Re: [computer-go] Tesuji
One would expect that a lot of Go-specific knowledge is required to develop a good program, but my impression is that some of the best Go programs so far have been actually written by people who know little about the game itself. Terry McIntyre <[EMAIL PROTECTED]> They mean to govern well; but they mean to govern. They promise to be kind masters; but they mean to be masters. -- Daniel Webster - Original Message From: Russ Williams <[EMAIL PROTECTED]> To: computer-go Sent: Monday, September 10, 2007 10:41:36 PM Subject: Re: [computer-go] Tesuji On 9/11/07, Joshua Shriver <[EMAIL PROTECTED]> wrote: > Was reading a page about Go and came across this term. Anyone know > what it means? With no disrespect intended, it seems like there are a fair number of go programmers who don't actually know go very much beyond the rules themselves. (I'm assuming from your question that you fall into this category.) So I'm curious why non-go-players (or minimal-go-players) are interested in programming go, instead of a game they know well. Is there a similar situation in chess (are there a lot of chess programmers who don't really know chess)? Hmm, maybe so. I also wonder whether experienced go programmers believe one needs to know go to be able to make a very strong go program. Or will some of the new Monte Carlo etc techniques sufficiently supplant expert domain knowledge that any good programmer with just a rudimentary knowledge of the rules of go will be able to make a strong go program? cheers, russ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ Be a better Globetrotter. Get better travel answers from someone who knows. Yahoo! Answers - Check it out. http://answers.yahoo.com/dir/?link=list&sid=396545469___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Congratulations to GNU Go!
In message <[EMAIL PROTECTED]>, Jason House <[EMAIL PROTECTED]> writes Has anyone verified "Hb04 now refused to proceed with the game, neither agreeing nor disagreeing about which stones were alive"? At the start of the tournament, HB04 supported genmove and "final_status_list dead", but not kgs-genmove_cleanup. HB04 played to what it considered the end of the game and then gave a legitimate response to "final_status_list dead" that I posted into the game chat (copied from kgsGtp's log file). As I understand the game end protocol, HB04 not supporting kgs-genmove_cleanup should not cause the game to hang. If both bots don't agree on "final_status_list dead", bots that support cleanup will cleanup and then all stones are considered alive. If anything, HB04 would simply get a worse score. I suspect that the issue was with MoGoBot2. Since I knew hb04 wouldn't clean up in the event of a disagreement, I tried to adjust when HB04 would pass between rounds 2 and 3 to avoid any future issues in the tournament. I overcompensated and HB04 played as described in round 2. I got the tuning right for round 3 (and beyond). Also, it appears that HBotSVN crashed before its second move in round 5. Thank you for telling me about these things - I have now corrected the page. Nick Nick Wedd wrote: Congratulations to GNU Go, the winner of both divisions of last Sunday's KGS bot tournament. MoGoBot might have done better but it was suffering from a bug that caused it to time out. My report is at http://www.weddslist.com/kgs/past/30/index.html. Nick ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- 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] Tesuji
I don't know how well other engine authors are in OTB gameplay, but personally I play the game a lot and trying to learn it to the best of my abilities. Not only because I enjoy the game, but to hope when I'm writing code it'll reflect in it . -Josh On 9/10/07, Russ Williams <[EMAIL PROTECTED]> wrote: > On 9/11/07, Joshua Shriver <[EMAIL PROTECTED]> wrote: > > Was reading a page about Go and came across this term. Anyone know > > what it means? > > With no disrespect intended, it seems like there are a fair number of > go programmers who don't actually know go very much beyond the rules > themselves. (I'm assuming from your question that you fall into this > category.) > > So I'm curious why non-go-players (or minimal-go-players) are > interested in programming go, instead of a game they know well. Is > there a similar situation in chess (are there a lot of chess > programmers who don't really know chess)? Hmm, maybe so. > > I also wonder whether experienced go programmers believe one needs to > know go to be able to make a very strong go program. Or will some of > the new Monte Carlo etc techniques sufficiently supplant expert domain > knowledge that any good programmer with just a rudimentary knowledge > of the rules of go will be able to make a strong go program? > > cheers, > russ > ___ > 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] Tesuji
