> I programmed this in college and the break-even
> point is 23 people (just barely over/under 50%,
> forgot which) and 39 is almost guaranteed (99.99+
> percent if memory serves). I think 36 hit the 99%
> plus a little.

It would not astonish me that the counts for 99% were that low, I just didn't have a calculator on which I could do it easily... 36 does "feel" a bit low for 99%, the incremental multiplications from 30 to 36 are all about 9/10, so one is multiplying 30% by (9/10)**6 which comes out about (9/10)**6=1/16, so I would think it was closer to 98% for 36 people with 99.5% for 39 people, but that's an off-the-cuff estimation, one should *always* do the real calculations, the numbers are often a surprise to one's intuition....