| |
BGonline.org Forums
Rollout: Several different plays here...
Posted By: Timothy Chow In Response To: Rollout: Several different plays here... (mikeg)
Date: Friday, 28 May 2010, at 2:32 a.m.
Good observation...it's worth checking to see if the duplication is important.
Let me remark, by the way, that I think it's quite "dangerous" to think along the lines of "now that I see the rollout it is obvious that Play A is best because of Reason X." If it really were so obvious that Play A is best because of Reason X, then wouldn't this be obvious before seeing the rollout? The rollout has only told you, "The bot thinks Play A is best." The rollout has not given you any reasons, nor does it even tell you that Play A really is best, because the bot might be wrong. Furthermore, even if Play A really is best, it might not be because of Reason X but because of Reason Y. What's likely going on is that you're just being intimidated by the bot and coming up with ex post facto rationalizations of why the bot must be right. Since the only function of the rationalization is to justify the bot, it doesn't matter whether the rationalization is true as long as it supports the bot's preference. When you discover that you made an error when performing the rollout and that when you correct the error the bot changes its mind, you can then promptly change your mind and say, "Ah, now it's obvious that Play A is wrong because of Reason Z." O.K., I'm exaggerating, but I think you see my point. One has to be careful not to accept Reason X just because it supports the bot's play. It's important to test whether Reason X is correct by varying the position in different ways to see whether the theory that Reason X is correct stands up to further experimental evidence.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.