You could estimate the importance of a word's hooks statically, but that requires knowing not just the number of hooks a word has, and their total scores, but also how often that word is played, and is played in a position that can be hooked, and how often hooking that word is then used to make future best moves. I found simulation to be the easiest way to be sure the resulting list took all these factors into account somehow.
Eugene ________________________________ From: Graham Toal <[email protected]> To: [email protected] Sent: Wednesday, January 21, 2009 3:41:11 PM Subject: Re: [quackle] Re: "hookability" list for OWL2 - 8 million games of simulation On Wed, Jan 21, 2009 at 4:14 PM, Eugene Deon <ejd...@yahoo. com> wrote: > I've now compiled 8 million games of play for estimating hookability. Here > are the lists for whoever might be interested: I hope you'll excuse a possible naive question, but why do you need to simulate many games to compile these lists? Can't they be determined statically from the wordlist and the tile frequency/values? Graham
