Re: [computer-go] Re: Interesting problem
At 23:17 03/01/2007, Don wrote: David, I thought of another way to put it which I think, in a way, defines the difference in the rule-sets. You are playing a game, and you think the opponent group is dead. But you are not 100 percent sure. What do you do? Chinese puts the emphasis on the actual truth of the situation. Japanese makes you gamble, and penalizes you for being wrong. It makes your opinion about the situation become a factor in the final result instead of the board position and your play leading up to it. Don, I can see that chinese rules let a player try a speculative invasion inside his opponents territory at the end of the game without risk, but you seem to be saying more than this. Could you give a 5x5 example or two please? I had heard that in some sense, chinese rules require more sophisticated understanding for perfect play. It might be best to construct the example by playing a pretend game so that each player has played the fair number of stones. Thanks ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
I assume that cannot be captured by the opponent means that the opponent, playing first, cannot capture it. I accept that it is unclear whether this opponent is the actual one present in the game, or a hypothetical competent one. In an unresolved semeai it is not clear who is the one trying to capture and should thus get the first move. One more vote for simple rules. :) -- Tapani Raiko, [EMAIL PROTECTED], +358 50 5225750 http://www.cis.hut.fi/praiko/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
In message [EMAIL PROTECTED], Tapani Raiko [EMAIL PROTECTED] writes I assume that cannot be captured by the opponent means that the opponent, playing first, cannot capture it. I accept that it is unclear whether this opponent is the actual one present in the game, or a hypothetical competent one. In an unresolved semeai it is not clear who is the one trying to capture and should thus get the first move. It is fairly clear to me. You ask the players for the status of each group (alive, or dead. Alive in seki is a special case of alive). Where they agree, you accept what they say. Where they differ, you have to find out whether it can be captured, with its would-be capturer moving first. Of course, if the players do the finding out themselves, there is a danger that you end up with two adjacent dead groups. If this happens, I am not sure what to do next. One more vote for simple rules. :) Agreed. 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] Re: Interesting problem
On Thu, 2007-01-04 at 08:01 +, Tom Cooper wrote: At 23:17 03/01/2007, Don wrote: David, I thought of another way to put it which I think, in a way, defines the difference in the rule-sets. You are playing a game, and you think the opponent group is dead. But you are not 100 percent sure. What do you do? Chinese puts the emphasis on the actual truth of the situation. Japanese makes you gamble, and penalizes you for being wrong. It makes your opinion about the situation become a factor in the final result instead of the board position and your play leading up to it. Don, I can see that chinese rules let a player try a speculative invasion inside his opponents territory at the end of the game without risk, but you seem to be saying more than this. Could you give a 5x5 example or two please? I had heard that in some sense, chinese rules require more sophisticated understanding for perfect play. It might be best to construct the example by playing a pretend game so that each player has played the fair number of stones. + + O + O O # # + + + O + O O # # + + # O O O O O # # + # + O O # # # + + # O O # + # O + + + O # # + # + + + O O O # + + # + # O O # # # # + + + O O O # # # + + Here is an example from 9x9 which illustrates a key conceptual different in the rule-sets. I admit this is a rather trivial example but it illustrates what I need to say. In the diagram, black has a chance to make a live group but only if white plays stupidly. Although this is a trivial example, we might imagine a much more interesting example where it's not so clear, or where the better player has a real chance to make this group live. In such a situation, Japanese is more about gambling skill, can I get away with it? The strong Japanese player is inhibited for trying to take advantage of his extra skill. The Chinese player can apply his skill to such a position without being penalized if the opponent is able to defend. Now imagine that diagram is played out more, so that there are no chances to save groups - there is a point in any game, where the game is conceptually over and a strong player can compute what the exact score should be using any unambiguous rule-set. With Chinese rules, when the game is LOGICALLY over, the ACTUAL result will be the same as the LOGICAL