On Sat, 5 Jun 2004, Michel Py wrote:

If your desktop support guys are half-organized, they have a replacement machine ready to install when the user calls for service, then the machine that as problem (which often is PICNIC: Problem In Chair Not In Computer) goes on the bench where the time it takes to reload a fresh Windows is 5 minutes, time to plug the cables in and pop the Ghost CD in.

This is where I agree with Paxson and Weaver, there isn't a deep supply line for replacement PCs. If 50 million PCs fall over in a single morning, its unlikely the PC repair people will have enough replacement machines ready instantly. Its the supermarket effect before a big snowstorm, they run out of milk and bread within a couple of hours. But then what happens? People starve? or is there product substitution? But I disagree with the repair costs and even the productivity impact. We have some experience with defective software, operator error and destructive maleware. My concern is over-estimates and under-estimates can lead to poor decisions. Unfortunately, we don't have a good way to get the proprietary data that does exist into a useful form for public decision making. The perfect storm scenario is interesting. But it needs to be validated against "business as usual" and cost of historical product recalls.