wcaclimbing wrote:wikipedia wrote: Dice probabilities
probabilities of winning a dice roll in Risk
(various dice combinations) Attacker
one die two dice three dice
Defender one
die Attacker wins 15/36 = 41.67% 125/216 = 57.87% 855/1296 = 65.97%
Defender wins 21/36 = 58.33% 91/216 = 42.13% 441/1296 = 34.03%
two
dice Attacker wins 55/216 = 25.46% 295/1296 = 22.76% 2890/7776 = 37.17%
Defender wins 161/216 = 74.54% 581/1296 = 44.83% 2275/7776 = 29.26%
Both win one 420/1296 = 32.41% 2611/7776 = 33.58%
these are the statistical probabilities for the dice throws. if you look at the dice analyzer thread, you will find that the most recent alalyzations of the dice come out very similar to this.
The dice analyzer has effectively proven that the long run trends reflect the statistical probabilities you would expect to see after taking very large numbers of samples.
All that means is, the defender and attacker roll every die a roughly equal amount of times in the long run, which should be expected.
That doesn't have anything to do with the topic at hand, which is the streakiness (if there is any) of the dice. Yes, random occurrences mean that very good and bad streaks will occur. But they will be outside of the norm and the total trend of streaks should follow a very even distribution. The dice analyzer does nothing to look at this type of data and therefore has done nothing to quantify the apparent problem that most people complain about.
Once again, I am not saying there IS something wrong with the dice. I'm just saying there COULD be and the dice analyzer wouldn't be able to catch it at all (at least under the current programming). If someone would make a program that looks at how many times in a row you win each of the battles and then creates a distribution based on those streaks- THEN we could see how random the dice are in the short term, and everyone who is unsatisfied would be able to see the truth.