Author: leonid
Date: 03:54:54 06/21/00
Go up one level in this thread
On June 21, 2000 at 01:03:12, blass uri wrote: >On June 20, 2000 at 17:19:58, leonid wrote: > >>On June 20, 2000 at 15:31:24, blass uri wrote: >> >>>On June 20, 2000 at 13:38:08, leonid wrote: >>> >>>>On June 20, 2000 at 08:39:38, blass uri wrote: >>>> >>>>>On June 20, 2000 at 07:48:31, leonid wrote: >>>>> >>>>>>On June 19, 2000 at 19:52:46, Tom Kerrigan wrote: >>>>>> >>>>>>>On June 19, 2000 at 19:38:10, leonid wrote: >>>>>>> >>>>>>>>Anyway, if simple 12 plies deep search is still not enough, much deeper search >>>>>>>>should give correct answer. Analysis in this situation is just like learning >>>>>>>>about our Moon through telescope. In practical and actual moment it almost >>>>>>>>unique way to know something about this satellite. But the best way is to go >>>>>>>>there personally. Material echange logic in chess is just like going to the >>>>>>>>Moon. Expensive but best way possible. >>>>>>> >>>>>>>I don't understand what you mean by material exchange logic. >>>>>>> >>>>>>>Your logic to find material wins should be the same as your logic to find >>>>>>>positional wins, except for the evaluation function. >>>>>>> >>>>>>>-Tom >>>>>> >>>>>>Material echange I use in order of not saying, what is so natural for me to say, >>>>>> "positional logic". For me this part of the logic (contrary to the "mate >>>>>>solving logic". Mate solving logic search for mate only and don't use any >>>>>>material avaluation) calculate how much material could be gain after given move >>>>>>at given depth. For instance, given move, after its deposition in upper ply, >>>>>>give 8000 and later, 8 plies deep give -2000. Final result is 6000. This >>>>>>evaluation say me practically everything about position. Problem is only that >>>>>>number of plies that you can search, after the time that you have, is limited. >>>>>> >>>>>>Leonid. >>>> >>>>>You could also say: >>>>>Mate or not mate say to you practically everything about the position. >>>>>Problem is only that number of plies that you can search is limited but if you >>>>>search deep enough you can decide if the position is a draw(no possible mate) or >>>>>a win for one side. >>>>> >>>>>I do not think that it is correct because practically you cannot search deep >>>>>enough and for the same reason only material evaluation tells you practically >>>>>almost nothing about the position(I agree that theoretically it can tell you >>>>>everything but I am not going to live 10000000000000000000000 years to see it). >>>>> >>>>>Uri >>>> >>>>Logic that see in what way will go material echange (who will take how much) can >>>>say everything in perfect way, including mate or draw. Reason for having two >>>>logics is mainly technical, consideration of speed and perfection. Mate logic, >>>>that goes in its thinking only after two factors: check and numbers of moves, >>>>permit easely verifiable and perfect logic that find mates (or draw). >>> >>> >>>It is hard to understand your last sentence and maybe part of the >>>misunderstanding is because of wrong translation to english(English is not my >>>first language but I believe that I know enough to not translate words in the >>>wrong way) >> >>Don't worry, English is not my first language either. > >I know. >I think that this is the reason for part of the misunderstanding. >The problem here is that it seems to me that you also did not learn english >in the right way. > >Learning by looking at a dictionary for translation of words is not a good way >to learn english because it may lead to a misunderstanding. > >There are other people that their first language is not english but I have no >problem to understand them. > >> >> >>>Translating words by a dictionary when you never learned the language can cause >>>a misunderstanding(it is possible that the same word in one language has more >>>than one meaning and when you tanslate it by a dictionary you get the wrong >>>meaning). >>> >>>I understand that when you say logic you mean to a chess program. >> >>Exact! Logic through which program solve this other problem. >> >> >>>(the word logic in english does not mean a program and it is about the brain >> >> >>Brain of the program is once again logic that was put into it. >> >> >>>I can say that a person does illogical move in chess if the move does not make >>>sense). >> >>Maybe. It could be that move is just illegal as well. >> >>>I guess that you mean: >>>It is easy to discover bugs in programs that consider only if there is checkmate >> >>Reason for this that moves that lead to mate are very often just one, seldom >>few. Smallest number of moves that lead to the mate is always fixed. Response is >>all the time clear: mate existe or not existe in the number of moves that you >>asked the program to search. In "positional logic" (material exchange) response >>if never 100% clear. You almost never can say that you found the best value for >>given move. It could be that one ply deeper