Author: Uri Blass
Date: 08:38:45 08/12/02
Go up one level in this thread
On August 12, 2002 at 10:24:43, Maurizio De Leo wrote: >>>>I never tried to play checkers seriously but at least >>>>my impression is that the game (if you use 8x8 board) >>>>is clearly simpler than chess. >>> >>>You definetely have to define simple. Any game that is beyond human solving is >>>complicated in the same way for human. In Othello the best programs are working >>>night and day for solving the game and maybe in the next years they will be able >>>to. But this doesn't make Othello "more simple" than chess : also in othello >>>there will be a world champion and also him will lose some games, as Kasparov or >>>Kramnik. Same for checkers, the drawing rate is high, but humans can't solve the >>>game and it is complicated for them. >> >>I rememeber that I read that there was some years go a match >>between computer and human in checkers when the computer >>won 2-1 and 67 draws. > >This doesn't show the fact that checkers is more simple than chess. It only >shows it is more drawish. >If we create a new game that is same as chess but the goal is "mate your >opponent with a bishop" or "mate your opponent in the first 30 moves" it would >be the same complexity as chess, but a lot more drawish. I think that in this case the game is more simple. Let take extreme case Assume that the target is to mate in 1 moves. all the games are drawn and the game is not interesting. I do not think that the complexity is the same as chess. I believe that even if the target is to mate in 30 moves I can draw even against kasparov and the game is simply not interesting. > >>I do not expect so many draws in a chess match. > >Neither do I. But this doesn't mean that checkers is simpler than chess. By the >way, it is full of game where the draw is impossible, but they aren't more >complex than chess. > > >>>If you wanted to say "computer simpler", then as long as we talk of "english >>>checkers" you are right. But know that international checkers (the one you call >>>10*10, but which has different rules) is computationally at the same level of >>>chess. >> >>I do not think that it is the same level as chess. >>10*10 means 50 squares when in chess >>there are 64 squares. >>The number of legal moves in 10*10 is still bigger >>in chess and the number of different >>pieces is bigger in chess(imagine chess when >>you have only queens and pawns). > >You should consider the largeness of the game tree in order to compute >complexity. So you shouldn't forget the average game lenght, the easyness of >cutting the branches and so on. >Not to mention that a program that sees 16 ply in chess is a lot stronger, from >a human point of view of a checkersprog which sees the same number. a program that see 16 plies in middlegame is also stronger than a program that can see 16 plies in the endgame but the endgame is more simple than the middlegame. As Marion >Tinsley said : chess is like looking over an endless ocean, checkers is like >looking inside a bottomless pit. >I can tell you that the average complexity is the same. > >>>Yes, all this refers to "computationally simple". However reasoning this way, Go >>>should be the most complicated game by a long distance. >> >>It is not clear. >>The number of different pieces in go is smaller >>than chess so in some meaning it is more simple than >>chess. > >So, if I create a game with 100 different pieces, but which can be won in just 2 >forced moves, that is the more complex ? I only say that this is a factor to consider and it is not clear what is more complex. >For the computational complexity you should look the game tree. >In chess you can choose among an average of 40/70 different possible moves >(maybe Dr.Hyatt or someone else can be more precise) for an average lenght of >the game of 50/80 moves. In go you can choose among about 150 moves for an >average lenght of 80/120 moves. >Of course this numbers are rough extimates, but I put them "on the chess side" >to show that without doubt go is computationally more complex. >But this is only to say that from an human point of view, every game which is >beyond our complete understanding is complex in the same way. > >>>>I guess that most of the intelligent people simply choose >>>>chess and not checkers. >>> >>>This statement is just too stupid to need an answer. Reading this I was going to >>>consider you a troll, but then I saw you post a good number of other meaningful >>>messages, so maybe you have just slipped. >> >> >>When I was young I heard from a chess trainer >>that he considers checkers to be too simple. > >I heard this statement a lot of times. And this is also a reason why checkers >have less players than chess. People consider it a "simple-game", a "child-game" >and never suspect of its deepness. > >>At that time I got the impression that checkers is >>not a game that people take seriously and only >>a game that childs play. > >This is common idea. And it's wrong. A lot of common ideas are just wrong. > >>I also read in a book of vladimir liberzon >>about chess that in checker there was a new player >>who became grandmaster after 2 years of playing when >>something similiar never happens in chess. > >It all depends on what you classify as a grandmaster, how many games he played >in that period and so on. It's full of chess grandmaster at the age of 14 or >less in this period, while there aren't in checkers, but this doesn't mean >anything. > >><snipped> >>>In international checkers a strong player (one of the first 10) convincigly won >>>against the program Buggy, winning also some games at 5 mins !! >> >>How much time did people devote to >>programming international checkers? > >A lot less than to chess. But best international checkers program have the same >effort of commercial chess programs. However, assume that Buggy had the same >strenght of a strong amateur chess engine : how many people do you know that can >beat it at 5 mins time ? The problem is that I suspect that there was not enough effort to investigate a good algorithm for checkers because it is considered as too simple so most people know almost nothing about it. The top amateurs in chess are based on ideas of other people that are good for chess (for example null move pruning) and I believe that with the right ideas for checkers humans are not going to be able to beat programs at 5 minutes per game. Note that I do not know if null move pruning is used in checkers and if it is used then it was not a good example. 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.