As I read it, his main response is that making the numbers available for any engine or motor problems would be unfair to the larger producers since statistically with more output and sales they are likely to have more failures. That may to the casual observer appear to mean the larger producers have more problems. Of course being scientifically minded most of the time, we typical rocketeers wouldn't be interested in raw numbers. More likely we would be interested in a failure rate or ratio. Their is a problem, of course, in just stating a number. How accurate is that number and what data was used to create it?

Let's just say that the current to date run for the B20-8 is 2000 released for sale. Additionally we find that each batch is of 500 engines and that they produce two batches per month. Let's also say there have been 4 MESS reports and of those four, two came from the same batch number janB, one came from the JanA, and one came from Feb A.

So how do we report our findings?

Using just the raw numbers we get:

Overall ratio 1 in 500

Jan only statistics 1 in 333

Feb only statistics 1 in 1000

Batch Jan A statistics 1 in 500

Batch Jan B statistics 1 in 250

Batch Feb A statistics 1 in 500

Batch Feb B statistics 0 in 500

Looking at those numbers alone, we might think, "boy I might not want to use a Jan engine especially batch B from that month. The thing is, we are not looking at a complete picture here. As most of us are aware, the entire run of production is not used to launch. The major contributing factor to a reliability statistic is not being collected, and that is an accurate number of

*successful launches*. Without that number, our results are skewed toward a more negative result. Imagine if you will that we get the information required on successful launches of the B20-8 and find out that 100 of the Jan A batch were launched, 400 of the Jan B were launched 350 of the Feb A were launched and 50 of the Feb B were launched. Our numbers would change significantly:

Overall ratio 1 in 175

Jan only statistics 1 in 167

Feb only statistics 1 in 200

Batch Jan A statistics 1 in 100

Batch Jan B statistics 1 in 200

Batch Feb A statistics 1 in 150

Batch Feb B statistics 0 in 50

Now what looked like the batch to stay away from, is possibly the safest with one batch not having enough data to make a statistical factor.

Since they can't get all the data, this is what the MESS people are concerned about. A certain product getting a bad rap because of uncertain statistical information. I can understand that and that is why the Cantina only provides anecdotal information. Anecdotal is going to be out there anyway and sometimes posters actually contradict the general perception.