Probabilities?

Interesting and concisely informative article. While I have great respect for the Gartner Group, I’m quite curious about how they developed and validated the probabilities for each stage of the Win2K development cycle.

Surely it can’t be a regression-extrapolation. A simple linear regression based on Microsoft’s prior schedule performance couldn’t possibly be that high. These guys are smart, savvy analysts with an excellent data base to work from. So how did they come up with these numbers?