Author: Vincent Lejeune
Date: 13:54:11 05/21/03
Go up one level in this thread
On May 21, 2003 at 16:36:05, Uri Blass wrote: >On May 21, 2003 at 15:42:17, Tom Kerrigan wrote: > >>On May 21, 2003 at 12:40:42, Uri Blass wrote: >> >>>On May 21, 2003 at 11:58:55, Tom Kerrigan wrote: >>> >>>>On May 21, 2003 at 09:07:24, martin fierz wrote: >>>> >>>>>On May 21, 2003 at 04:29:31, Tom Kerrigan wrote: >>>>> >>>>>>On May 21, 2003 at 00:00:11, Russell Reagan wrote: >>>>>> >>>>>>>As far as I know, computer chess is the most popular computer board game. There >>>>>>>are dozens of commercial programs, and hundreds of amateur programs. >>>>>>> >>>>>>>What do you think are the main contributing factors to computer chess being the >>>>>>>most popular? Why not checkers/draughts, or go, or any other game? >>>>>>> >>>>>>>I think there are two main factors. >>>>>>> >>>>>>>1. Chess is one of the most popular games in the world, if not THE most popular. >>>>>>> >>>>>>>2. The computer chess community has created a good environment that is inviting >>>>>>>to potential computer chess programmers. >>>>>>> >>>>>>>One example of #2 can be seen in the nice GUI's that we amateur computer chess >>>>>>>programmers have. Between Winboard and Arena, we have nice looking interfaces, >>>>>>>and networking support for internet chess servers, and we get it all for free. >>>>>>> >>>>>>>I also think that chess servers such as ICC and FICS do a great deal to attract >>>>>>>potential programmers to chess instead of other board games. >>>>>>> >>>>>>>For instance, I have tried to make other game playing programs in the past >>>>>>>(checkers and amazons), but after a while my program slaughters me, and there >>>>>>>isn't anyone to play my program. If I had a nice GUI like Arena that would let >>>>>>>me organize a tournament with a dozen other amazons programs, then I would >>>>>>>maintain my interest in computer amazons. Or if there was an "Internet Amazons >>>>>>>Server", and I had a GUI that would handle the networking for me, I would be >>>>>>>much more motivated to continue programming amazons. >>>>>>> >>>>>>>So what do you think contributes to the popularity of computer chess? >>>>>> >>>>>>1. It seems the game itself might be the west's most popular board game, and >>>>>>computers are more widespread in the west. >>>>>> >>>>>>2. The rules and complexity of the game are difficult but not prohibitive, so >>>>>>it's not too simple to be uninteresting with current compute power (as is >>>>>>Connect 4, and arguably checkers) >>>>> >>>>>connect 4 can be solved on a fast pc in 1 hour. checkers is nowhere near being >>>> >>>>Really? I didn't know that. Not by "brute force," i.e., only scoring positions >>>>as win, lose, or draw. I believe that takes closer to a month. >>> >>>I read that this game was solved even without computer programs. >> >>I don't think so. It was solved pretty much simultaneously by two different >>approaches, but both did involve computers searching. > >You may be right > >I read that it was solved many years ago and here is a link for an article but I >admit that I did not read most of the article but only got the impression that >it was solved by strategic rules without a computer program. > >http://citeseer.nj.nec.com/cache/papers/cs/3978/ftp:zSzzSzftp.cs.vu.nlzSzpubzSzvictorzSzconnect4.pdf/allis88knowledgebased.pdf Here you can download "Velena", a free program who play perfect connect4 with some doc >> >>>I never tried to write a program for that game but how do you get the estimate >>>of one month? >> >>On my Athlon/800 it took about 30 minutes for my program to solve the position >>where the center column had 4 checkers in it. Let's see, figure that 15 min is >>average for a 4 checker position, >> >>7^4 = 2401, * 15 = 36015 minutes = 25.01 days >>I didn't do any symmetry stuff, so / 2 = 12.5 days >>Then figure that you can get a 2.2GHz Athlon now, / 2.75 = 4.54 days >>(Also, my program uses 64-bit ints a lot, so it might go, say, 25% faster with >>an Opteron/Athlon 64.) >> >>So you're right, a month is pessimistic nowadays but it's still way more than an >>hour. >> >>>Even without knowing the solution I believe that programs >>>can solve it relatively fast if you make the following asumptions: >>> >>>1)I assume that hash tables are used. >> >>Yes, my program did. Otherwise it craps out at ~16 ply. >> >>>2)I assume that symmetric is also used and the program consider 2 positions >>>as the same if they are symmetric even if they are not the same. >> >>Nope, didn't do this. I guess this could be done by computing two hash keys. >>Lowers NPS a little, but I imagine it's still a big win. >> >>>3)I also assume that programs also can evaluate >>>positions when all the moves are forced correctly >>>as win,draw,loss >> >>This seems like a good idea but actually doesn't help any. > >I think that it can help because you need less plies to prove that white wins >and you will have less positions that you will need to store in the hash tables. > >This kind of singular extension in evaluation means that you can see that one >side won even some plies before he practically won the game. > >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.