| |
BGonline.org Forums
Best online data source for studying opening replies?
Posted By: Timothy Chow In Response To: Best online data source for studying opening replies? (Matt Ryder)
Date: Thursday, 24 December 2009, at 7:18 p.m.
Matt Ryder wrote:
I remember there was some controversy at one point, with some experts claiming the Snowie live-cube rollout option was a bust.
One major problem with a live-cube rollout is that if the bot does make an error, then the effect of the error tends to be magnified since you're doubling the value of the game.
People have more experience with the bots now, and have found that GNU's 2-ply cube action is really not so bad. Of course in some positions one still has to be cautious. I recently posted a GNU rollout which, if taken at face value, says that White is too good to double after any play by Blue—in a money-play position with the Jacoby rule and a centered cube! In other positions, the bot will claim that you're too good to double even when the opponent has already borne some checkers. And David Rockwell recently mentioned that GNU 2-ply cube errors may be affecting its assessment of how to play 41$-44. But the relevant question is not whether live-cube rollouts are perfect; it's whether they're better than the alternative. Currently, the answer seems to be yes.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.