Author: Uri Blass
Date: 03:54:48 05/05/02
Go up one level in this thread
On May 05, 2002 at 06:22:37, Pekka Karjalainen wrote: >On May 04, 2002 at 07:19:48, Vine Smith wrote: > >>>http://www.buggy-online.com/english/match_samb_buggy/match_samb_buggy_e.htm > >>If you look at the analyzed games on the site, you'll notice a striking >>similarity to the situation in computer chess -- the program was a dangerous >>tactician, but had no "feel" for the position. Samb won the tiebreak by >>exploiting a positional weakness he had noticed in an earlier game. Since >>draughts is a more tactical game than chess, this would seem to indicate that >>positional problems will continue to plague chess programs far into the future. > > I am not sure I know what you mean by more tactical, but draughts would seem >to be less tactically complex to me than chess, since there are only two types >of pieces and the board is actually smaller. Only 50 squares are used in >draughts of the 10x10 board. > > Currently it seems that board games fall into a few major categories in human >vs. computer play: > >1) too simple > >tic-tac-toe - which is always a draw with competent human or computer play > >2) solved by computers only > >nine-mens-morris, go-moku > >where good human players can draw at best, but cannot beat the computer any >more, unless the game is a win with the first move advantage (or similar). even >then for tactically complex games the computer will be likely to win anyway, >since it does not make mistakes. > > (It was once claimed on the net by a nine-mens-morris player that the program >that solved the game cannot necessarily win a nmm tournament, since it is too >easy to draw against, while the best human knows how to swindle other human >players.) > >3) dominated by computers > >8x8 checkers, 8x8 othello > >where the top humans have not been able to demonstrate ability to beat the >computer for some time, yet the games are not yet solved or likely to be solved >in the near future. how many possible positions are in 8*8 checkers? I think that there are only 32 squares and 4 kind of pieces and it means that 5^32 is an upper bound for the number of position(a square may be empty) 5^32 is only an upper bound and the number of practical positions to analyze may be clearly smaller. If you also remember that you probably do not need to search until the end of the game in order to play the best move then it seems that checkers is a game that is likely to be solved in the near future if it is not practically solved today. In chess it is easy to find for every program a legal position when it cannot find the right move in a reasonanble time. I do not know if it is possible to do it in checkers for chinook and even if it is possible then it does not mean that the game is not practically solved because if you cannot get that position in a game against chinook then the position is not relevant. > >4) contested by humans at top level > >chess, 10x10 draughts, a few others > >games where the top computers do not yet clearly demonstrate superiority to >humans. the indications might be that they will get there, but we cannot know >how hard the last hurdles are to overcome. > >5) dominated by humans > >shogi, go > >games where there is still no question of seriously challenging the very best >human players by computers. shogi might fall to category 4) some day soon, but >I don't think computer go is going anywhere :-/ >Now, looking at this list, what does it mean that a game A is more tactical than >B? Can you explain? Does the term positional play in draughts have any >relation to positional play in chess, since they are two completely different >games? I know nothing about draught but I can explain the meaning of more tactical game. A tactical game is a game when search is relatively more important and a positional game is game when the evaluation function is more important. If searching one ply deeper gives almost nothing in rating the game is not a tactical game. I understood that it is the case in go. Uri
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.