[metapost] Drawing a urn

[Troy Henderson]

> with this binary experiment, there shouldn't be any practical difference
> between normal distribution and any other distribution (like uniform) as
> long as probability for one outcome is 0.57, right?

Well, if I understand his problem correctly now, he's saying that he
ALWAYS wants 57 out of 100 balls to be painted "light" as opposed to
having the probability of a ball painted "light" is 0.57.  Those are
clearly two different problems, and their outcome will almost always
be different no matter what probability density function is used
(normal, uniform, etc.).  I'm not a probability expert, but I have to
believe that there should be some difference in using either normal or
uniform.  However, for the problem that we wants to solve, I believe
the randomness (no matter what probability density function is used)
is only to determine the position of the balls and not how many balls
are painted a particular color.

Troy