[dbcurtis]

At the risk of meandering very far off topic.....

When qualifying a new semiconductor process, it is common to do "schmoo" testing, that is to intentionally drive the process steps away from nominal temperature, vapor pressure, doping levels, process times, etc, etc, in order to characterize the expected yield with variations from nominal conditions.

When my wife was in General Mills' Betty Crocker division, it was common to characterize the behavior of a new cake mix formulation under varying conditions to account for inaccurate measurement of added liquids, variation in the size of eggs added by consumers, variation in home oven temperatures, variation in cook time, consumers adding milk instead of water, etc.

I found the parallels fascinating.