result. With Japanese rules the game might be LOGICALLY over but the actual OUTCOME is needlessly delayed. In other words Japanese rules gets very petty about what happens AFTER the game is LOGICALLY over - the point where good players know what the result SHOULD be. Chinese rules is more intellectual about that - it doesn't care about things that are not important - Japanese is juvenile about this. That's why in my opinion Chinese rules are superior. They give more scope for skill, once a game is logically decided it's OVER and it doesn't place juvenile emphasis on what should be non-issues. Japanese is very petty about what happens AFTER the game is logically over and to me this isn't GO, it's poker. I wold point out that this is not a virtue, it is is a necessity designed to make the scoring come out right. It wasn't designed purposely to punish you for not passing. - Don Thanks ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
In message [EMAIL PROTECTED], Petri Pitkanen [EMAIL PROTECTED] writes All these are rather imaginary problems really. How many times you end arguing about the outcome of a game at the club? I rarely do. But 15-kyu players do; they generally ask a stronger player for help. This year, as referee at the London Open, I was not required to deal with any status problems. But I was summoned to deal with a game-end status argument there the previous year. Japanese rules are de-facto rules in international go and hence computer programs should implement them best they can. Humans can find it difficult enough. Requiring programs to do something that humans don't know how to do is unreasonable. If I am to referee a human event, I prefer area rules, which don't lead to these problems. If I am to referee a computer event, I greatly prefer them. Nick And they problems doe exist as Robert has pointed out, but simple counting procedure out weights any problems encountered so far. And besides on normal game difference is just 1 pt. Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. Petri -- Nick Wedd[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Fw: [computer-go] Re: Interesting problem
In the diagram, black has a chance to make a live group but only if white plays stupidly. there's a nice rule of thumb that says that you should only play moves whose outcome results in your opponent playing *what you think is the best move*. there's simply nothing more irritating than someone attempting an unreasonable invasion at the end of a game in order to try to turn a loss into a win. either they're assuming that you're unable to respond correctly, or hoping that you'll run out of time. exactly when this is the case -- that all reasonable people would stop playing -- is of course determined by the relative skill level of the players involved. many games in practice are resigned far before yose. many computer programs can determine when unambiguous end of game has occurred (i.e. when point-making yose has finished), and it would be most friendly of them at that point to discontinue playing. s. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: Fw: [computer-go] Re: Interesting problem
steve uurtamo wrote: there's simply nothing more irritating than someone attempting an unreasonable invasion at the end of a game in order to try to turn a loss into a win. I try this during the opening, the middle game, and the endgame. The only difference is in YOUR perception. -- robert jasiek ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: Fw: [computer-go] Re: Interesting problem
I try this during the opening, the middle game, and the endgame. The only difference is in YOUR perception. :) fair enough. s. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
On 4, Jan 2007, at 5:57 AM, Petri Pitkanen wrote: Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. This is the heart of my argument. I still consider it a feature when my program passes 100+ times in the endgame. I do think that a bot that plays hundreds of endgame moves that amount to nothing and that their opponent does not even need to answer should pay a point for each of those moves. I see it as perfectly fair that the bot with the better ability to read, and thus knows it can pass, should be rewarded for that reading skill. Cheers, David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
