Re: [computer-go] Fast Board implementation
You just can try few, and look whether the percentage of playout wins doesn't change too much. You probably need more that one starting position to test the rule. Lukasz On 1/15/07, George Dahl <[EMAIL PROTECTED]> wrote: What should the mercy threshold be for other board sizes than 9 by 9, particularly 19 by 19? - George Dahl > > > > Here are a few speedup tricks that have helped me. > > > > 1. The mercy rule. Since I'm incrementally keeping track of a list of empty > > points, it's no real extra pain to keep track of the number of black and > > white stones on the board. If the difference between them exceeds a > > threshold, the game is over. Ending early has an added bonus that I know the > > outcome without needing to score the board. (You can shoot yourself in the > > foot here. Best to pick a more conservative threshold the closer you are to > > interior nodes of the tree.) For exterior nodes far from any interior nodes, > > I use a threshold of 25 stones. ___ 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: MC thought
I tried to finish simulation as soon as dead stone difference was greather than a thresold but I got worst result. Its a totally preliminary. I tried that only one day. I suppose because at the end of the day, the MC simulation will try to fill all the board positions with stones so almost every simulation ends with a quite big quantity of dead stones in order to totally fill the board. Finishing the simulation earlier seems to add strong biases to it. I suppose, nothing is for sure. Eduardo --- Dave Dyer <[EMAIL PROTECTED]> escribió: > I wonder if MC programs shouldn't prune game > branches when > sufficiently large captures occur. The loss/win > might not > be strictly allocated to the right player, but it > certainly > means that the current game has entered sillyspace. > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > __ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Go and IQ training
I have the same IQ than Kasparov, He is millionarie and famous and I'm a moron that have to spend almost all his day coding for food. This is a good proof that IQ tests do not work :-P --- Aidan Karley <[EMAIL PROTECTED]> escribió: > In article > <[EMAIL PROTECTED]>, Mike > Olsson > wrote: > > This is a bit off topic, but I am wondering if a > person can play Go > > to increase their IQ or improve their > intelligence. >If one is going to discuss the extremely > slippy concept of > "intelligence" (or it's far, far slippier distant > relative > "Intelligence Quotient"), then it's practically > required to have read > Stephen Jay Gould's "Mismeasure of Man" (various > editions from about > 1980 to at least 1996, including ISBN-10: 0393314251 > / ISBN-13: > 978-0393314250). While it may not "blow out of the > water" the whole > subject of "intelligence testing", it does make one > very well aware > that the whole subject is a minefield of assumptions > and prejudices > (both conscious and unconscious. >I read what was probably the original edition > back in the > mid-80s, and loaned my copy to a university friend > who was studying > psychology ; 15 year later she declined to return it > because she was > still regularly using it to deflate novice > opinionated staff working > under her with the "learning impaired". That would > have been about the > time of the infamously neo-racist tract "The Bell > Curve". > > > From what I have read Kasparov's IQ is around 135 > so playing Chess > > doesn't really increase a person's IQ. > > >About 2.3 standard deviations above the norm. > That would imply > he's in the top 1½% or thereabouts of the population > in performance on > IQ tests. Sounds like there's be 3 > Kasparov-equivalents per couple of > full "Clapham Omnibuses". [Note 1] Or several per > average chess club. > Or maybe IQ test results are not a terribly good > predictor of chess > strength. I wouldn't really expect it to be much > better a predictor of > Go strength either. > > >For what it's worth, the Aberdeen University > Go Club was set up > in the early 1980s by ... a carpenter. Always a good > memory for > deflating one's potential to self-aggrandisment. > > > [Note 1] Standard British English idiom refers many > questions to the > opinion of the "man on the Clapham Omnibus", which > seats about 75 > people and stands another couple of dozen. > > -- > Aidan Karley, > Aberdeen, Scotland > Written at Mon, 15 Jan 2007 09:40 GMT, but posted > later. > > > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > __ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] UCT memory issues and scalability
