Erik: "if you look only at the "computational" behaviour, 1 hour (read: any time) flawless operation on 100°C won't guarantee the chip is still OK - it might have already developed weaknesses"

Lynn: "if you bake a chip for 1 hour at 100°C (or whichever the maximum temperature given by manufacturer is) as a screening for early failures, it shouldn't affect the reliability afterwards (assuming today's standard silicon processes)".

You are talking about two different things. Lynn's bias towards production testing is and understandable. Correct me if I am wrong.

JW