It seems to be the case that you don't need to be a great player to write a great program but it certainly doesn't hurt. I feel that it holds me back since I learned the rules just so that I could write a program. I believe it's this way with other games too. Over the decades, some of the very best chess programs were written by non-masters. A few were written by very strong players and often strong players were involved as advisers. It seems that it's good to have a reasonable level of skill, but it's more important to be a skillful programmer. All other things being equal, I'm sure playing strength is important. I personally believe there is a certain skill-set that is somewhat related to teaching ability and not highly correlated to playing strength. The skill-set involves being able to articulate what you DO KNOW. I don't know how to explain it, but I can recognize it in others - I know how the really good chess programmers think and I'm sure similar skills are involved for those who know how to write good Go programs. Who has the best Go programs at 19x19 level? I think David Fotland is only 2 Dan and his is one of the best. I know the old handtalk program was written by a very strong player. How strong is Michael Reiss? And the other top guys? I'm not an expert on this but I would just guess that it's a bit more important in GO to be strong than in games like chess. - Don On Tue, 2007-09-11 at 07:41 +0200, Russ Williams wrote: > On 9/11/07, Joshua Shriver <[EMAIL PROTECTED]> wrote: > > Was reading a page about Go and came across this term. Anyone know > > what it means? > > With no disrespect intended, it seems like there are a fair number of > go programmers who don't actually know go very much beyond the rules > themselves. (I'm assuming from your question that you fall into this > category.) > > So I'm curious why non-go-players (or minimal-go-players) are > interested in programming go, instead of a game they know well. Is > there a similar situation in chess (are there a lot of chess > programmers who don't really know chess)? Hmm, maybe so. > > I also wonder whether experienced go programmers believe one needs to > know go to be able to make a very strong go program. Or will some of > the new Monte Carlo etc techniques sufficiently supplant expert domain > knowledge that any good programmer with just a rudimentary knowledge > of the rules of go will be able to make a strong go program? > > cheers, > russ > ___ > 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] Tesuji
> > Who has the best Go programs at 19x19 level? I think David > Fotland is > only 2 Dan and his is one of the best. I know the old > handtalk program > was written by a very strong player. How strong is Michael Reiss? > And the other top guys? The programs that reached the top quickly were all written by strong players. Nemesis - Bruce Wilcox - 5 Dan Goliath - Mark Boon - 6 Dan Handtank - Chen Zhixing - 6 Dan Go Intellect - Ken Chen - 6 Dan I was improving from 4 kyu to 1 dan while I was writing most of Many Faces, and It typically finished 3rd or 4th. Michael Reiss was about 1 Kyu or 1 Dan. His program became very strong against other programs over a long period of time with a lot of tuning against those programs. So I'd say that programmer go strength gives a small edge, enough to push the program from strong to best. I agree with Don that most important thing is the ability to turn your unconscious go knowledge into explicit knowledge that you can articulate. David > > I'm not an expert on this but I would just guess that it's a > bit more important in GO to be strong than in games like chess. > > > - Don > > > On Tue, 2007-09-11 at 07:41 +0200, Russ Williams wrote: > > On 9/11/07, Joshua Shriver <[EMAIL PROTECTED]> wrote: > > > Was reading a page about Go and came across this term. > Anyone know > > > what it means? > > > > With no disrespect intended, it seems like there are a fair > number of > > go programmers who don't actually know go very much beyond > the rules > > themselves. (I'm assuming from your question that you fall > into this > > category.) > > > > So I'm curious why non-go-players (or minimal-go-players) are > > interested in programming go, instead of a game they know well. Is > > there a similar situation in chess (are there a lot of chess > > programmers who don't really know chess)? Hmm, maybe so. > > > > I also wonder whether experienced go programmers believe > one needs to > > know go to be able to make a very strong go program. Or > will some of > > the new Monte Carlo etc techniques sufficiently supplant > expert domain > > knowledge that any good programmer with just a rudimentary > knowledge > > of the rules of go will be able to make a strong go program? > > > > cheers, > > russ > > ___ > > 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/
Re: [computer-go] Tesuji
In message <[EMAIL PROTECTED]>, Don Dailey <[EMAIL PROTECTED]> writes Who has the best Go programs at 19x19 level? I think David Fotland is only 2 Dan and his is one of the best. I know the old handtalk program was written by a very strong player. How strong is Michael Reiss? And the other top guys? Ken Chen is 6-dan. Chen, Zhixing, the writer of HandTalk/GoeMate is about 5-dan by European standards. Martin Müller and Robert Rehm are 5-dan. Daniel Bump and Arnoud Rutgers van der Loeff are 4-dan. David Forland is 3-dan. Joachim Pimiskern is 2-dan. Michael Reiss is 1-kyu. Wang, Yizao is 2-kyu. Guillaume Chaslot and Ivo Tonkes are 3-kyu. I have heard that Bruce Wilcox learned Go so as to be able to write a program to play it, and became 5-dan himself. 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] Tesuji
On 9/11/07, Nick Wedd <[EMAIL PROTECTED]> wrote: > > In message <[EMAIL PROTECTED]>, Don > Dailey <[EMAIL PROTECTED]> writes > > >Who has the best Go programs at 19x19 level? I think David Fotland is > >only 2 Dan and his is one of the best. I know the old handtalk program > >was written by a very strong player. How strong is Michael Reiss? > >And the other top guys? > > Ken Chen is 6-dan. Chen, Zhixing, the writer of HandTalk/GoeMate is > about 5-dan by European standards. Martin Müller and Robert Rehm are > 5-dan. Daniel Bump and Arnoud Rutgers van der Loeff are 4-dan. David > Forland is 3-dan. Joachim Pimiskern is 2-dan. Michael Reiss is 1-kyu. > Wang, Yizao is 2-kyu. Guillaume Chaslot and Ivo Tonkes are 3-kyu. > > I have heard that Bruce Wilcox learned Go so as to be able to write a > program to play it, and became 5-dan himself. It may be important to distinguish the ratings that people are now with their ratings when they started coding their program. I've improved by 9 stones (by kgs ratings) since I started my bot. (I wish I could say I'm a dan, but at 3k I think I'm still far from it). I find that sitting down to work on my bot often encourages me to play go instead... naturally leading to improved playing strength. I partly decided to write a bot so that I would be forced to solidify my go knowledge (through the process of teaching the computer what I know in a systematic way). Sadly, I don't think my coding of go has taught me anything about go yet. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Tesuji
