Re: [computer-go] MoGo
Oooh, another of my favourite topics. I realized pretty early on, and haven't seen any counter-evidence over the past decade of study: go skill is transitive to almost all board sizes, and that is why 9x9 computer go is so important. (IMHO :-) Call me picky if you want, but I spend a lot of time processing go knowledge and none of it is board size independent. ... but I constantly repeat the idea that in a two day professional game the first day must be dedicated to the first 50 moves. ... My feeling is they spend so much effort in the opening (trying to get a 1 or 2 pt advantage) because they have nothing better to do with the time. The gap between a professional and, say, a 1-dan amateur is all to do with tesuji knowledge, life/death knowledge, and (to a lesser extent) tactical reading skill, accurate endgame counting and joseki knowledge. All except joseki-knowledge is board-size independent. And the same applies even between top professionals: look at how many really top-class players claim their endgame is the strongest part of the game. Darren -- Darren Cook http://dcook.org/mlsn/ (English-Japanese-German-Chinese free dictionary) http://dcook.org/work/ (About me and my work) http://dcook.org/work/charts/ (My flash charting demos) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re:[computer-go] MoGo
Sylvain Gelly wrote: You should also know that we never claimed that MoGo plays 9x9 go near the level of a professional go player, . . . Just curious: Do 9x9 professionals exist? When we say professional we mean 19x19 professional. Of course, there must be a correlation. One expects an Olympic final level 100 meter runner to run 400 meters much faster than any of us, but not faster than a 400 meter runner. Call me picky if you want, but I spend a lot of time processing go knowledge and none of it is board size independent. The relative value of the corners, the sides and the center are too different. Josekis do not work at different sizes either. I am repeating myself, but I constantly repeat the idea that in a two day professional game the first day must be dedicated to the first 50 moves. The precise value of these moves cannot be determined by search as in the end of the game. Professional players use study and experience. And that is board size dependent. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On 4/5/07, Sylvain Gelly [EMAIL PROTECTED] wrote: Sylvain, could you run the same test on 7x7 to verify that there the 'correct' komi would be 9 (try 8.5 vs 9.5)? If MoGo wouldn't converge to 9 we probably shouldn't have much confidence in the generalisation of the above results for higher levels of play on 9x9 (or you could be on your way to discover an error in the human solution for 7x7 :-). Either way I think it would be interesting to know what comes out... Ok, here are the results (again MoGo_3k against itself): 6.5 3399/5200 65% 7.5 2397/4800 50% 8.5 2603/5200 50% 9.5 1848/5200 35% That looks quite good! Assuming no seki's (with neutral intersections), which IIRC is in agreement with the human solution, the winner on 7x7 follows from: Black intersections (7*7 + komi)/2 so for 7.5 komi black needs to get at least 29 intersections, and the same holds for 8.5 komi. Consequently, your statistics confirm 9 komi for jigo. Thanks, Erik ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
Don Dailey wrote (about big/small wins) It actually surprises me that go players care about this ... One difference with chess is that you don't play chess after the game is over. The comparison could be: the king is captured, the loser keeps playing and then the winner gives the queen for nothing. Bad moves hurt! And we get to the really important part of a any program: The user. Many users feel stolen by UCT programs. I have read that in the KGS chatrooms. Normal users do not count with +/- 0.5 point precision. They have the impression the program blundered and they caught up. But when the program counts, surprise!, it wins by 0.5 points Chinese. The users were thinking Japanese even if they accepted Chinese rules. In fact, they did not have the choice. They get the impression the game was stolen by technicalities after they saw the engine blunder many times. Of course, I know there is a good reason for how UCT works. And improving style is much less important than improving strength. But many players don't want to adapt to computer settings. They want computers to win the game as they have always played it. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