Oh ... I should have been more complete ... I think that the things said below should be the case when the tournament is not announced as playing under Chinese rules, as are all KGS computer tournaments. I do think that the TD gets to set the rules that they prefer. I just hope that someday the extra skill required as mentioned below is applied to computer programs, and rewarded accordingly. Cheers, David On 4, Jan 2007, at 12:53 PM, David Doshay wrote: On 4, Jan 2007, at 5:57 AM, Petri Pitkanen wrote: Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. This is the heart of my argument. I still consider it a feature when my program passes 100+ times in the endgame. I do think that a bot that plays hundreds of endgame moves that amount to nothing and that their opponent does not even need to answer should pay a point for each of those moves. I see it as perfectly fair that the bot with the better ability to read, and thus knows it can pass, should be rewarded for that reading skill. Cheers, David ___ 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: [computer-go] Re: Interesting problem
On Thu, 2007-01-04 at 12:53 -0800, David Doshay wrote: On 4, Jan 2007, at 5:57 AM, Petri Pitkanen wrote: Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. This is the heart of my argument. I still consider it a feature when my program passes 100+ times in the endgame. I do think that a bot that plays hundreds of endgame moves that amount to nothing and that their opponent does not even need to answer should pay a point for each of those moves. I see it as perfectly fair that the bot with the better ability to read, and thus knows it can pass, should be rewarded for that reading skill. Chinese views all this as a clean-up phase that is not important to the real game and so do I. I'm certainly not interested in winning points that way and would take no delight in it. I have a question. With perfect play, obviously a 9 stone handicap game is dead lost. If 2 perfect players played a game where one was given the 9 stones, and they played for maximum territory (obviously it doesn't make sense to play for a win) would the handicapped player be able to hold some territory at the end of the game?Could he carve out a little piece for himself even against his perfect opponents wishes? - Don Cheers, David ___ 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: [computer-go] Re: Interesting problem
OK, now I see your perspective ... the invader has the right to ask the defender to prove their skill, which I must say seems very much like a gamble to me, but should not be punished if their attempt is refuted. As such, I claim only that in this case we have to assume that it will be the norm for our programs because this is an unequal situation: no possible cost but some possible benefit. And indeed, it is what we see most programs do. Again, it only comes down to points when the defender tries things that the opponent can repeatedly ignore! If the invader is trying things that have to be answered move for move, then there is no penalty for trying. To me, this shows that there is balance in the risk/reward equation when the defender can pick up a point for properly evaluating the logical reality of the board position and then pass. This says to me that the one point loss per move you play that for which a defense is not required is indeed measuring skill and punishing a gamble. The Japanese player you mention below does not have to decide in advance if their opponent's defense of an invasion is possible or not, he just needs to determine if the opponent needs to answer at all. And to me, if this happens in the opening, the midgame, or the endgame, it is a standard part of determining the value of a move, and is a very good way to determine the strength of play. If my opponent keeps playing tenuki when I think my moves are meaningful, then I know that I am either going to win very big or get slaughtered by somebody who knows much better than me that those moves really did not matter. If they pass multiple times I have to ask why and look more carefully. In fact, it seems to me that saying PASS is the bigger gamble: you can easily just answer the invasion move for move and not change the score at all ... it takes greater faith to pass in order to pick up that point. I do not see why the situation should not be symmetric, and thus the invader must have equal faith that their probe *must* be answered. And while we are evaluating gambling in games of reason, I think that the skill level of everyone on this list is such that they have tried things they are not sure are going to work. It is the norm in very hard games, and we all know that Go is hard. This kind of gambling is even required in high handicap games. Cheers, David On 4, Jan 2007, at 9:08 AM, Don Dailey wrote: On Thu, 2007-01-04 at 15:57 +0200, Petri Pitkanen wrote: Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. No, this inhibits the application of skill. A silly invasion that wastes time is punished in all rules sets, but in Chinese it may not be silly if it doesn't waste time - Japanese rules unfairly defines these moves as silly. Chinese is better in this regard. You can try these invasions and put your opponent under pressure to refute them. When a Japanese player has a possible invasion that he knows is difficult but possible to defend, he must decide whether to play correctly or whether to gamble that his opponent won't be able to find the defense. With Chinese you can attack without inhibition in this situation and force your opponent to prove his skill.You can play more exciting games with Chinese rules. - 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: [computer-go] Re: Interesting problem
