Author: KarinsDad
Date: 21:22:44 03/01/99
Go up one level in this thread
On March 01, 1999 at 19:37:13, Stuart Cracraft wrote: [snip] > >I guess my question to all of this is when would you not care? You >always have to care. Though there are degrees of caring... That is the point. I care a lot if my king is exposed or my pawn structure is compromised. This is the element that I was missing in my evaluation. >The only >time I would think you would not need to care is when the opponent >doesn't have enough material to mate! You do not need to care (a lot) if your opponent's pieces are on the opposite side of the board and it is a closed position. I'm sure that there are other possibilities as well. Most of these should be handled via the "who controls the squares around the king" concept (see below) so that you do not have to have special code for each case you can think up. > >What I've seen mostly is factors applied to the overall king safety >score based on phase of game, whether enemy queen is present, etc >that would heighten or lessen the king safety term's value. Actually, I currently do not worry about which enemy pieces are on the board when it comes to king safety. I worry about who controls the squares immediately adjacent to the king and how many pieces of each color attack the squares immediately adjacent to the king and two squares away. So moves that increase the safety of my king by minimizing the number of attacking pieces or increasing the number of defending pieces have a higher evaluation. However, I have a feeling that once I test this, moves such as g3 when White has castled kingside may receive a higher score (say for example a bishop attacks the h2 square, g3 protects the h2 square by having the h2 and f2 pawns protect the g3 pawn, however, this is rarely a good move) and I'll have to adjust for that. One final note. Check to see if your king is trapped during the evaluation. If your king cannot move, the chances of a sacrifice being more worthwhile increases. Trapped kings should be worth less than untrapped kings in the evaluation. KarinsDad :)
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.