*sigh* This never ends, does it?
1. The algorithm used in 8.9 is the
Mersenne Twister
2. I posted the million dice rolls, becuase I wanted to see how
KingSnakeeyes would respond to it.
3. The response by
Kingdom is utterly meaningsless as a statistical proof of whether or not the algorithm used is random or not. I could generate a file with "12345612345612..." and it'd have roughly the same results as the file I originally posted.
4. People far better than me (and I'd say most people on this site) have made statistical tests on the algorithm used. For example, the Mersenne Twister passes the
Diehard series of tests. That's a much better statistical analysis than what you are likely to come up with.
5.
DrDiscoStu suggested that I (or someone) rolls a million dice so a comparison can be made. He also stated that the fact that you've rolled a number of 6es in a row with a RNG, the probability of rolling another one is reduced. The problem with this line of thought is that random number generators don't work that way. There simply isn't a memory of past rolls. The fact that you get equidistribution is an effect of randomness; not because the system remembers what you rolled.
6. Even if the RNG actually did remember your rolls, you fail to realize an important aspect of the RNG. It produces random numbers in the range of [0, 2^32-1]. That is, 0 - 4294967295. The number generated is then converted to 0-5 by using a modulo 6 operation, and 1 is added to get 1-6. Now, in order to not taint the equidistribution, numbers that are on the high end of that scale (that is, the top few where the numbers 1-5 aren't fully represented) will be discarded. That means that the low numbers (1-4 or so) are not represented more than the high ones (5-6). So.. The numbers 1-6 are represented in about 715827882 different ways. Essentially, it means that you can roll 715827882 rolls before you're guaranteed to actually get the "same" 1 result. This is likely to be more rolls than you have made in your entire BB career.
7. The RNG is not broken. It's highly probable that it's more random than your real-life dice are.
8. Regardless of how much you tell yourself that "[t]here is no pattern. There is no pattern." (From what I can tell, this quote is attributed to Ed Miller, with regards to poker randomness), your mind will keep looking for them. There is no pattern. You will still see patterns everywhere. Double skulls, snake-eyes, lots of failed dodges in a row. There is no pattern. Isn't it funny how most of the streaks you remember are the bad ones? There is no pattern. If you convince yourself that the RNG is broken, your mind will notice the events that reinforce this belief (there is no pattern), and tend to forget the other events. There is no pattern.
Had enough yet? I've spent a LOT of time researching randomness and RNGs. I challenge those of you who claim that the RNG is broken to actually take the time to do some research as well before bringing it up. Show me the statistical evidence you have. I don't care about your thousand-odd matches you play here. Show me a proper statistical test that proves beyond doubt that the RNG the client uses is flawed. I haven't been able to find one. Prove me wrong. Please?