where these people are, and a host of other factors. 5 people with HIV could conceivably do more damage that 100 if they are located in the right region of the country and are more sexually active that the 100.

This is where Gartner Groups numbers cannot tell you the level of severity of the y2k bug because it does not merely depend on the percent complete of a certain industry or government.

-- James Chancellor (publicworks1@bluebonnet.net), October 19, 1998

Answers

And remember, both the Gartner Group's and recent NERC reports are basically taking UNQUESTIONED what they are being told by the responders (with a large amount NOT responding to NERC). Again, one must balance these reports with common sense: Does it not seem odd that, at this very late date, that we do not seem to have a single significant "Y2K success story", in spite of some organizations working many years on this problem? If nothing else, the Gartner and NERC reports verify that Y2K compliance seems to be a very elusive animal.

-- Jack (jsprat@eld.net), October 19, 1998.

In other words you need an adequate statistical universe. 100 people are far too few for a disease analysis, 10,000 would give a much better figure. It will never be perfect, but it can come very close to actual numbers if the sample is unbiased and large enough.

The Gartner information may have some bias, as it reflects information from companies that Garner has had contact with in the past. That may not be a random representative sample in all cases, esp. in the case of the non-US data. I would trust their US data more than the non US data due to more info from the US. Still, it isn't a perfect sample. But it is large enough that we can say it should reflect a pretty good idea of what the conditions in the US are.

-- Paul Davis (davisp1953@yahoo.com), October 19, 1998.

Paul, everytime I read one of your posts, I just feel so PROUD to be an AMERICAN!!!

-- Jack (jsprat@eld.net), October 19, 1998.

Paul,

I agree with your analysis if you are simply trying to find the existing status of a particular attribute. However, if you are trying to predict the future current of events and their severity based on cascades, domino effects and interdependencies, the survey tells you much less.

-- James Chancellor (publicworks1@bluebonnet.net), October 19, 1998.

My general problem with the Gartner Group polls is that they are not unbiased in any way. They have a 12 page pamphlet they will sell companies about the horrors of Y2K. The cost? $750.00. Yes folks...a mere $62.50 per page. Or how about the fact that they have now started up a new branch of their company that will help you with your contigency planning.

I don't trust any poll conducted by a company with such obvious gains to be made from it.

Rick

-- Rick Tansun (ricktansun@hotmail.com), October 19, 1998.