I have been doing a lot of experiments with the scalability and memory usage of UCT.I'm using the exact scheme that was described like this in a previous posting by someone: Here is a summary of how it works: - Use probability of winning as score, not territory - Use the average outcome as position value - Select the move that maximizes v + sqrt((2*log(t))/(10*n)) v is the value of the move (average outcome, between 0 and 1), n the number of simulations of this move, and t the total number of simulations at the current position. In case a move has n = 0, it is selected first. For these experiments, I use random play-outs. This version should be a relatively generic implementation of what many of us are doing, without the frills and a relatively stable bug-free implementation. Each version does 2x the number of play-outs as previous version and the level of the program is defined as the number of play-outs * 1024 - so level 2 does 2048 playouts, etc. I am going to test as far as I can stand it like this: level 1 level 2 level 4 level 8 level 16 Additionally, I am testing 2 variants, because I am interesting in saving memory. The base version does the basic thing - always expanding a node in the tree once it has been visited one time, or in other words each probe is guaranteed to add 1 new node to the tree. The second variant will not expand a node until it has been visited 5 times - expanding on the 6th visit. This version is significantly more stingy about memory use. So far, after running well over 1000 games I cannot detect an advantage over using the less memory hungry version. There seems to be no point in wasting the memory. In a separate less systematic test I did last week, I also tested values as high as 100 visits before expansion at levels comparable or higher to CGOS time controls. In that test, I tested 1 visit, 5 visits, 10 visits and 100 visits before expansion. After playing a massive round robin (where each player played a total of over 200 games) I could not prove empirically that even 100 visits is too much. The ELO rating difference varied only 40 ELO points from high to low but the 1 visit version, which is supposed to be the best, was the second lowest rated version and the 100 visit version score 10 ELO points higher. This is all statistical noise. In this implementation the 100 visit version was substantially faster which makes it a no-brainer - however I replaced my node management routines with a fixed pool of nodes so there is no malloc or free operations. Now the slowdown is very minor due to excessive node expansions. Does anyone have any reason to believe we should be expanding so aggressively as after only 1 visit? My tests indicate that you can easily get away with anything less than 100. Here are the results of the current more systematic test. I will continue to expand this by introducing higher levels to the autotester and dropping the lower ones in order to perform a scalability study.We are seeing about 200 ELO per doubling at these levels. s05_032 means: 5 visits before expand - level 32 which is 32K play-outs per move. Perf Rating Win perc Tot Gms Ave Time Player --- --- --- 2107.190.749 227 50.3 s05_032 2102.790.323 217 52.2 s01_032 1942.179.452 219 32.1 s05_016 1916.378.027 223 32.6 s01_016 1764.365.447 246 16.1 s01_008 1753.161.200 250 15.3 s05_008 1542.140.000 245 6.8 s01_004 1526.539.914 233 6.3 s05_004 1284.322.273 220 1.8 s05_002 1274.822.594 239 1.8 s01_002 1000.0 7.623 223 0.2 s05_001 972.4 5.932 236 0.3 s01_001 Black wins: 652 46.9 % White wins: 737 53.1 % - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] UCT memory issues and scalability
There is a good argument why 100 is ok. When You have about 50 children, then waiting 100 playouts before start of attaching them results only in 2 playouts per child loss, so I guess even higher threshold should be OK. Lukasz On 1/16/07, Don Dailey <[EMAIL PROTECTED]> wrote: I have been doing a lot of experiments with the scalability and memory usage of UCT.I'm using the exact scheme that was described like this in a previous posting by someone: Here is a summary of how it works: - Use probability of winning as score, not territory - Use the average outcome as position value - Select the move that maximizes v + sqrt((2*log(t))/(10*n)) v is the value of the move (average outcome, between 0 and 1), n the number of simulations of this move, and t the total number of simulations at the current position. In case a move has n = 0, it is selected first. For these experiments, I use random play-outs. This version should be a relatively generic implementation of what many of us are doing, without the frills and a relatively stable bug-free implementation. Each version does 2x the number of play-outs as previous version and the level of the program is defined as the number of play-outs * 1024 - so level 2 does 2048 playouts, etc. I am going to test as far as I can stand it like this: level 1 level 2 level 4 level 8 level 16 Additionally, I am testing 2 variants, because I am interesting in saving memory. The base version does the basic thing - always expanding a node in the tree once it has been visited one time, or in other words each probe is guaranteed to add 1 new node to the tree. The second variant will not expand a node until it has