Kinda like how the discussion is on this mundane stuff instead of the interesting stuff? On 1/4/07, Don Dailey [EMAIL PROTECTED] wrote: On Thu, 2007-01-04 at 13:16 -0800, David Doshay wrote: I just hope that someday the extra skill required as mentioned below is applied to computer programs, and rewarded accordingly. I hope the programming effort isn't spend on this mundane stuff, but instead is applied to playing the game well - not trying to get rewards for passing. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
On 4, Jan 2007, at 1:37 PM, Don Dailey wrote: I'm certainly not interested in winning points that way and would take no delight in it. I do not take delight in picking up the points, but in my feeling that this shows true understanding of the reality of what is on the board. Whenever it looks like my program is playing like it really understands the board, I am delighted. Cheers, David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
Thanks Chris! that's all from me this time ... ;^) Cheers, David On 4, Jan 2007, at 1:46 PM, Chris Fant wrote: Kinda like how the discussion is on this mundane stuff instead of the interesting stuff? On 1/4/07, Don Dailey [EMAIL PROTECTED] wrote: On Thu, 2007-01-04 at 13:16 -0800, David Doshay wrote: I just hope that someday the extra skill required as mentioned below is applied to computer programs, and rewarded accordingly. I hope the programming effort isn't spend on this mundane stuff, but instead is applied to playing the game well - not trying to get rewards for passing. - 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: [computer-go] Re: Interesting problem
The discussion isn't mundane, it has helped me understand the rule-set differences even better. I also think it's an important discussion for the future of GO, I believe it's generally understood that Japanese rules is traditional, but the future is Chinese - that's the direction things have been moving. Most of the mediocre Chinese programs understand when the game is over and know what groups are dead. This isn't rocket science except in extreme cases it can get tough. In those cases the Japanese programs are equally clueless. Trying to determine the exact moment to pass seems like a tedious unimportant exercise that at best will give you a stone or two if you have a reasonable program. - Don On Thu, 2007-01-04 at 16:46 -0500, Chris Fant wrote: Kinda like how the discussion is on this mundane stuff instead of the interesting stuff? On 1/4/07, Don Dailey [EMAIL PROTECTED] wrote: On Thu, 2007-01-04 at 13:16 -0800, David Doshay wrote: I just hope that someday the extra skill required as mentioned below is applied to computer programs, and rewarded accordingly. I hope the programming effort isn't spend on this mundane stuff, but instead is applied to playing the game well - not trying to get rewards for passing. - 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: [computer-go] Re: Interesting problem
I'm done too ;-) - Don On Thu, 2007-01-04 at 13:58 -0800, David Doshay wrote: Thanks Chris! that's all from me this time ... ;^) Cheers, David On 4, Jan 2007, at 1:46 PM, Chris Fant wrote: Kinda like how the discussion is on this mundane stuff instead of the interesting stuff? On 1/4/07, Don Dailey [EMAIL PROTECTED] wrote: On Thu, 2007-01-04 at 13:16 -0800, David Doshay wrote: I just hope that someday the extra skill required as mentioned below is applied to computer programs, and rewarded accordingly. I hope the programming effort isn't spend on this mundane stuff, but instead is applied to playing the game well - not trying to get rewards for passing. - 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 mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Cheap multiprocessing
Those of you looking to wring more performance out of your MonteCarlo Go programs might be interested in this article about installing Linux on the Sony PlayStation 3 and programming the 6 available SPE coprocessors on its Cell cpu: http://www-128.ibm.com/developerworks/power/library/pa-linuxps3-1/ regards, -John ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
