Author: Dave Gomboc
Date: 02:19:17 09/10/02
Go up one level in this thread
On September 10, 2002 at 03:10:43, martin fierz wrote: >duh again! my statement, conveniently including the next sentence is: >"In my opinion, the title which Nemesis won is worth more than the title Chinook >currently holds. The challenge should actually be the other way round!" > >show me "belittle" in this. as ingo says: doing it first is much harder than >doing it better afterwards. i don't mean to say "programming nemesis was more of >an accomplishment than programming chinook". i think the opposite. but i do mean >to say "nemesis plays checkers better than chinook did in 1994". >besides, my report is free... nobody is stuffing it down your throat... If you can't see the belittlement in there, then you have less intellect then I give you credit for. I'm not going to argue about it further. It's pretty obvious to several graduate students of a 20+-person games group here at the U of A. And no, Jonathan isn't my supervisor. I do think that assuming Nemesis plays checkers better than Chinook is a mistake, though. (It might be true, but I wouldn't assume it.) In such a drawish game, two really good players are tough to rank one way or the other at all. >i'm happy to say that it is not in my throat: i am using my own 8-piece db and >my own access code. my database is smaller and my access code is faster. and >again, this is not belittling their achievement... i had their paper to start >with. i had a computer with lots of ram. i had their 6-piece database to compare >with to verify my 6-piece database and make sure that my generator was working >properly. they did something i never could have done. >and if you never *really* looked at that code, don't defend it just because >you're at the same university as schaeffer... he didn't even write it himself. >crap is crap is crap, and i say it when i see it. you would say the same *if* >you took the trouble to look at it. I took the time to look at it over a week before I posted here about it. Again, the state of the code is irrelevant. There is documentation of the database format provided -- so you don't have to even use that code to access the databases. You can write your own access code. Or, you can use the (old) code and verify your results against the results of another program. Jonathan has caught errors in his database because he specifically verifies them. There's no duty for Jonathan to update his web code just because he decided out of the goodness of his heart to make his eight-piece database available. I'm sure it didn't occur to him that the code would have some arcane overflow. It's unfortunate that the problem happened, but dumping on Jonathan isn't going to change it. >>That does sound rather odd, but you know, when I read your tournament report I >>was surprised because it sounded like the operators picked all the openings as >>the round came up, I mean, like including the first move(s?) that the opponent >>had to play. What's up with that? >standard checkers practice - the first 3 moves are chosen at random. here, to >increase the chances of a win, we agreed to let everybody choose openings where >he thought he "had something". It seems to me that it'd be best if it was the program itself doing that sort of thing (which I guess would be the same as you guys choosing the 3 moves for all the matches before the tournament started, as opposed to while it was in progress). You guys must be really desperate to avoid draws! >hmm, but you're a chess player, right? happens in every tournament you play: in >the last round, you get black and a strong opponent, and your friend gets white >and a 1600 player and ends up in front of you, and it wasn't your fault. or vice >versa. only if you are really better than your friend, you will end up in front >of him. >besides, if nemesis finds and knows about this line, and kingsrow does not, then >nemesis is the better program. and that's how it was. kingsrow would have lost >the same game cake lost. i could argue that the format is wrong, since murray >could choose to play this line against kingsrow or against cake. had he chosen >kingsrow, cake would have finished in second place. but that is not the point. >the tournament is supposed to find the correct winner, the rest is not really >important. it did. But that's the point, in a tournament that kind of stuff happens, but in a match it doesn't. Which is why matches are better for deciding a championship (IMO). >>(So why did Cake++ play the losing move anyway? Did the other programs see that >>it wasn't good?) >the others had the moves in their book. they were very deep traps. i think they >would have played the same bad moves that cake played without book - but i'm not >100% certain. it depends on how much time exactly they would have alloted for >this move, what parts of the db are in memory, what is in the hashtable etc. I emailed the position to Jonathan and asked him if Chinook would play the right move -- he replied that it did. I assume that the search was cold (no filled hash from previous searches). So I guess it's not that bad a program. <shrug> Dave
This page took 0.01 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.