been visited 5 times - expanding on the 6th visit. This version is significantly more stingy about memory use. So far, after running well over 1000 games I cannot detect an advantage over using the less memory hungry version. There seems to be no point in wasting the memory. In a separate less systematic test I did last week, I also tested values as high as 100 visits before expansion at levels comparable or higher to CGOS time controls. In that test, I tested 1 visit, 5 visits, 10 visits and 100 visits before expansion. After playing a massive round robin (where each player played a total of over 200 games) I could not prove empirically that even 100 visits is too much. The ELO rating difference varied only 40 ELO points from high to low but the 1 visit version, which is supposed to be the best, was the second lowest rated version and the 100 visit version score 10 ELO points higher. This is all statistical noise. In this implementation the 100 visit version was substantially faster which makes it a no-brainer - however I replaced my node management routines with a fixed pool of nodes so there is no malloc or free operations. Now the slowdown is very minor due to excessive node expansions. Does anyone have any reason to believe we should be expanding so aggressively as after only 1 visit? My tests indicate that you can easily get away with anything less than 100. Here are the results of the current more systematic test. I will continue to expand this by introducing higher levels to the autotester and dropping the lower ones in order to perform a scalability study.We are seeing about 200 ELO per doubling at these levels. s05_032 means: 5 visits before expand - level 32 which is 32K play-outs per move. Perf Rating Win perc Tot Gms Ave Time Player --- --- --- 2107.190.749 227 50.3 s05_032 2102.790.323 217 52.2 s01_032 1942.179.452 219 32.1 s05_016 1916.378.027 223 32.6 s01_016 1764.365.447 246 16.1 s01_008 1753.161.200 250 15.3 s05_008 1542.140.000 245 6.8 s01_004 1526.539.914 233 6.3 s05_004 1284.322.273 220 1.8 s05_002 1274.822.594 239 1.8 s01_002 1000.0 7.623 223 0.2 s05_001 972.4 5.932 236 0.3 s01_001 Black wins: 652 46.9 % White wins: 737 53.1 % - 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] UCT memory issues and scalability
On Tue, 2007-01-16 at 20:23 +0100, Łukasz Lew wrote: > There is a good argument why 100 is ok. > When You have about 50 children, then waiting 100 playouts before > start of attaching them results only in 2 playouts per child loss, so > I guess even higher threshold should be OK. And I haven't bothered to test over 100. But at 100 I still get good node expansion and fairly deep principal variations. > Lukasz ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Can Go be solved???... PLEASE help!
On Sat, 13 Jan 2007, Don Dailey wrote: One of the theoretical limitations to computing power (which was layed out in someones posts) and I have always understood to be the case, is related to space - the physical size of the universe. The problem with higher dimensions is that they are small AND they do NOT increase the 3-dimensional volume of our universe. Imagine a 2 dimesional (finite) surface and bend it in some way (eg. cylinder) ... even though your 2-dim "universe" exists now in 3 dimensions, it did not increase in area. If a computer can exist in 3 dimensions, couldn't an infinite number of them exist with 1 more dimension? Nope; see above. Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Can Go be solved???... PLEASE help!
On Tue, 2007-01-16 at 16:21 -0800, Christoph Birk wrote: > On Sat, 13 Jan 2007, Don Dailey wrote: > > One of the theoretical limitations to > > computing power (which was layed out in someones posts) and I have > > always understood to be the case, is related to > > space - the physical size of the universe. > > The problem with higher dimensions is that they are small AND they do > NOT increase the 3-dimensional volume of our universe. > Imagine a 2 dimesional (finite) surface and bend it in some way > (eg. cylinder) ... even though your 2-dim "universe" exists now in > 3 dimensions, it did not increase in area. > > > If a computer can exist in 3 > > dimensions, couldn't an infinite number of them exist with 1 more > > dimension? I'm suggesting computers that might exist outside our 3 dimensional space, not confined to our 3 dimensional space. Perhaps there are beings that see our space as flat from their many dimensions and any physical objects they deal with, are infinitely bigger that we can observe. For instance if there existed 2 dimensional beings, we could not show them 3 dimensional objects, just reflections of them and any of our objects would be infinitely large to them.If we could build 2 dimensional computers, we could stack any number of them on top of each other and they would not take up any extra space, no? - Don > Nope; see above. > > Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Can Go be solved???... PLEASE help!