Many users feel stolen by UCT programs. I have read that in the KGS chatrooms. Normal users do not count with +/- 0.5 point precision. They have the impression the program blundered and they caught up. But when the program counts, surprise!, it wins by 0.5 points Chinese. If the program blundered as you said and still wins, it means that the program already won much earlier in the game. It is not a matter of chinese or japanese rules, and it is not a steal. Also, as you speak about KGS chatrooms, this behavior has been explained many times by developers, and now I feel many people know that, and when someone is surprised, there is almost always someone explaining why. The informations in the program profile are also here for that. Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 10:49 +0100, Jacques Basaldúa wrote: Don Dailey wrote (about big/small wins) It actually surprises me that go players care about this ... One difference with chess is that you don't play chess after the game is over. The comparison could be: the king is captured, the loser keeps playing and then the winner gives the queen for nothing. Bad moves hurt! And we get to the really important part of a any program: The user. Many people DO play chess after the game is over. They continue to play a game long after they could have resigned.The only difference is that GO is easier. You can easily lose a dead won game in chess but in GO there is point where it is played out and it's very hard to lose unless you are doing it on purpose. Many users feel stolen by UCT programs. I have read that in the KGS chatrooms. Normal users do not count with +/- 0.5 point precision. They have the impression the program blundered and they caught up. This is human nature and ego. I have seen it many times in computer chess - every loss to a computer it seems was some kind of irregularity in the minds of the loser. But when the program counts, surprise!, it wins by 0.5 points Chinese. The users were thinking Japanese even if they accepted Chinese rules. In fact, they did not have the choice. They get the impression the game was stolen by technicalities after they saw the engine blunder many times. How does Japanese make any difference? I would think that a Japanese rule player would appreciate UCT style of play more than Chinese players? My impression of go players has changed a little as a result of this discussion. There has always been this idea (usually when Japanese/Chinese rules comes up) that it's more about the hidden truths, not playing out the game to get the score. But it seems it's not that at all, even for strong players if they think it's odd to protect the win more than try to win big. I think this may be an ego thing too. It's probably humiliating to win by 1 point against a much weaker player, even if you have the win in the bag very early on. Of course, I know there is a good reason for how UCT works. And improving style is much less important than improving strength. But many players don't want to adapt to computer settings. They want computers to win the game as they have always played it. They may want this, but it's a bit selfish. After all, does anyone force THEM to change their playing style to suit someone else? I know how human nature works. I don't know GO culture, but I know chess culture and there are many people in chess who are immersed in the cultural trappings of chess, not so much the game although they like to pretend.They get real annoyed if you don't play their kind of game. It's more a cooperative dance than a battle for them. Ever play tennis recreationally? Same thing but worse. Many players will not play a game of tennis with you if you don't play in a way that is comfortable for them. They want to play someone that will always hit to their strong side with comfortable knee level strokes. To make a long story short they want to play someone who makes their own game look pretty. They want you to cooperate with them (even if they lose they want it to be their kind of game.) If I'm playing them and I happen to be the better player, I usually just cooperate with them to make them feel like they are playing the game the way it's supposed to be played. - Don Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] 12th Computer Olympiad
Dear Go programmers, The ICGA has concluded the negotiation for organizing the WCCC 2007, the 12th Computer Olympiad, and an accompanying scientific workshop . The events will take place in Amsterdam, The Netherlands, 11-18, June 2007. The workshop will be held on Friday 15. - Sunday 16. June 2007. The Call for papers can be found on the website of the ICGA: http://www.icga.org/ An excursion is scheduled for all interested participants. A detailed schedule will be announced as soon as possible. The Computer Olympiad will include a 9x9 and a 19x19 Go tournament. More information will be available soon on the tournaments page: http://tournaments.icga.org/ Best Regards, Guillaume winmail.dat___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 12th Computer Olympiad
On 4/5/07, Chaslot G (MICC) [EMAIL PROTECTED] wrote: The workshop will be held on Friday 15. - Sunday 16. June 2007. Must be a leap Saturday... regards, -John ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
