GameScience Dice: Rolling More True, Kinda

Awesome Dice has a really interesting blog post up about the claims that "sharp" GameScience dice that haven't been tumbled like most other dice produce more even, "true" results. To test that theory, they rolled two d20s, one from GameScience and another from Chessex, 10,000 times each (and double checked by rolling two more 1600 times each) and compared the results. You can go read the article for the specifics, but the bottom line is that GameScience's dice don't roll true enough to be depended upon to act as a high-quality random number generator, but they do roll "true-er" than dice that have been tumbled, with one notable exception: GameScience d20s don't roll 14 nearly as often as they roll every other number, almost certainly because of the "flash" left on the opposite face that sticks out from the die. I've been using GameScience dice for a while now as my personal set (except for my d30, which I haven't seen a GameScience version of); it's nice to have that choice confirmed, but it looks like I need to figure out how to take the flash off of my dice…

[Disclaimer: This isn't exactly a review, especially of Awesome Dice, but I do want to point out that Brian of Awesome Dice is the one that brought this to my attention.]