By Shalosh B. Ekhad and Doron Zeilberger
(Exclusively published in the Personal Journal of Shalosh B. Ekhad and Doron Zeilberger and arxiv.org )
Written: Jan. 4, 2018
A certain sampling process, concerning an urn with balls of two colors, proposed in 1965 by B.E. Oakley and R.L. Perry, and discussed by Peter Winkler and Martin Gardner, that has an extremely simple answer for the probability, namely the constant function 1/2, has a far more complicated expected duration, that we discover and sketch the proof of. So unlike, for example, the classical gambler's ruin problem, for which both `probability of winning' and `expected duration' have very simple expressions, in this case the expected number of rounds is extremely complicated, and beyond the scope of humans.
The following package is also needed in the same diretory
Doron Zeilberger's Home Page