The one.cgi file has a ratio of (head:tails) of (10:11). Therefore p(heads) = 10/(10+11) = 0.4762, close to fair, but not fair.
It's going take a big n to measure this difference in an experiment.
It turns out that for a binomial with probability p and trials n, the mean is \( n p \) and the \( \sigma \) (the standard deviation) is \(\sqrt{n p (1-p)}\).
So ... how about how many trials will our experiment need to see that this is not a fair coin ?
Answer: Find n such that \( 2 \sigma \) is the difference between n/2 and \( n p \) ... which says that n needs be bigger that 1760 .
