322 CHAPTER 10 Simulation Techniques make ridiculous predictions. Consider the following modification to our previous example. EXAMPLE Consider a situation in which the processors are checked every T days, for some large T (say T = 300). Suppose we find, on the very first check, that all 10 processors have failed: this means that all 10 have had lifetimes less than T days. The likelihood function associated with this observation is L(µ) = F(T) 10 = 1 - e -µT 10 The value of µ that maximizes this function is µ = ; our estimate is thus that ^ the average processor lifetime is zero! What this means is that T was set so high that we were not able to obtain much information from checking after T days. 10.2.4 The Bayesian Approach to Parameter