On the need for new reliability parameters

It is argued that we must decide which parameters are worthy of comparison not over a time span for one specimen, but among specimens derived from diferent times, different designs, and different configurations.