On 16, Jan 2007, at 5:45 PM, Don Dailey wrote: For instance if there existed 2 dimensional beings, we could not show them 3 dimensional objects, The answers to this are in "Flatland: A romance of many dimensions" a nice short book by E.A. Abbott. just reflections of them slices and any of our objects would be infinitely large to them. No, they would only be aware of the slice that intersects their world. It would still have finite extent in their world. If we could build 2 dimensional computers, we could stack any number of them on top of each other Which is essentially what we really do ... and they would not take up any extra space, no? No. Cheers, David ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Go and IQ training
Another question that I would like to ask is what distinguishes a student at Harvard or at a top school from the rest of the students. And how can one develop the aptitude to reach that level. Or is it just that some people are born with a gift. Eduardo Sabbatella <[EMAIL PROTECTED]> wrote: I have the same IQ than Kasparov, He is millionarie and famous and I'm a moron that have to spend almost all his day coding for food. This is a good proof that IQ tests do not work :-P --- Aidan Karley escribió: > In article > <[EMAIL PROTECTED]>, Mike > Olsson > wrote: > > This is a bit off topic, but I am wondering if a > person can play Go > > to increase their IQ or improve their > intelligence. > If one is going to discuss the extremely > slippy concept of > "intelligence" (or it's far, far slippier distant > relative > "Intelligence Quotient"), then it's practically > required to have read > Stephen Jay Gould's "Mismeasure of Man" (various > editions from about > 1980 to at least 1996, including ISBN-10: 0393314251 > / ISBN-13: > 978-0393314250). While it may not "blow out of the > water" the whole > subject of "intelligence testing", it does make one > very well aware > that the whole subject is a minefield of assumptions > and prejudices > (both conscious and unconscious. > I read what was probably the original edition > back in the > mid-80s, and loaned my copy to a university friend > who was studying > psychology ; 15 year later she declined to return it > because she was > still regularly using it to deflate novice > opinionated staff working > under her with the "learning impaired". That would > have been about the > time of the infamously neo-racist tract "The Bell > Curve". > > > From what I have read Kasparov's IQ is around 135 > so playing Chess > > doesn't really increase a person's IQ. > > > About 2.3 standard deviations above the norm. > That would imply > he's in the top 1½% or thereabouts of the population > in performance on > IQ tests. Sounds like there's be 3 > Kasparov-equivalents per couple of > full "Clapham Omnibuses". [Note 1] Or several per > average chess club. > Or maybe IQ test results are not a terribly good > predictor of chess > strength. I wouldn't really expect it to be much > better a predictor of > Go strength either. > > > For what it's worth, the Aberdeen University > Go Club was set up > in the early 1980s by ... a carpenter. Always a good > memory for > deflating one's potential to self-aggrandisment. > > > [Note 1] Standard British English idiom refers many > questions to the > opinion of the "man on the Clapham Omnibus", which > seats about 75 > people and stands another couple of dozen. > > -- > Aidan Karley, > Aberdeen, Scotland > Written at Mon, 15 Jan 2007 09:40 GMT, but posted > later. > > > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > __ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ - Finding fabulous fares is fun. Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains.___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Go and IQ training
Getting way off topic now. On 1/16/07, Mike Olsson <[EMAIL PROTECTED]> wrote: Another question that I would like to ask is what distinguishes a student at Harvard or at a top school from the rest of the students. And how can one develop the aptitude to reach that level. Or is it just that some people are born with a gift. Eduardo Sabbatella <[EMAIL PROTECTED]> wrote: I have the same IQ than Kasparov, He is millionarie and famous and I'm a moron that have to spend almost all his day coding for food. This is a good proof that IQ tests do not work :-P --- Aidan Karley escribió: > In article > <[EMAIL PROTECTED]>, Mike > Olsson > wrote: > > This is a bit off topic, but I am wondering if a > person can play Go > > to increase their IQ or improve their > intelligence. > If one is going to discuss the extremely > slippy concept of > "intelligence" (or it's far, far slippier distant > relative > "Intelligence Quotient"), then it's practically > required to have read > Stephen Jay Gould's "Mismeasure of Man" (various > editions from about > 1980 to at least 1996, including ISBN-10: 0393314251 > / ISBN-13: > 978-0393314250). While it may not "blow out of the > water" the whole > subject of "intelligence testing", it does make one > very well aware > that the whole subject is a minefield of assumptions > and prejudices > (both conscious and unconscious. > I read what was probably the original edition > back in the > mid-80s, and loaned my copy to a university friend > who was studying > psychology ; 15 year later she declined to return it > because she was > still regularly