plasmoid wrote:Just asking,
Mike - you said you've looked at the data extensively.
Does that include some sort of frequency calculation?
Because - as you know - a million 1's, followed by a million 2's, then a million 3's etc. would still yield the magical 3.5
I've looked at the code used to generate the random numbers - the random numbers are not in a range of 1 to 6, they're in the range of 0 to 4,294,967,295 - those random numbers are then run through a second piece of code that converts it into a number from 1 to 6. The RNG in question is a well known algorithm that has been studied rigorously and which is considered to be a source of high quality random numbers. Thus, the only thing that needs to be questioned is the functions that convert the RNG output into the ultimate results, which is why I posted a chart of all possible RNG outputs run through those functions.
When you have the de facto algorithm used to create data, you don't need data samples. The point of statistics is to acquire information about a system you are unable to access directly and fully: you can roll a die a million times to investigate its balance... or you could use special equipment to physically analyze it to determine if it has an imbalance. The latter precludes the need for the former, while the former can only give you clues as to what the latter could conclusively find.
Analysis of the code supported what statistical data had already pointed to - that there was not an issue with the RNG or its conversion into dice rolls.