Please stop this confusion. Chinese scoring != Chinese rules Japanese scoring != Japanese rules Moreover, both Japanese and Chinese rules are to be considered traditional rules. E. On 1/4/07, Don Dailey [EMAIL PROTECTED] wrote: I also think it's an important discussion for the future of GO, I believe it's generally understood that Japanese rules is traditional, but the future is Chinese - that's the direction things have been moving. Most of the mediocre Chinese programs understand when the game is over and know what groups are dead. This isn't rocket science except in extreme cases it can get tough. In those cases the Japanese programs are equally clueless. Trying to determine the exact moment to pass seems like a tedious unimportant exercise that at best will give you a stone or two if you have a reasonable program. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
In message [EMAIL PROTECTED], Don Dailey [EMAIL PROTECTED] writes snip I have a question. With perfect play, obviously a 9 stone handicap game is dead lost. If 2 perfect players played a game where one was given the 9 stones, and they played for maximum territory (obviously it doesn't make sense to play for a win) would the handicapped player be able to hold some territory at the end of the game?Could he carve out a little piece for himself even against his perfect opponents wishes? I used to play a game with someone much (~8 stones) stronger than me, where he started by placing eight stones where he wanted them, and I then tried to live anywhere on the board. Usually I failed, but sometimes I succeeded. So I think the answer to your question must be, yes. 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] Re: Interesting problem
On Thu, 2007-01-04 at 23:28 +0100, Erik van der Werf wrote: Chinese scoring != Chinese rules Japanese scoring != Japanese rules So you can play with Chinese rules, but score the Japanese way? Please explain the difference so that I can use the correct terminology. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
On Thu, 4 Jan 2007, Don Dailey wrote: I have a question. With perfect play, obviously a 9 stone handicap game is dead lost. If 2 perfect players played a game where one was given the 9 stones, and they played for maximum territory (obviously it doesn't make sense to play for a win) would the handicapped player be able to hold some territory at the end of the game?Could he carve out a little piece for himself even against his perfect opponents wishes? Between equal players that's easy. I talked about this with very strong amateur ( 6d) from Taiwan and he told me that professionals estimate the handicap where white cannot live to be about 17 stones. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Cheap multiprocessing
Amazon Elastic Compute Cloud might be cheaper @ $0.10 per instance-hour consumed. doesn't the 'amazing amazon elastic waistband' require you to write all of your code using windows-based hooks? that kind've turns me off. s. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Cheap multiprocessing
Amazon Elastic Compute Cloud might be cheaper @ $0.10 per instance-hour consumed. doesn't the 'amazing amazon elastic waistband' require you to write all of your code using windows-based hooks? that kind've turns me off. You may be confusing with Amazon Simple Storage Service (which I've not studied). The compute cloud gives you a linux instance: http://docs.amazonwebservices.com/AmazonEC2/gsg/2006-10-01/ This is suited to go programs that can work on a cluster. The playstation multiprocessing looks very different: you get 1 general purpose CPU and 6 specialized CPUs. Their key feature is they have 256K of local memory - this is not cache, it is all the memory they can access. Not useful for UCT designs (which seem memory-limited currently) but fine for normal monte-carlo. It may also be ideal for running tactical search. In fact tactical search is around 6/7ths of the CPU cycles in a traditional go program, so a perfect fit? Darren ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Cheap multiprocessing
On 1/5/07, Darren Cook [EMAIL PROTECTED] wrote: The playstation multiprocessing looks very different: you get 1 general purpose CPU and 6 specialized CPUs. Their key feature is they have 256K of local memory - this is not cache, it is all the memory they can access. Not useful for UCT designs (which seem memory-limited currently) but fine for normal monte-carlo. The UCT tree is kept in main memory by the single PPE. The 6 SPEs which do the individual MC simulations don't need any access to that tree and should run perfectly fine in 256k... regards, -John ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
