|
BGonline.org Forums
Got Variance?
Posted By: Daniel Murphy In Response To: Got Variance? (Timothy Chow)
Date: Sunday, 23 May 2010, at 11:00 p.m.
Well .. if one's claim is that the CI is underestimated in only some positions but perfectly fine in many (most? gazillions?) others, I'd like to know what one is thinking the bot is doing that would make that possible -- and how we'd identify such problem positions, and what to do about it.
It seems odd that a bot would a have problem with "typical" positions, but not with others.
Also, should we expect the CI to be correct in all rollouts of all positions after X trials? If so, why? If not, then picking/finding one such position only supports that "not."
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.