In message [EMAIL PROTECTED], Don Dailey [EMAIL PROTECTED] writes On Thu, 2007-04-05 at 10:49 +0100, Jacques Basaldúa wrote: But when the program counts, surprise!, it wins by 0.5 points Chinese. The users were thinking Japanese even if they accepted Chinese rules. In fact, they did not have the choice. They get the impression the game was stolen by technicalities after they saw the engine blunder many times. How does Japanese make any difference? I would think that a Japanese rule player would appreciate UCT style of play more than Chinese players? People are aware that Japanese and Chinese scoring can sometimes differ by a point. So when a player loses to a program by half a point, he can try claiming that he would have won if the other scoring method had been used. I like to think that MoGo deliberately beats such people by half a point, so as to annoy them more :-) Nick -- Nick Wedd[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
I like to think that MoGo deliberately beats such people by half a point, so as to annoy them more :-) this isn't uncommon in teaching games -- the idea (i think) is to give the student opportunities to make good moves, providing them with opportunities to learn through good play, rather than winning by the maximum margin possible, where they may not even be able to understand their own bad moves or their opponent's good moves. of course, making the realization that you're playing against someone who can win by 0.5 points at will is pretty frightening. s. Looking for earth-friendly autos? Browse Top Cars by Green Rating at Yahoo! Autos' Green Center. http://autos.yahoo.com/green_center/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 09:17 -0400, Don Dailey wrote: How does Japanese make any difference? Because the vast majority of games use Japanese rules on KGS, I think many players do not notice if they are playing Chinese rules. If they then find out that dame is worth 1 point, they may feel cheated if the opponent plays dame while they pass. I always play by Chinese rules on KGS and I can verify that many players aren't aware that they are playing by Chinese rules, because in my games they often pass when they could be playing dame (and I doubt they counted odd vs even). In that case I pass too; I wouldn't want to take a bunch of points while they kept on passing. Some players who play by Chinese rules will take these points, and so people have become somewhat aware of this issue. I've seen this discussed as a kind of cheating, so perhaps people are thinking of this when they lose by half a point? There is another reason for the negative reaction with regard to monte carlo endgame play -- it is completely unhuman and unaesthetical. It is natural to make safer plays when ahead, but the monte carlo plays are *so* ultra-safe as to look ugly. They are plays no human except an absolute beginner might make. So I think the reaction by humans is to be expected. Obviously playing strength is the most important thing, though if play could appear more reasonable without impacting strength and without too much effort then that would be best. -Jeff ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] professional knowledge
The gap between a professional and, say, a 1-dan amateur is all to do with tesuji knowledge, life/death knowledge, and (to a lesser extent) tactical reading skill, accurate endgame counting and joseki knowledge. All except joseki-knowledge is board-size independent. There's also what you might call the meta-knowledge of joseki: Which joseki will accomplish worthwhile goals for me in this position vs which joseki might (while giving perfectly balanced results in a generic corner) entirely fail to make the best of this situation? That is a separate type of knowledge, as in the proverb Learn joseki, lose ten ranks in strength. The chief difference between a 9X9 game and a 19X19 is in the demands the larger board makes on our _strategic_ reading ability. And that is not merely another board-size-dependent skill, among many. That is the most significant difference between a competent player and a strong one. Forrest Curo San Diego ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
I like to think that MoGo deliberately beats such people by half a point, so as to annoy them more :-) Sylvain, I think it would be quite humorous if you could tune KGS MoGo to do exactly that without hurting its win rate too much. Perhaps one way would be to evaluate playouts normally near the beginning of the game and gradually narrow the requirements so that by the end of the game, only a 0.5 point win is considered a win. I don't know if that would really work or not. