From: michael facer LAVA NET> Date: 12 dec 1996 Subject: Re: Evaluation (was Re: Computer Shogi) >> >How about this for pieces on the board: >> > >> >Unpromoted: Promoted: >> > >> >Pawn = 10 40 >> >Lance = 30 50 >> >Knight= 40 55 >> >Silver= 50 60 >> >Gold = 60 - >> >Bishop= 100 150 >> >Rook = 110 160 >> > > >> Why is a promoted pawn worth less than a gold? In fact a tokin is worth >> more. > >A promoted pawn is worth less than a real gold; while they are >functionally the same, when you take a promoted pawn you only get a pawn. >It is worth more than a regular pawn because of its enhanced mobility. O.K., suppose we use your suggested values--each side starts the game with material worth 660 points. One side makes a tokin (a promoted pawn) and succeds in trading it for a gold. what are the new point totals? For the tokin side; 660+60-10=710. For the other side 660-60+10=610. So now there is a difference of 100 points, which according to your table equals 2 silvers. So the value of creating a tokin and exchanging it for a gold is a 2 silver advantage. Of couse as Jereon pointed out this depends on the tokin side being able to force the exchange. The tokin side may not always be able to get a gold, but it can usually force some kind of exchange and get something for nothing--otherwise it would probably have been a waste of time to make the tokin. But occassionally for example one makes a tokin to chase the rook away, and then the tokin ends up off in a corner by itself with little chance of capturing anything or of being of any further use. What is it worth? Close to nothing? Probably, and yet it would still be worth more than a gold would be worth if it was in the exact same position. Why? Because your opponent can gain nothing by capturing your tokin, but he very well might take the time to capture a gold in the same position and then use it against you. So even in a position where a tokin is useless, it is still better than a gold would be in the same position. > >> A promoted Bishop is supposed to be worth 3 silvers? Is there anybody out >> there who wouldn't jump at the chance to take the three silvers? > >Let's put it another way: is getting a bishop promoted worth losing a >silver or a promoted lance? (True: all things are situation dependant, but >how about in general?). There is no general case. There is an endless number of possible situations in which the answer would be either yes,no or maybe. > >> The value of any piece can change with every move. In Shogi, the essence >> of good strategy is to attempt to refute material value tables such as you >> have listed. At least that's how I see it. > > This is quite probably true, but it doesn't make it any easier to >make the computer any smarter :) Hey, nobody ever said it would be easy. If you want easy, stick with a simple game like chess. >I am trying to get a stable engine working with a good foundation first. >After that I will be adding strategical assessment etc... If you want a good foundation , forget about basing it on material profit. Make good shape your foundation by using Joseki. Even in a simple game like chess, most (if not all) strong programs are heavily dependent on their books to provide the foundation of a strong position from which to play. But anyway, once again, Good Luck! Mike