Author: Peter McKenzie
Date: 16:43:20 09/08/02
Go up one level in this thread
On September 08, 2002 at 18:02:13, Rolf Tueschen wrote: >On September 08, 2002 at 17:30:15, Peter McKenzie wrote: > >>On September 08, 2002 at 16:10:24, Rolf Tueschen wrote: >> >>>On September 08, 2002 at 15:33:16, Peter McKenzie wrote: >>> >>>>On September 08, 2002 at 13:08:33, Rolf Tueschen wrote: >>>> >>>><snip>> >>>>>But the game is already over here. >>>> >>>>Yes agreed. In case you didn't read it, I said that already. >>>>I will spell out my reasoning: >>>>The point is that the program should be able to STATICALLY evaluate this type of >>>>position as good for black in order to use search EARLIER in the game to avoid >>>>this type of position. >>>> >>>>>Don't know who made yor book, but no PC can >>>>>find the necessary 12.e3. Rd2 is bull. >>>> >>>>Rd2 is not best, but white is not lost after that move. Of course it would be >>>>nice if the book had something other than Rd2, but you can't ALWAYS rely on the >>>>book to get you through the opening! >>>> >>>>> You simply were outbooked in the game. >>>> >>>>The book was a small factor, more important was how Warp played after book which >>>>wasn't very good! Brutus understood the position, Warp did not, therefore >>>>Brutus won. >>>> >>>>>You simply don't have the time to make such nonsense as Rd2. The black Bc4 is a >>>>>known motif. Without book comps would be easy bully. How coul a comp calculate >>>>>the later king safety?? >>>> >>>>Its not so hard, you can see from the replies that crafty, isichess and diep >>>>have a reasonable idea. Some other amateur programs aren't so good here. >>>> >>>>If you continue to assume I am an idiot I won't bother replying to your posts >>>>:-) >>> >>>Of course not! Did my English sound so negative? It was simply my judgement >> >>Yes, somewhat. Not to worry, thanks for your comments anyway. >> >> that >>>Rd2 is wrong. e3 must be played instead. Rd2 loses tempi. If you want, the >>>"later" position is complex and easy at the same time. The main point is the >>>lacking o-o in a position where Black is fully developped and, and that is the >>>most important point, White cannot make the o-o because of the Bc4! But the >>>position has many more weaknesses of White. Normally you would like to play Nd2 >>>to exchange the Ne4. But this is not possible due to tactics. Because the Q is >>>not protected. >> >>I know Rd2 is dubious, but that was never the point of my post. >>No need to lecture chess to me (well not tooo much), I had 2300 fide rating some >>(ok many) years ago you know :-) > >Sorry, if I went too far, but I didn't know your number. It's surely good >enough. :) > >But honestly, I wasn't teaching you; I was arguing for my position. The move is >so ugly that you must forbid it in the book. I see a little contradiction in >your position. It is a book position but you want to examine why the later >position must be seen as weak by your program to be able to see - without book - Its not a contradiction. This problem (misevaluation of centralised king which cannot castle in opening/middlegame) can manifest itself in many many different positions, and it is impossible to have them all in book! I think the theme is common enough to warrant me fixing it, or at least improving my program's understanding of such positions. >that Rd2 is wrong? As I said the possibility Ne4 and the Rd2 must move by force, >and the lost of two tempi Rd1, Rd2 and the need of further three tempi for >castling (e3, B~, o-o) should be implemented as weak. Why Rd2 shouldn't be >treated like h3x~g4 or h6x~g5? There you can't wait for the evaluation of later >position either. These comments don't fit with the search + evaluation paradigm used by most if not all modern chess programs. > >I must add if necessary that I'm really a complete alien to chess programming. >So please do never take anything I write as a direct critic of your job of a >programmer. On the other side I would hope that you might understand why such >comments from a complete outsider could perhaps give interesting ideas at times. >Hope this was the case here. Food for thought. cheers, Peter > > > > >> >>> >>>Let me summarize your interest and the analyses this way. The move Rd2 is a move >>>you should never play if you have _not_ made deep analyses. Rd2 is an odd move >>>so to speak. That this move is odd is not so much a question of the "later" >>>position you gave us. Rd2 could be shown wrong in the position it was played, I >> >>What? You are proposing I should have pruned out Rd2 at ply 1 using general >>principles? For a human that is possible, but if you try to do that with a >>program you will have a disaster! >>Your comments seem to indicate a lack of understanding about how a chess program >>chooses a move?! > >Misunderstanding. I meant your book should contain e3 as the correct motif. >Rd1-d2 is a bad motif. > > > >> >>There is a big difference between being able to say, as a chess player, 'this >>move is bad because of X, Y, and Z' and knowing what to change in a program to >>avoid such moves. My judgement is that most of the bad moves played by Warp vs >>Brutus were caused by poor static evaluation of the centralised white king (this >>position is just one example), therefore it did not have enough incentive to >>castle. I don't see any reason to change my judgement. > >But this question is way beyond me of course. King safety is direct programming. > >Thanks. >Rolf Tueschen > > >> >>cheers, >>Peter >> >>>would pretend. Simply because White hasn't o-o yet. Also because bNe4 is always >>>a direct attack against the R. More chess advice is wanted. >>> >>>I hope you ask more questions, not that I could always give the answer. BTW it >>>was the first time that I wrote my opinion to a programmer in a direct question >>>about chess. I must take care that it's not me in the end who is the chess idiot >>>after all. :) >>> >>>Rolf Tueschen >>> >>>> >>>>Peter >>>> >>>>> >>>>>Rolf Tueschen >>>>> >>>>> >>>>>> >>>>>>> >>>>>>>Regards, >>>>>>>Gerd
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.