Author: Vasik Rajlich
Date: 03:40:56 01/14/05
Go up one level in this thread
On January 14, 2005 at 02:39:42, Mridul Muralidharan wrote: >On January 13, 2005 at 14:00:05, Drexel,Michael wrote: > >>On January 12, 2005 at 14:02:53, Mridul Muralidharan wrote: >> >>>On January 12, 2005 at 13:01:29, Drexel,Michael wrote: >>> >>>>On January 12, 2005 at 12:42:05, Anthony Cozzie wrote: >>>> >>>>>So let me see if I understand this conversation correctly. >>>>> >>>>>1. I state that the 6 man tables are worth 100 elo >>>> >>>>I thought you were joking, but obviously I was wrong. >>>> >>> >>>I have no quantitative way of accurately guessing this too - but "depends on >>>program" maynot be a wrong statement ? >>>And both are definitely agreeing that there is a non-trivial improvement in >>>performance - right ? Then why disagree for the sake of disagreeing !!!! >>> >>> >>>>> >>>>>2. You disagree, and state they are worth 50 elo >>>>> >>>>>3. You do this by pulling numbers out of your *** >>>>> >>>>>4. Since the full 6-man set hasn't been generated, and the elo gain is almost >>>>>certainly different for different programs, we are both guessing. >>>> >>>>Yes, but in this case Uri's guess is much more educated. >>> >>>Hmm , I dont see how - just 'cos there was a "women" reference ? :) >>>Jokes apart - the point to be taken is - they could be a SIGNIFICANT improvement >>>: and would be the world of difference between a loss and a draw (or a draw and >>>a win). >>>Depends on how you eval , and what you do in your search (extensions and qsearch >>>/threat detection). >>>Ofcourse , if you have a junk endgame eval with quiet decent middle game eval - >>>your improvement can be much higher than what both of them quote ! >> >>Certainly we don´t talk about the rare programs which throw away pieces every >>time they reach a 6 men ending. >> >>The strength of the programs in the late endgame is on average already very >>high. There is simply not much improvement possible. >> > > >I would like to differ on this. >EGTB's help you in taking a safer path into endgame from middle game. >Most of the positions at end of your pv in a middle/late middlegame would be >actually an endgame position. (especially so if you extend passed pawn pushes , >checks , recap , etc - I dont do all these , but that is a different matter :) >). > >What I mean by "safer" is that most of the programs now need atleast a few ply >to evaluate an endgame as a win/loss - their static eval is miserable in most >respects in this aspect (triangulation , locked pawns with king battling for >that single opening , rooks and bishops dancing around the board , etc come to >mind :) ) - note : I have seen lot of programs make remarkable improvement in >this respect , but they are not perfect ... >Rude heurestics like rook vs bishop is draw , (or similar) have fallen short in >cases since it is not perfect (I am talking about end of pv eval here - not >start of pv search) ! >So you might be taking a path into endgame which according to static eval is >draw/won - which if you did an egtb probe would be a loss/draw. >There is no substitute for perfect knowledge - and I have seen it time and again >happening for mine (lots - I personally sucks badly at endgame , so do all my >progs!) as well as really good programs. >Have we not seen so many positions being posted here where the version with EGTB >find solution in like millisecs while without it takes 25+ ply ? >Now , take this from the view of this position having occured at end of your pv >in a superlative middle game position ? you would actually be giving up all your >advantage to get into an inferior position 'cos it was just too tough to >evaluate this resulting position. >And I have not even mentioned all the interior node cuts that you will be >getting 'cos of this ! > >I might actually be preaching to choir here - most people here are much more >knowledgable about endgame and TB interactions with search , along with the >tradeoffs and benifits in this forum than I am , but my 2 cents :) > >Mridul > In practice I think it's pretty rare to have these "critical decisions" where you have to either choose some endgame, or some completely different middlegame. Usually the game just flows from one move to the next, and it's enough for your eval to like things which are generally good and dislike things which are generally bad. Getting the endgame scores "really right" might not help that much. Vas > >>Michael >> >>> >>>But are we not quibbling over nitty gritty details ?? 50 , 100 , 125 - what does >>>it matter : it would be a substantial improvement !!! >>> >>> >>>Mridul >>> >>><snip>
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.