you will obtain even bigger value. >>Very often number of moves that have identical value goes beyond 10. And >>finally, value that you obtain depend on the value for each piece that you >>indicated in your data. The only reasonable way, until now, to debug "positional >>logic" I found in making my minimax positional logic and compare its response to >>the advanced alpha-beta. All other suggestions, that people said me until now, I >>found not valid. >> >>>in a fixed number of movesand these programs are perfect for finding mates or >>>draws. >> >>Word "perfect" stay here in the the sense that program never make mistake. >>It will never say mate in so moves ahead when this mate not existe, or existe in >>different number of moves. I speak about search by brute force. Usually this >>part in chess programs are buggy. I suspect that the reason for this is that >>they search their response through the material echange. >> >>> Logic that >>>>is based on echange of material is prone to mistake and very difficult to be >>>>verified for bugs. Slight change in value given to each piece on the board can >>>>change instantly its final result. >>>> >>>>Now practical side of all this. Search done by mate solving logic and logic that >>>>look into material echange is limited in time for each game, but so is for >>>>human. In this moment mate solving logic see better and find sonner mate on the >>>>board that every world champion can see. It don't make mistake as human make. I >>>>do remember few books with mate positions that had more that few mistakes in it. >>>>One book that I used was printed in the 30th. Few books that was printed in the >>>>80 had many mistakes as well. Human is no more the best to recognize rapidly >>>>mate in chess game. >>>> >>>>Positional logic is still not as performant as mate solving logic. >>> But it could >>>>be that its basic thinking is also as perfect as in mate solving logic. Actual >>>>computer power is not enough to make it uniquely based on its raw speed, that is >>>>probably is as good as it could be. Meantime, all the endless astuce of the weak >>>>must be implement to make it work even now. >> >> >>>I do not understand the meaning of all the endless astuce of the weak. >> >>"Astuce" is French word that I put wrongly in English text. Astuce is clever >>invention in order of saving you from difficult situation. >> >>"Astuce" is everything that you must invent in real life in order of lifting 200 >>killos by one hand. If it will be only 10 killos, for instance, you will do this >>without that much thought. In this moment my "positional logic" permit me to >>lift 10 killos where I would like to lift 200 killos. To order to complet my >>dream I must invent some lifting artifact. Do some "astuce". In chess it is >>called tactics, and by so many other names that I still don't know. >> >>>I also do not see the point of "computer power is not enough" because it will >>>probably never be enough for chess in the next 1000 years if you do not evaluate >>>things that are not material. >>> >>>Uri >> >>Uri, everything is MATERIAL. Not everything we can calculate. >> >>I doubt that we must wait for so long to beat human by almost raw power of our >>computers. Computer must only reach human capacity and no more that this. >>Actually, I am more pessimist that you are. My predication is that computer will >>become so quickly, and so badly strong that human will abandon chess game for >>ever. Why even to try to play against somebody that win from you all the time? >> >>Leonid. > >I do not believe that the best humans are going to have no chance against >programs in tournament time control in the near future(maybe we need to wait 20 >years for it). Ura, I remember that some 10 years ago I was talking with one guy in the park about chess programs. I was not even near to my idea to write a program. In this park still now many people come during the Summer to play chess. The man was very offended when I said that human will finally loose its first position as a chess player and soon. But this really happened only few years later. The same I expect to take place everywhere in chess game. Maybe you still remember some phenomenal men able to calculate many numbers and at every high speed. They was exibiting their capabilities up to the late 60th, but now no more. Reason for this is superhuman capability to calculate for very cheap calculator. With chess game finally should happened the same. I expect that this will be not that long to happened, maybe even 30 years down the road. The next big strength in chess program I expect to happened almost overnight, with the arrival of the next generation of 64 bits computers. About 2.8 Ghz computers (even not 64 bits) people already started to talk. Leonid. >I also do not believe that material only evaluation can beat GM's even if we >wait 20 years and computers are 1000000 times faster. > >Uri
This page took 0.02 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.