I should get a rating on KGS just to get a rough idea of how weak I am. I know that when I started I could not tell if a group was dead in even simple cases. I thought my own program was blundering many times when the moves were actually good. Even though I haven't played a game (outside of my own program for testing) I know I've improved enormously just by watching my program play itself and other programs and from working out ways to improve it. I would expect that I might be around 20 kyu - much better than raw beginner but still not very good. Of course I'm not a top Go programmer either so I guess this is not very relevant ;-) - Don On Tue, 2007-09-11 at 13:32 -0400, Jason House wrote: > > On 9/11/07, Nick Wedd <[EMAIL PROTECTED]> wrote: > In message > <[EMAIL PROTECTED]>, Don > Dailey <[EMAIL PROTECTED]> writes > > >Who has the best Go programs at 19x19 level? I think David > Fotland is > >only 2 Dan and his is one of the best. I know the old > handtalk program > >was written by a very strong player. How strong is Michael > Reiss? > >And the other top guys? > > Ken Chen is 6-dan. Chen, Zhixing, the writer of > HandTalk/GoeMate is > about 5-dan by European standards. Martin Müller and Robert > Rehm are > 5-dan. Daniel Bump and Arnoud Rutgers van der Loeff are > 4-dan. David > Forland is 3-dan. Joachim Pimiskern is 2-dan. Michael Reiss > is 1-kyu. > Wang, Yizao is 2-kyu. Guillaume Chaslot and Ivo Tonkes are > 3-kyu. > > I have heard that Bruce Wilcox learned Go so as to be able to > write a > program to play it, and became 5-dan himself. > > > It may be important to distinguish the ratings that people are now > with their ratings when they started coding their program. I've > improved by 9 stones (by kgs ratings) since I started my bot. (I wish > I could say I'm a dan, but at 3k I think I'm still far from it). I > find that sitting down to work on my bot often encourages me to play > go instead... naturally leading to improved playing strength. > > I partly decided to write a bot so that I would be forced to solidify > my go knowledge (through the process of teaching the computer what I > know in a systematic way). Sadly, I don't think my coding of go has > taught me anything about go yet. > > ___ > 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] playing strength of programmers
1) The strongest players I've known personally have been extremely articulate about what they were doing & why; one used to practice explaining every move he made to his opponent. 2) I'm not that strong, myself--so you may take this with suitable seasoning. I think the main advantage of personal go strength is having enough experience of the game that you have a sense of the size of the tree-space, the critical importance of correct move-order & of the need to prune that search-tree. Anyone who thinks his program can afford to make a slack move... just doesn't understand the game's requirements. Forrest Curo - This email was sent using AIS WebMail. http://www.americanis.net/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] playing strength of programmers
There are some subtle distinctions to make when thinking about slack moves, though. Some strong moves simply solidify a connection enough to make a large region of the board come under more influence to be used later. This is really difficult to measure, because these moves often can serve several purposes, depending upon what happens later in the game. They aren't necessary, they don't immediately threaten anything, and they're not providing life for a group under threat of death. Instead, they are just setting the stage for changing the direction of play later. They may look like slack moves, but they aren't, and it can be difficult to time them, much less find them. Some stronger players can laugh at my description now. :) s. - Original Message From: "[EMAIL PROTECTED]" <[EMAIL PROTECTED]> To: computer-go@computer-go.org Sent: Tuesday, September 11, 2007 3:45:18 PM Subject: Re: [computer-go] playing strength of programmers 1) The strongest players I've known personally have been extremely articulate about what they were doing & why; one used to practice explaining every move he made to his opponent. 2) I'm not that strong, myself--so you may take this with suitable seasoning. I think the main advantage of personal go strength is having enough experience of the game that you have a sense of the size of the tree-space, the critical importance of correct move-order & of the need to prune that search-tree. Anyone who thinks his program can afford to make a slack move... just doesn't understand the game's requirements. Forrest Curo - This email was sent using AIS WebMail. http://www.americanis.net/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ Sick sense of humor? Visit Yahoo! TV's Comedy with an Edge to see what's on, when. http://tv.yahoo.com/collections/222 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/