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 10:18 -0400, Jeff Nowakowski wrote: There is another reason for the negative reaction with regard to monte carlo endgame play -- it is completely unhuman and unaesthetical. It is natural to make safer plays when ahead, but the monte carlo plays are *so* ultra-safe as to look ugly. They are plays no human except an absolute beginner might make. So I think the reaction by humans is to be expected. Of course. If it were easy to fix, I would want my program playing the more natural moves. Obviously playing strength is the most important thing, though if play could appear more reasonable without impacting strength and without too much effort then that would be best. Yes, I agree. As I mentioned earlier it affects our perception of how strong a program is. Many people see these endings and conclude the programs are pretty weak and really get surprised. Someone once accused my older program Botnoid of just making random moves and they saw that it was winning a lot more games that a random player would win and they couldn't believe it.This showed that they didn't really understand what they were looking at as a whole. It's the human way, like many of the older go programs the emphasis is mostly on local tactics, as if it has nothing whatsoever to do with the big picture.So humans have the same failings as computers, just not as pronounced.Maybe that's why UCT is getting so good? It no longer has that local tactics is everything sense of the game and so at least in this regard these programs are surpassing many humans. Wouldn't that be a hoot? To get better at GO you will be taught to see the big picture - think more like a computer!:-) - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] April KGS Computer Go tournament
In message [EMAIL PROTECTED], Remi [EMAIL PROTECTED] writes Hi Nick, I'd like to register CrazyStone for formal, and StoneCrazy for open. Also, could you please confirm the date of the tournaments ? KGS still says this week-end, but there were discussions of postponing them. Thank you for these entries. CrazyStone is now registered for the Formal division (9x9), and StoneCrazy for the Open (13x13). This tournament will take place this Sunday, Easter Sunday, April 8th. The May KGS tournament has been postponed, from Sunday May 6th to Sunday May 13th. http://www.weddslist.com/kgs/future.html should give the correct dates, though it currently says Sunday May 15th - I shall change it now. On Monday I posted a statement about this postponement in which I mixed up various months. I apologise for the confusion this caused. Nick -- Nick Wedd[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 14:42 +0100, Nick Wedd wrote: I like to think that MoGo deliberately beats such people by half a point, so as to annoy them more :-) I like that! I think I will program Lazarus to have to goal of wining by EXACTLY 0.5 points! If it looks like it will win big, it will fill it's eyes and try to help the opponent catch up! - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 11:08 -0400, Chris Fant wrote: Sylvain, I think it would be quite humorous if you could tune KGS MoGo to do exactly that without hurting its win rate too much. Perhaps one way would be to evaluate playouts normally near the beginning of the game and gradually narrow the requirements so that by the end of the game, only a 0.5 point win is considered a win. I don't know if that would really work or not. That would be hard because you cannot expect your opponent to cooperate. It would be pretty much impossible to force the opponent into a 1/2 point loss. I'm pretty sure that anything drastic in this regard would weaken the program. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
That would be hard because you cannot expect your opponent to cooperate. It would be pretty much impossible to force the opponent into a 1/2 point loss. I'm pretty sure that anything drastic in this regard would weaken the program. Didn't you just say you were going to try to make Lazarus do the same thing? Maybe you were just joking. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
That would be hard because you cannot expect your opponent to cooperate. It would be pretty much impossible to force the opponent into a 1/2 point loss. I'm pretty sure that anything drastic in this regard would weaken the program. Didn't you just say you were going to try to make Lazarus do the same thing? Maybe you were just joking. Or maybe it was 1 hour before, and then realise that it is difficult. I also think it is quite difficult, because in the tree the opponent level will try to be far from 0.5 and give you points to make you miss your goal... Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
Or maybe it was 1 hour before, and then realise that it is difficult. I also think it is quite difficult, because in the tree the opponent level will try to be far from 0.5 and give you points to make you miss your goal... Ahh, true. The opponent levels would need goal=win while the self levels would need goal=(win by 0.5). That complicates the search too much to continue to be amusing. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo
