| |
BGonline.org Forums
Two things that need to be done with XBG
Posted By: Rich Munitz In Response To: Two things that need to be done with XBG (phil Simborg)
Date: Tuesday, 30 June 2009, at 3:47 p.m.
It all depends upon what you are trying to do as to whether a 0.02 error is meaningless or huge.
If you are rolling out benchmark positions and need to know exact answers and want to achieve some fundemental understanding of a position type, then 0.02 is huge.
If you want to simply improve your game by having your blunders pointed out so that you can try to understand where the heck you went so wrong, then 0.02 is largely irrelevant.
I tend to be of the latter camp. I let matches of interest roll out over night on Snowie (so speed has not been a huge concern), and I review the results "whenever". Instant feedback can be a distraction if I play the bot. I used to get that from Jellyfish because there was no other choice and when Snowie came along and allowed you to just play and analyze later, I found that to be an advantage.
Speed is of huge value for people in the former camp (and I am sometimes in there). In that case, you do want to have substantial trust in the results. I am all for a new bot on the block that is 5 times as fast, but lets verify that its results are trustworthy. In my opinion, there is little value to getting wrong answers in 20% of the time. I certainly hope that is not the case because if XBG passes the test, it would mean a tremendous tool at our disposal and I would be right in line to fork over $50.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.