On Fri, 15 Aug 2003 08:23:31 -0400, Paige Miller
<[EMAIL PROTECTED]> wrote:

>Jeroen Donkers wrote:
>> -- CALL FOR PARTICIPATION ---
><...snip...>
>> What is RoShamBo?
>> -----------------
>>
>> RoShamBo is a game played all over the world. In Japan, where it is
very
>> popular, it is called Jan Ken and in the USA and England they call
it
>> Rock-Paper-Scissors. The game is quite simple. Two players sit
facing each
>> other and on a sign they first stamp three times with the right
fist in the
>> open left hand and then they simultaneously make a gesture with
their right
>> hand.
><...snip...>
>> Computer-RoShamBo!
>> ------------------
>> The purpose of the competition is to write a Java program that can
play
>> RoShamBo. Your program will play against all other participating
programs
>> and the program winning the most games is the champion.
>
>Forgive me for being skeptical, but enlighten me, please ... won't
the 
>wins be based purely on random luck rather than any outstanding 
>mathematical logic, or rather than any outstanding computer
programming? 
>  If so, what is the point of declaring a champion?

If everyone plays according to a uniform distribution of plays then
who wins is just based on random luck.  But, if one person deviates
from that uniform distirubtion than someone may be able to exploit
that and gain an edge in expected value.  


>
>So, if you had an infinite number of turns, the best strategy is to
use 
>equal probability on each outcome. No problem, I understand that. 

But, it's not correct unless everybody plays that way.  If one person
deviates from the uniform play then you won't be harmed by it, but
your failure to exploit it might cause you to now do as well as
someone who does attempt to exploit the one player who deviated from
uniform.




.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to