Hacker News Reader

Here's someone testing d6s to see if 1's really do come up more often. He used student labor instead of a machine.

http://www.dakkadakka.com/wiki/en/That%27s_How_I_Roll_-_A_Sc...

That "study" claims rolling a 1 has a probability of 29% for the tested dice. I saw it when it was published and was very sceptical, and so where others. One guy decided to roll 1000 D6 and count the number of 1's, he got 169. Pretty close to the expected number and far off 290 as the study would predict.

I calculated that the chance of rolling 169 or less ones out of 1000 rolls, assuming the chance rolling a 1 is 29% is 0.0000000000000000000459 (far far below the p-value needed to prove a new particle btw).

To quote the guy rolling the 1000 dice:

"Basically, I'm forced to assume that there is three options

A) Said students decided 'bugger that old idiot telling us to roll a bunch of dice', made up some numbers and went for a beer B) He can't do data processing worth poop C) He is outright lying

Take your pick, but it's one of those three."

Complete discussion thread is here: http://warhammer.org.uk/phpBB/viewtopic.php?f=3&t=31419

Reply on news.ycombinator.com