[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

How well would a bot do in a human tournament?

Posted By: Rich Munitz
Date: Monday, 20 September 2010, at 1:24 p.m.

In Response To: How well would a bot do in a human tournament? (Julia Hayward)

Let's imagine that the weak player averages 0.01 error per move worse than the strong player. The strong player has a choice of plays. The "wrong" play is a 0.02 error in that half the time when it works it gains .05 over the right play and the other half when it doesn't work it loses .07. However when it doesn't work, the resulting positions are more complicated and on average the game is likely to last 5 moves longer. Aside from the fact that a more complicated game is likely to produce larger errors, anything that adds moves to the game will create more errors. In this example, the lengthening of the game by itself nearly compensates for the error in the play. Add in the complication factor and the 0.05 upside when the wrong play does not get punished and we see that it could well be a blunder to make the "right" play. On top of all that, the value of cube ownership in complex positions increases for the stronger player and decreases for the weaker player, whereas in simple positions the reverse is true. The right play is the play that wins against this opponent.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.