I was going to avoid more postings ... but it seems that any beauty of omission that might be achieved would be offset by the rudeness of not answering specifically posed questions. Answers embedded below. Cheers, David On 4, Jan 2007, at 4:29 PM, Jeff Nowakowski wrote: On Thu, 2007-01-04 at 12:53 -0800, David Doshay wrote: On 4, Jan 2007, at 5:57 AM, Petri Pitkanen wrote: Also It is good that unsound invasions are punished. This is supposed to be game of skill. If someone make silly invasion that does not require answer, the more skilled player i.e player that correctly passes should be awarded a point for his skill. This is the heart of my argument. I still consider it a feature when my program passes 100+ times in the endgame. Is it also a feature when a program cannot play out bent-4, because it knows that it is dead, but not why? Which program has more skill, the one that understands how to play it out, or the one that doesn't? It is difficult to discern the difference. But if you look at the SlugGo MoGo game in the slow KGS tournament, you will see that SlugGo avoided playing any time possible, and even avoided simple captures in a way that led to MoGo filling space in a way that avoided SlugGo having to play the extra capture stones. You can say that SlugGo understands nothing about endgame counting, or you can say that it shows signs of doing something well. Your choice. Again, almost all of those moves were pure GNU Go moves, so it speaks more to the quality of their endgame counting than anything I wrote for SlugGo. Japanese rules, in their pursuit of efficiency and beauty of omission, have thrown out the baby with the bathwater. This point I do not understand. However, I do understand how I can find that efficiency and beauty of omission lovely while others do not. I am a physicist, and very many of the equations I learned to understand are written explicitly in a max efficiency / min energy form. If the universe really works that way then it is lovely that this game captures it. You can no longer force an opponent to demonstrate his skill on the board; instead you must agree off the board what is alive or dead. I believe that this point was covered best by On 4, Jan 2007, at 2:28 PM, Erik van der Werf wrote: Please stop this confusion. Chinese scoring != Chinese rules Japanese scoring != Japanese rules I only wish to address how we should do our scoring, not the entire set of formal Japanese rules. Specifically, how we should score the result of a game when one bot passes and the other keeps playing. That is where this thread really got started, Lukaz's suggestion that a pass cost one point, because that will lead to the same result with Chinese or Japanese COUNTING. As Archivist of the AGA I have several volumes of Japanese rules. It is astounding how long those documents are. The only solace I get is that they are written in Japanese, and I don't read Japanese, so I do not have to worry about all of the details. And please, for once address this argument: When a player is *losing* under Japanese rules, how does it hurt him to make unreasonable invasions? Your argument is no argument at all. Japanese rules provide no benefit in this department. The only thing that happens is that they loose by more points to the extent that their opponent does not answer move for move. If your argument is that there is nothing beyond loosing, then yes, there is no clear motivation to avoid invasions that might bring the win back. I do not see a problem with that. To try is fine, perhaps even showing a tenacious spirit. That was my evaluation of the previously mentioned KGS tournament game between botnoid and SlugGo. SlugGo was going to win by 368.5 points, but botnoid kept playing and SlugGo kept passing, but eventually botnoid made things too complicated for SlugGo and lost by only 180 or so points. I had no problem with that. It was interesting and pointed out where SlugGo had evaluation problems. Even if SlugGo had lost the game, it would have been the same: a clear indication of a problem in counting in a liberty race. Winning may be more fun than loosing, but I usually learn more from loosing. In this case I had the lucky circumstance of both winning and learning, although there was the loss of 180 or so points. Cheers, David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Allocating remaining time
How much time should a program spend on each move? If my program has t milliseconds left to use in a game, and there are an estimated m moves left on the board (e.g., this many vacant spaces), one reasonable choice is t / m. In practice, this seems to spend too much time on early moves, which (under UCT/MC) is largely wasted time. Would it be better to use something like t / m**k, for some constant k? (Looking at graphs of such functions, k = 1.5 seems reasonable.) It would also be interesting to look at the graphs of how much time humans spend on each move; is it usually less for the opening moves than for middle / endgame moves? Is there a smooth curve, or is there a relatively abrupt shift from joseki to analysis? Peter Drake Assistant Professor of Computer Science Lewis Clark College http://www.lclark.edu/~drake/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Allocating remaining time
