Author: Nicolas GUIBERT
Date: 06:02:25 02/09/01
Go up one level in this thread
>How many programs are there for the 10x10 game? Jean-Bernard Alemanni has published on his site (the address is given in a message he posted) an impressive list of draughts program (for 10x10 international version mainly, but also recently for the other games). Around 100 programs, I guess. This number is surely even much higher, as students are very given the task to build a draughts-playing program as a homework ! (Often, lazy students ask for some source code). > >I suspect that the reason that there are no good program may be the fact that >people simply did not try to do programs for this game. It is clear that the number of programmers is smaller than for chess. Though, the international variation (10x10) called international draughts, is played in 50 countries (number of countries belonging to the international federation, the FMJD), we can't say that it is as much played as chess. Only in Africa, the game is clearly more popular than Chess. There are not many programmers there. The programmers often meet in the Netherlands (and sometimes in France, tomorrow we have a tournament !!). Two competitions a year is the usual frequency of these tournaments. >The game seems to me simpler than chess even in cases that the board is 10*10. >The number of legal positions is clearly smaller than 5^50. >The branching factor is also smaller. After playing draughts for more than 10 years now, and having programmed for 2 years and a hal, I still don't know if this statement is true or not. I feel that it is, but I am not sure. Clearly, the number of moves in a given position is smaller (10-13 in a middle-game position). On the other hand, the length of the game is similar to what it is for chess (50-70), maybe even a bit longer. The openings are not very important. On the contrary, the endgame is. Currently, a 6 pieces database is fully available to anyone. It is 1.5 Gb big. The strategic aspect of the game is quite hard to understand, and can sometimes be very subtle. Fortunately, most strategical knowledge can be implemented. So far, I pulled my hair many times wondering how "this damned feature" could be implemented. Most often, after days (and good night sleeps), I managed to find a reasonable solution that took into account enough aspects of the considered problem. (As we say in France, "Impossible is not French") Therefore, my conclusion is that most features that the humans know from the theory of the game can be implemented. It is often very hard, but it is possible. Patience required ! Another consideration : the game has to be split into different types of positions with completely different evaluation functions. For those who know the game, an example : a classical position (central positions with limited number of moves) is not the same at all, as an open attack position. In the former case, it is extremely important to have fewer tempos (which means more moves before being blocked), in the latter, it is important to have more tempos (this way, you have more chance to get a king first). This makes the evaluation function(s) more complex to build, but not impossible. Lastly, there are some strategic features that are very long-term ones. I don't know Chess enough, but from what I've read here so far, I guess we have things that can be compared to what you call "Passed pawns". Is it possible from these 3 white pieces against these 2 black pieces, to promote one of them to a king ? And how long will it take ? This is just an example of these strategic elements. Strategy is essential. In draughts, the computer are now strong enough to avoid most of the tactical tricks. The battle is on the field of strategy, and thus on the field of the evaluation function. At least, this is my opinion. The game is full of very nice combinations. These combinations don't happen any more nowadays. We have gone past the "tactical barrier" or "tactical boundary" . I think that now the top programs have roughly the same levels as the top 20 grandmasters. I don't know much more, since no real one-to-one matches between humans and computers has ever been played. I hope this will change soon. In blitz games, the top programs (and also the others) are surely close to being world champion, if not world champion. (IMO, they are) >I do not play this game but I read that an advantage of one pawn is usually >decisive when it is not the case for chess when the side that is pawn down can >often draw. Yes in 10x10 international draughts, that's true. And we also have too many draws at top level. The Dutch magazines and online forums often refer to the possibility of changing the rules of the game for getting fewer draws. In the Netherlands championship, 70% of the games are draws. Fortunately, in the other competitions, it is usually less. Still, we have fewer draws than in the 8x8 variations (Checkers, Spanish, Russian...) I hope these lines were somewhat interesting. Feel free to ask for more info. I'll be pleased to give an answer. Nicolas Guibert. (Program name : Buggy)
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.