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

BGonline.org Forums

Reaching DMP

Posted By: Timothy Chow
Date: Monday, 9 June 2014, at 2:31 p.m.

In Response To: Reaching DMP (scotty)

An ambitious, but potentially very valuable, project would be to extract a computer model of human errors from such a database. If a computer had a good understanding of what kinds of errors humans tend to make, or even better a way of modeling different types of human play, then there are at least two very interesting applications I can think of:

1. You could program the computer to play like a human. The kinds of errors it would make would be more like the errors that humans make. Currently, it seems that the only way to make a computer play weaker is for it to deliberately sacrifice equity in a more-or-less random fashion. This can lead to strange plays that don't yield a satisfying playing experience.

2. You could program computers to exploit human errors. Currently, the BG community has largely fallen into the rut of regarding "the right play" as synonymous with "the bot play," i.e., the equilibrium strategy play, with the only widely recognized exception being fish METs, or perhaps the general principle that you should complicate play and lengthen a match against a weaker opponent. A computer that is capable of rolling out a position with various human styles of play would undoubtedly yield fresh insights about the game, in particular about how to extract maximum equity from an imperfect 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.