randomSample

[BCS]

Hello BCS,

> Check my reasoning on this:
> 
> It will be skewed because as soon as the system gets behind (and
> statistically it will half the time) the elements start being selected
> with a higher probability.
> 
> Take the related case, select each element with equal probability
> until you have enough selected or have just enough left to get the
> correct amount. The last element will get selected about half the time
> no matter what percentage you want.
> 

after some ad-hoc tests ("fun time with Excel!") it seems that the above 
is off in the weeds.