I'm sorry to have to beat up on you here, but you're showing a basic misunderstanding of statistics and the law of large numbers here. Anything can and will happen over the next 36 rolls. The shooter may toss six 12's in a row for all you know. If that were to happen - it would not necessary show any evidence of bias. It would simply indicate that you need additional data. It's when you look at a LARGE number of rolls that you find that there is no significant deviation from the norm over the entire body of rolls.