On Wed, 2007-01-03 at 22:04 -0800, Peter Drake wrote: How much time should a program spend on each move? If my program has t milliseconds left to use in a game, and there are an estimated m moves left on the board (e.g., this many vacant spaces), one reasonable choice is t / m. Excellent question. I think this really is a reasonable choice. I did a LOT of tests to determine this and it makes sense to front load quite heavily. I used a constant, I did not try to estimate how many moves might be left. If you use t/m you really should make m much smaller than the number of vacant points left. It's not a waste to spend a lot of time on early moves. From casual observation, I noticed that most games were decided very quickly, after just a few moves in 9x9. Another reason to front load is that the game gets easier and easier to play correctly as more stones get placed. It's a matter of concentrating the most energy where it's needed. My program notices when the game is pretty much a forgone conclusion and when this happens it plays even faster - I do this so that I can be even more aggressive about earlier moves. In practice, this seems to spend too much time on early moves, which (under UCT/MC) is largely wasted time. Would it be better to use something like t / m**k, for some constant k? (Looking at graphs of such functions, k = 1.5 seems reasonable.) You should test all of this. That's what I do. I think self-testing of different formula's and constants is fine for this kind of thing. It would also be interesting to look at the graphs of how much time humans spend on each move; is it usually less for the opening moves than for middle / endgame moves? Is there a smooth curve, or is there a relatively abrupt shift from joseki to analysis? Peter Drake Assistant Professor of Computer Science Lewis Clark College http://www.lclark.edu/~drake/ ___ 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: [computer-go] Re: Interesting problem
Ok, since you broke the truce so will I :-) On Thu, 2007-01-04 at 13:55 -0800, David Doshay wrote: I guess we will just have to leave it as a disagreement about what is important and what is mundane. I do not find the question of correct endgame reading to be mundane. What does this have to do with correct reading? Most of the reasonable programs, whether using area or territory scoring know what is going on, they know what is dead or alive. I don't think this discussion has anything to do with reading. If SlugGo passes 100+ times and in the process the opponent builds something that is then mis-evaluated (as happened in a game against botnoid in a KGS tournament) this is a very important thing for me to fix. If it turns out to be correct as it hangs itself way out on the edge, counting every liberty and cut correctly, then I am happy. It is not the winning, but the appearance of understanding that is important to me. Im not in to this. I would be programming chat-bots if I were. I'm not that interested in the aesthetics unless it comes for free. I just want to make the program play stronger. I don't care one whit if it can pass the Turing test or not. But how is this related to territory scoring? It's just as easy to make an area scoring program pass. I don't get it? I think I just stick with the more logical rule-set. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Cheap multiprocessing
The PS3 is a bit starved for memory - 512 megabytes, half seems to be for video, half for the main CPU. I just got a PS3 and hope to do some exploration with Linux programming. My own personal supercomputer :D Terry McIntyre UNIX for hire software development / systems administration / security [EMAIL PROTECTED] - Original Message From: Don Dailey [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Thursday, January 4, 2007 7:43:37 PM Subject: Re: [computer-go] Cheap multiprocessing Darren, The 6 CPU's don't have to keep the tree - they only have to do useful work. You could run a simulation on each of them. The question is how much memory is available for the whole system to run Linux and general purpose software on? - Don On Fri, 2007-01-05 at 09:07 +0900, Darren Cook wrote: Amazon Elastic Compute Cloud might be cheaper @ $0.10 per instance-hour consumed. doesn't the 'amazing amazon elastic waistband' require you to write all of your code using windows-based hooks? that kind've turns me off. You may be confusing with Amazon Simple Storage Service (which I've not studied). The compute cloud gives you a linux instance: http://docs.amazonwebservices.com/AmazonEC2/gsg/2006-10-01/ This is suited to go programs that can work on a cluster. The playstation multiprocessing looks very different: you get 1 general purpose CPU and 6 specialized CPUs. Their key feature is they have 256K of local memory - this is not cache, it is all the memory they can access. Not useful for UCT designs (which seem memory-limited currently) but fine for normal monte-carlo. It may also be ideal for running tactical search. In fact tactical search is around 6/7ths of the CPU cycles in a traditional go program, so a perfect fit? Darren ___ 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/ __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Interesting problem