using it to deflate novice > opinionated staff working > under her with the "learning impaired". That would > have been about the > time of the infamously neo-racist tract "The Bell > Curve". > > > From what I have read Kasparov's IQ is around 135 > so playing Chess > > doesn't really increase a person's IQ. > > > About 2.3 standard deviations above the norm. > That would imply > he's in the top 1½% or thereabouts of the population > in performance on > IQ tests. Sounds like there's be 3 > Kasparov-equivalents per couple of > full "Clapham Omnibuses". [Note 1] Or several per > average chess club. > Or maybe IQ test results are not a terribly good > predictor of chess > strength. I wouldn't really expect it to be much > better a predictor of > Go strength either. > > > For what it's worth, the Aberdeen University > Go Club was set up > in the early 1980s by ... a carpenter. Always a good > memory for > deflating one's potential to self-aggrandisment. > > > [Note 1] Standard British English idiom refers many > questions to the > opinion of the "man on the Clapham Omnibus", which > seats about 75 > people and stands another couple of dozen. > > -- > Aidan Karley, > Aberdeen, Scotland > Written at Mon, 15 Jan 2007 09:40 GMT, but posted > later. > > > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > __ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ Finding fabulous fares is fun. Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains. ___ 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] Go and IQ training
Can Go be used to increase a person's aptitude. Chris Fant <[EMAIL PROTECTED]> wrote: Getting way off topic now. On 1/16/07, Mike Olsson wrote: > Another question that I would like to ask is what distinguishes a student at > Harvard or at a top school from the rest of the students. And how can one > develop the aptitude to reach that level. Or is it just that some people are > born with a gift. > > Eduardo Sabbatella wrote: > I have the same IQ than Kasparov, He is millionarie > and famous and I'm a moron that have to spend almost > all his day coding for food. > > This is a good proof that IQ tests do not work > > :-P > > --- Aidan Karley escribió: > > > > In article > > <[EMAIL PROTECTED]>, Mike > > Olsson > > wrote: > > > This is a bit off topic, but I am wondering if a > > person can play Go > > > to increase their IQ or improve their > > intelligence. > > If one is going to discuss the extremely > > slippy concept of > > "intelligence" (or it's far, far slippier distant > > relative > > "Intelligence Quotient"), then it's practically > > required to have read > > Stephen Jay Gould's "Mismeasure of Man" (various > > editions from about > > 1980 to at least 1996, including ISBN-10: 0393314251 > > / ISBN-13: > > 978-0393314250). While it may not "blow out of the > > water" the whole > > subject of "intelligence testing", it does make one > > very well aware > > that the whole subject is a minefield of assumptions > > and prejudices > > (both conscious and unconscious. > > I read what was probably the original edition > > back in the > > mid-80s, and loaned my copy to a university friend > > who was studying > > psychology ; 15 year later she declined to return it > > because she was > > still regularly using it to deflate novice > > opinionated staff working > > under her with the "learning impaired". That would > > have been about the > > time of the infamously neo-racist tract "The Bell > > Curve". > > > > > From what I have read Kasparov's IQ is around 135 > > so playing Chess > > > doesn't really increase a person's IQ. > > > > > About 2.3 standard deviations above the norm. > > That would imply > > he's in the top 1½% or thereabouts of the population > > in performance on > > IQ tests. Sounds like there's be 3 > > Kasparov-equivalents per couple of > > full "Clapham Omnibuses". [Note 1] Or several per > > average chess club. > > Or maybe IQ test results are not a terribly good > > predictor of chess > > strength. I wouldn't really expect it to be much > > better a predictor of > > Go strength either. > > > > > > For what it's worth, the Aberdeen University > > Go Club was set up > > in the early 1980s by ... a carpenter. Always a good > > memory for > > deflating one's potential to self-aggrandisment. > > > > > > [Note 1] Standard British English idiom refers many > > questions to the > > opinion of the "man on the Clapham Omnibus", which > > seats about 75 > > people and stands another couple of dozen. > > > > -- > > Aidan Karley, > > Aberdeen, Scotland > > Written at Mon, 15 Jan 2007 09:40 GMT, but posted > > later. > > > > > > > > ___ > > computer-go mailing list > > computer-go@computer-go.org > > > http://www.computer-go.org/mailman/listinfo/computer-go/ > > > > > > > > > > __ > Preguntá. Respondé. Descubrí. > Todo lo que querías saber, y lo que ni imaginabas, > está en Yahoo! Respuestas (Beta). > ¡Probalo ya! > http://www.yahoo.com.ar/respuestas > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > > > > > Finding fabulous fares is fun. > Let Yahoo! FareChase search your favorite travel sites to find flight and > hotel bargains. > > > ___ > 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/ - Be a PS3 game guru. Get your game face on with the latest PS3 news and previews at Yahoo! Games.___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