On Thu, 2007-04-05 at 12:46 -0400, Chris Fant wrote: Didn't you just say you were going to try to make Lazarus do the same thing? Maybe you were just joking. Of course I was joking, but it made me think about how it would be done. I think it would be really difficult to do this without weakening the program. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] MoGo
Every go book says that to get better you need to see the big picture :) The big difference between low kyu and high dan players is seeing the big picture. Low kyu players are already pretty good at local tactics. If you read commentaries you will see a lot of waords about direction of play, which is a big picture concept, and not so much about tactics. I agree with you that a big strength of UCT is its ability to see the big picture. Older go programs were stronger at local tactics than sam-strength people, and weaker at big picture. UCT seems to be stronger at big picture and weaker at tactics. David Wouldn't that be a hoot? To get better at GO you will be taught to see the big picture - think more like a computer!:-) - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: Re:[computer-go] MoGo strength
I tried a few games against Mogo 9x9 on KGS, and it's not professional strength, but it is very strong. When I played fast when I was tired it beat me every time, and when I made a more careful try, I beat it, but it wasn't easy. I'm AGA 3 Dan, KGS 2 kyu, so it seems to be about my strength or a little stronger. It's very strange to usually lose by 0.5 point, but win by capturing every stone. I don't think I could win even one game in 100 against a professional player at this time limit. David -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Sylvain Gelly Sent: Wednesday, April 04, 2007 1:52 AM To: terry mcintyre; computer-go Subject: Re: Re:[computer-go] MoGo Hello Terry, Sylvain, Were you aware of this challenge from the American Go Association? The following is from the latest AGA newsletter; you can send corrections or replies to [EMAIL PROTECTED] Yes I was aware, Roy Laird asked me to put MoGo on KGS to play these games, and it is what I did. Of course as experts, you should have noticed errors on this newsletter, as e.g. MoGo developed by the inventors of UCT in hungary :-). You should also know that we never claimed that MoGo plays 9x9 go near the level of a professional go player, which is of course false, and even if it was true should ask for many many experiments, and we would have never say that. These games are in some extends interesting, especially for part of you who can understand them. I can't, so the best I can do is count, and there are so few to count ;-). To answer peigang, the 10 minutes is here because players on KGS usually don't want to wait, and want not too slow games. For the komi, it may be right that MoGo wins more often with white than black, I don't know. Sylvain GO ONLINE: MoGo -- No-Go, So-So or Uh-Oh? Go has been called The fruit fly of IT, and for a good reason -- although software engineers have created programs that can defeat the strongest chess players, the strongest go programs are routinely defeated by talented children. In fact, go is the lone holdout, the only classic game that has not yet been solved (so to speak) for the computer. If you wonder why, the Wikipedia article on computer go http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL2VuLndpa2lwZWRpYS5v cmcvd2lraS9Db21wdXRlcl9nbw%3D%3D is a good place to start. One way to simplify the problem is to work with a smaller board, an approach followed by Levente Kocsis and Csaba Szepesvari, who are working together at the Hungarian Academy of Science on a 9x9 program called MoGo. Their recent claim that MoGo plays 9x9 go near the level of a professional go player made international news http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL2FiY25ld3MuZ28uY29t L1VTL3dpcmVTdG9yeT9pZD0yODkxNjc5 so we decided to investigate. Sylvain Gelly, a contributor to the program, clarified the one-armed bandit strategy, a variation of the ancient Chinese proverb Rich men don't pick fights. Gelly told the EJ that MoGo tries to maximize its winning probability. When behind, MoGo will play 'strange' moves to try to catch up, and when ahead, it will prefer safe moves which secure victory instead of keeping score. Usually it loses points when ahead, trading profit for safety, aiming to win by +0.5. To learn more see the Sensei's Library MoGo http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL3NlbnNlaXMueG1wLm5l dC8%2FTW9Hbw%3D%3D page. MoGo has played extensively http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL3d3dy5nb2tncy5jb20v Z2FtZUFyY2hpdmVzLmpzcD91c2VyPW1vZ29ib3QmYW1wO3k9MjAwNyZhbXA7bT0x on the Internet but evidence that it plays beyond the mid-kyu level is not compelling, so we're going to put MoGo to the test. Philip Waldron -- a solid 6-dan with a current AGA rating of 6.47 who has reviewed go software for the EJ -- will play a best-of-seven series against MoGo in the computer go room on KGS. Game times will not be announced in advance, and times will vary to eliminate the possibility of human interference on the MoGo side. The results, and possibly a few of the games, will appear in a future EJ, so stay tuned! - Roy Laird _ It's here! Your new message! Get new email alerts http://us.rd.yahoo.com/evt=49938/*http://tools.search.yahoo.com/toolbar/fea tures/mail/ with the free Yahoo! Toolbar. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] MoGo