2007/1/4, Don Dailey [EMAIL PROTECTED]: No, this inhibits the application of skill. A silly invasion that wastes time is punished in all rules sets, but in Chinese it may not be silly if it doesn't waste time - Japanese rules unfairly defines these moves as silly. It is silly if opponents best reply is pass Chinese is better in this regard. You can try these invasions and put your opponent under pressure to refute them. Is the refutation is pass even then? When a Japanese player has a possible invasion that he knows is difficult but possible to defend, he must decide whether to play correctly or whether to gamble that his opponent won't be able to find the defense. It it is severe enough that opponent has to reply It does not matter in any rule set. In Japanese if silly invasions needs a real refutation player gains point for extra prisoner and loses a point reply inside his/her own territory. No gamble there. BUT if it is so silly that PASS only thing that is needed, why in earth obviously the more skilled player i.e the one who knew that move does not even need an answer should not be awarded a point for it? Remember Chinese and Japanese rules give same outcome as long as players made same number of moves. -- Petri Pitkänen e-mail: [EMAIL PROTECTED] Phone: +358 50 486 0292 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Allocating remaining time
How much time should a program spend on each move? I think this is one of the most important and also difficult questions in game programming. Much effort is done to speed up the node-count by 10%, but a good time control is a much more effective speedup. If my program has t milliseconds left to use in a game, and there are an estimated m moves left on the board (e.g., this many vacant spaces), one reasonable choice is t / m. One should at least use t/(m+1). There is also a locial reason for this. If m is very small, especially m==1 one should have some extra time if the programm recognizes a problem. In this case it should search deeper. Generally this t/(m+k) should only be a target time. The final decision should be based on the results of the search. It is important to recognize trivial/forced moves and to stop in this cases search earlier. If the programm sees a problem than it should search longer. I have made recently a simple (but strong) UCT backgammon programm. UCT gives much better information for time-control than Alpha-Beta. E.g. if almost all search effort is concentrated on the best move, one can reasonable conclude that its a trivial/forced move. If the eval of the best moves decreases in the last period constantly and there are some chances that the second best becomes best, one should search on In practice, this seems to spend too much time on early moves, which (under UCT/MC) is largely wasted time. Would it be better to use something like t / m**k, for some constant k? (Looking at graphs of such functions, k = 1.5 seems reasonable.) Go-Programmers like it complicated. It would also be interesting to look at the graphs of how much time humans spend on each move; is it usually less for the opening moves than for middle / endgame moves? Is there a smooth curve, or is there a relatively abrupt shift from joseki to analysis? One should forget human behaviour. If I would have to make a Turing test - is the player human or a programm - I would not look at the moves but on the time behaviour. The fundamental difference is that (good) humans know when the position is difficult and when its easy. Programms have no understanding of this at all. Humans play Chess/Go, programm make chess/Go moves. Consequently humans think for a few moves very long, and play other moves rather fast. But I think that the time-control of humans is not at all optimal. Its very human to try to solve an urgent problem even at the risk that it makes solving a further problem more difficult. Humans tend therefore to get into Zeitnot. When playing against GM Adams I proposed 40 Moves in 2 hours. He proposed 40 Moves in 1 hour 40 minutes plus 30 sek/move. In the first moment I could not see the difference. In both cases one has 2 hours for 40 moves. But at move 30 its different. The flag is falling there already at 1h 55 minutes. Its a psychological trick to avoid extreme Zeitnot. But if the human would have a good time-control algorithm there is no need for this trick. He could save this 30 seks for himself. Chrilly Note: One should forget human behaviour generally. A programm is a programm is a programm. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/