Brumby
Major
I think dingyibvs and latenlazy were making a caveat that a simple mean failure rate count isn't always accurate depending on whether the distribution of failures was a result of something like improving the system and/or incorrect initial testing conditions.
Reading over some of your replies on the last page I'm not sure if you missed the rationale of what they were suggesting?
For the sake of bringing closure to this point over mean rate and let's just say as you pointed out, the mean rate is not accurate, how does it change the conversation? The mean rate in the report is just a means to an end, and the end point is that reliability is not where it needs to be. So even if the mean rate is not accurate does it mean reliability has suddenly gone up or conversely it is worst, with the latter being exactly what the report is emphasizing i.e. reliability needs to improve. So what is the point of arguing over whether the mean rate is correctly determined or not?