Of course my comment was tongue in cheek, but I agree with you. UCT programs are not as good at tactics as they are in understanding the game as a whole. I'm really not a good player, not qualified to speak about this, but I have an impression about how they manage to beat program that are tactically better and I would like to get your impression as well as others: Simply put, UCT seeks a pathway that keeps it out of trouble. They don't have to fully understand all tactics to have enough sense to simply go a different way. But they are still good enough that they won't turn down a good fight if the position calls for it. They play go the way I used to win in dodge ball when I was a kid. I was not the most athletic, but I just stayed alert and avoided the battle until almost everyone had knocked each other out. Usually who was left was a few geeks like myself and it was easy to take them out (because they were geeks!) I don't think I was actually smarter than anyone else, I just though it was more fun to see if I could win. To them, running TOWARD the ball was the way to go because that's where all the fun was! That's how I think UCT program play - the same strategy - avoid the fights you might not win. Apparently they have added the strategy to also annoy the opponent by making him think it's still close! :-) - Don On Thu, 2007-04-05 at 11:07 -0700, David Fotland wrote: Every go book says that to get better you need to see the big picture :) The big difference between low kyu and high dan players is seeing the big picture. Low kyu players are already pretty good at local tactics. If you read commentaries you will see a lot of waords about direction of play, which is a big picture concept, and not so much about tactics. I agree with you that a big strength of UCT is its ability to see the big picture. Older go programs were stronger at local tactics than sam-strength people, and weaker at big picture. UCT seems to be stronger at big picture and weaker at tactics. David Wouldn't that be a hoot? To get better at GO you will be taught to see the big picture - think more like a computer!:-) - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Turing test
I don't play go, so apply whatever discount seems appropriate. Go is a zero sum game - except when humans are involved. People are clearly dealing with a multi-criteria optimization task. Losses can be moral victories; wins can be humiliating; style and tradition matter. Virtually every KGS tournament has had a case where the author of a bot has said I don't want my bot to win this game under these circumstances! Go, as played by most people, can *almost* be described algorithmically. There are variants of go that can be described algorithmically. And people can be persuaded to play that way, albeit with much grumbling and quibbling. This can lead some to assert that anything else doesn't exist. (Due to the influence of Internet servers, maybe some day that will be true.) I won't generalize to everyone, but my own work is focused on a mathematical abstraction of the game. When I'm ignoring 3000 years of tradition, should I still call the game I'm modeling go? When a bot racks up a lot of wins against human players, through what they see as gimmicks like winning by 1/2 point, you can explain that 2 ways. In a sense, the humans are irrational, they play with pride and try to salvage their dignity or grandstand: they are making emotional mistakes. Or... the bot is maximizing a single criteria (winning percentage) at the expense of everything else. It gets an artificially high rating by playing churlishly. The friend, who first suckered me into trying my hand at computer go, once told me: a game of go is like a conversation. I think that has meaning on many levels. His immediate point was that playing a computer program felt like arguing with a chat-bot. Perhaps, at the current level of play, it doesn't matter. But when one of the engines reaches shodan at 19x19 (not so far away, I think) , I wonder if it should try to play the way people do. Or if maybe we should choose another name for the game it's playing that doesn't have so much history. Do strong chess programs pass turing tests? Should they? - Dave Hillis Check Out the new free AIM(R) Mail -- 2 GB of storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] professional knowledge
