Author: Uri Blass
Date: 06:35:38 01/30/06
Go up one level in this thread
On January 30, 2006 at 09:08:09, Michael Diosi wrote: >On January 30, 2006 at 08:58:24, Uri Blass wrote: > >>On January 30, 2006 at 08:51:12, Uri Blass wrote: >> >>>On January 30, 2006 at 08:40:09, Uri Blass wrote: >>> >>>>On January 30, 2006 at 07:49:53, Michael Diosi wrote: >>>> >>>>> >>>>>Hello. >>>>> >>>>>I have a few, this one is from Russel/ Norvig : Advances in Artificial >>>>>Intelligence A Modern Approach (page 130) >>>>>[d]8/1p1P4/k1p5/8/8/3PPPPP/r7/7K b - - 0 1 >>>>> >>>>>You can see here that black can give a series of checks "pushing" the inevitable >>>>>queening over the horizon. So programs how don't see it will have a positive >>>>>score for black. >>>>> >>>>> >>>>>I can look for more if you want. >>>>> >>>>>MD >>>> >>>>This is a bad example >>>> >>>>Amir Ban is correct. >>>> >>>>The horizon effect is effect that cause the program to play worse move because >>>>it does not see deep enough. >>>> >>>>This example is not the horizon effect but only not seeing deep enough. >>>> >>>>Uri >>> >>>Note that not every case that the program play worse move because of not seeing >>>deep enough means the horizon effect. >>> >>>The idea is that the program has some illusion that it can prevent some problem >>>but only cause itself a bigger problem because it delay the problem to position >>>that it does not see. >>> >>>In the example the program has an illusion that it can prevent the pawn from >>>promotion but it does not cause it to do mistakes and sacrifice material so I do >>>not think that it is good example. >>> >>>Without check extensions if the program sacrifice the rook then it can be a good >>>example. >>> >>>Uri >> >>The following sentence from the article summerize it: >> >>"The horizon effect is said to occur when >>the delaying moves unnecessarily >>weaken the position or give up additional material >>to pospone the evantual loss" >> >>Uri > >Hi, > >This is exactly what its happenening. It is delaying moves unnecessarily giving >checks and in the end material (the rook). The pawn promoting move is beyond the >horizon of the program (not really in this case). In other words the program >can't see that white will promote anyway. Each black move brings white closer to >promotion and to the horizon. > > > >MD The delaying checks in your example do not weaken the position and there is simply no better choice. Uri
This page took 0.01 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.