The chief difference between a 9X9 game and a 19X19 is in the demands the larger board makes on our _strategic_ reading ability. Agreed. And that is not merely another board-size-dependent skill, among many. That is the most significant difference between a competent player and a strong one. Disagreed, sorry. As I said, I think go skill applies to almost all board sizes. If you go to a go club, try holding a no-handicap tournament at a weird board size (e.g. 9x9, 13x13, 17x12, whatever). If the difference between competent players was all about 19x19 strategic knowledge you'd expect everyone stronger than about 10 kyu to tie for first place. But what you will find is the ordering will be very close to 19x19 go rank. (You can also see this on go servers that maintain a different rank for each board size.) Darren -- Darren Cook http://dcook.org/mlsn/ (English-Japanese-German-Chinese free dictionary) http://dcook.org/work/ (About me and my work) http://dcook.org/work/charts/ (My flash charting demos) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] professional knowledge
On Fri, Apr 06, 2007 at 07:52:34AM +0900, Darren Cook wrote: The chief difference between a 9X9 game and a 19X19 is in the demands the larger board makes on our _strategic_ reading ability. Agreed. And that is not merely another board-size-dependent skill, among many. That is the most significant difference between a competent player and a strong one. Disagreed, sorry. As I said, I think go skill applies to almost all board sizes. If you go to a go club, try holding a no-handicap tournament at a weird board size (e.g. 9x9, 13x13, 17x12, whatever). If the difference between competent players was all about 19x19 strategic knowledge you'd expect everyone stronger than about 10 kyu to tie for first place. But what you will find is the ordering will be very close to 19x19 go rank. (You can also see this on go servers that maintain a different rank for each board size.) I see two ways that something can be board-size-dependent. In the first way, the thing vanishes for small boards and grows progressively more important for large boards. I don't remember the original message, but from the quoted snippet above, I'd guess that this is what was meant by the person that you quoted but didn't attribute. That is, the importance of strategic reading ability essentially vanishes on a 4x4 board, but grows to be significant on a 19x19 board. Similarly, most kinds of endgame skill essentially vanish on a 4x4 board: it isn't very important to know the difference between sente and gote when there's only one meaningful game on the board. Thus, a large part of the difference between competent players is about strategic reading ability which vanishes on tiny boards but grows important on large boards. Strong players tend to be very clever about when they drop one fight and play to affect another, or when they steer fights into each other. No matter how much insight they have about such things, though, it doesn't help them much on a 4x4 board. Another way that something can be board-size-dependent, the way that you seem to be disagreeing with, is when what you learn on an NxN board hardly applies to an N'xN' board, even when N' is approximately equal to N. That seems strongly true of most Chess opening knowledge, for example. It's probably somewhat true of most Go opening knowledge, too: at least the acceptable joseki tradeoffs between territory and influence seem likely to change significantly. But as I said, I doubt that is what the original unnamed poster was talking about. -- William Harold Newman [EMAIL PROTECTED] PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C Ubi saeva indignatio ulterius cor lacerare nequit. -- Jonathan Swift's epitaph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] professional knowledge
Similarly, most kinds of endgame skill essentially vanish on a 4x4 board: Yes, my thesis crumbles on the tiny boards: I think 9x9 is the smallest board size where 19x19 playing strength is very significant. (Endgame skill is important at 9x9: I've found games where a mid-dan player has lost to a high-dan player by making a 1pt mistake in the last 10 moves.) influence seem likely to change significantly. But as I said, I doubt that is what the original unnamed poster was talking about. If you missed the start of this thread, the first two messages were in the Mogo thread: a post by Jacques Basaldua (asking what is meant by a 9x9 professional), and a reply by me. Darren ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Turing test
At 02:47 PM 4/5/2007, you wrote: I don't play go, so apply whatever discount seems appropriate But when one of the engines reaches shodan at 19x19 (not so far away, I think) ... probably still *very* far away. the best programs are rated at about 10-kyu. 10 stones is a *long* way. it's non-linear thanks --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
