Re: [computer-go] an idea for a new measure of a computer go program's rank.
Personally, I use the terminology in much the same way as Heikki. I use the word mistake to describe (for example) a move that loses a large group, but does not change the game from a win to a loss. It makes sense to me to generally apply mistake to any move that loses points relative to the best move on the board. The term blunder means, essentially, a move that lost the game. It can be quite difficult, of course, to determine unambiguously whether or not a particular move is a blunder. In an otherwise close match, a large mistake (i.e., loses many points) is probably a blunder. Toward the end of a close game, it may be possible to find unambiguous blunders, and some of these could be single point mistakes. Weston On 1/23/07, Heikki Levanto [EMAIL PROTECTED] wrote: On Sun, Jan 21, 2007 at 08:16:07PM -0800, Ray Tayek wrote: I don't know the percentage of blunders. It also depends on what you call a blunder. Is a 1 point mistake a blunder? no, maybe 10 or more points My gut feeling is that a real blunder is enough to loose the game. Between equally strong players, a one point mistake can be a blunder, if it was late in the yose, and the game was won by half a point. On the other hand, throwing away a 20-stone group may not be a blunder if you were already going to loose by 100 points. It could even be a (mis?)calculated risk, ignoring a threatening move in order to get an attack on an even larger group, even if that attack later turns out not to work... Just my uninformed gut feeling, of course. -H -- Heikki Levanto In Murphy We Turst heikki (at) lsd (dot) dk ___ 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] an idea for a new measure of a computer go program's rank.
Yes, we heard that argument for years in computer chess and it never happened. Do you have some kind of basis for believe that? i wouldn't argue that future algorithms can't be time-doubled beyond the existing skill level of people, just that the current evidence is weak that we already have such algorithms in hand. s. TV dinner still cooling? Check out Tonight's Picks on Yahoo! TV. http://tv.yahoo.com/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On 21-jan-07, at 19:27, Don Dailey wrote: not considering biological factors which would cut into this a bit. There was a time when there were no time-limits in Go, which was abused by many players by turning a game into a stamina contest. I believe this practice was abandoned when someone collapsed at the board and died. Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
- Oorspronkelijk bericht - Van: Ray Tayek [EMAIL PROTECTED] Datum: zondag, januari 21, 2007 4:18 am Onderwerp: Re: [computer-go] an idea for a new measure of a computer go program's rank. also i suspect that at least 33% of the moves (at my 1-dan level) are wrong (what you might call in chess a blunder?). what do other people of different strengths think about this 33%? I don't know the percentage of blunders. It also depends on what you call a blunder. Is a 1 point mistake a blunder? But on average it would seem that a player loses about 13 points per game per grade separation from perfect play (11d?), implied by the definition of grade difference in relation to compensation by handicap stones. I don't know what the distribution of these mistakes related to their size would be (it would be interesting to find out), but I suspect the small mistakes would be more numerous. For a 1d this would imply a loss of about 130 points over the course of about 130 moves played by him in a game. So on average he loses 1 point per move. I would guess that a handful of mistakes would be big, but most moves lose just a little bit or nothing at all. Dave ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Sun, 2007-01-21 at 13:34 -0200, Mark Boon wrote: Don, I agree that more time generally leads to better moves. Also in Go. Where I think Go differs from Chess is the qualitative difference between a move that was thought about for 10 sec. or 2 hrs. is much smaller in Go than in Chess. And that's really because of the different nature of the games. Chess really is a tactical game, so looking at more positions improves the results considerably. To move up 200 ELO points in Go is usually not achieved by looking at more positions but by acquiring new concepts. To acquire a new concept in just a few hours is a rare thing. Some of these concepts would maybe take years to acquire if there wasn't someone to teach it to them. The gist of my whole argument is that the human brain is not limited. The idea that after you think a few moments and then you are at a dead end is preposterous and I'm glad you agree with me on this. I would like to say that I don't think this has anything to do with tactics. When you are in a tactical situation that might well be where the time is best spent but when you are not, you spend your time on what is most appropriate. I take a more meta-view of what a new concept is and I think the human brain is capable of acquiring them as you go. Indeed, the process of study and experience is a farce if you are gaining new concepts as you do this. And you can gain new insights or concepts in a single short study period. I know about this, I do it in chess and it's not always about tactics. I experience waves of understanding the longer I look at a chess position I do not understand. This happens in endgames for example where tactics is not much of an issue. I consider a given strategy with the belief that it is winning, but I continue to discover new things that cause me to modify my understanding. It is sometimes very like a process of elimination. It is not pure tactical thinking - it's noticing that certain things can't happen given the current configuration. Or using your imagination to try to determine how to compel the opponent to let you have your desired configuration. I can't believe go players don't have this thrill, because it's a wonderful process. The move or course that you thought was most productive proves to be wrong and you continue to narrow your focus.If this isn't true in Go and it's only about pattern recognition, (your either know the answer or you don't) then GO is a sterile uninteresting game. But I don't believe that. You wrote: If you are given twice as much thinking time, there is bound to be 2 or 3 moves in a 300 move game where it makes a difference in the quality of those 2 or 3 moves. And that is worth 1 or more ranks of strength. I don't know the exact formula, that was a for instance type of statement. In go I suspect extra time gives you MANY moves to improve on, not just 2 or 3 and perhaps that is worth a rank. Two or three superior moves would most likely be worth a few points, not a whole rank. Two or three blunders, that would maybe make a rank or two difference. But I really don't think doubling the thinking time would reduce the number of blunders by 2 or 3. And definitely not another 2 when doubling again. (Also I think the numbers are deceiving, the vital part of a Go game rarely lasts more than 200 moves and that is only 100 moves each. And many moves are forced.) The example you gave about studying a position for two hours and then showing it to someone 600 ELO points stronger. I think in Go someone who is 600 ELO points stronger can let the other player think about every move for a whole day and still beat him using on average just 10-20 sec. per move. It doesn't scale the way it does with Chess. I don't believe this at all. But it's difficult to argue about it since it is extremely difficult to construct a fair experiment in this regard.But I continue to be amazed that so many people think GO cannot be approached in a methodical logical way or that the human mind cannot break it down with the application of time and effort. I must admit this opinion is not very scientifically based, just on personal observations of seeing players of many different levels play. My opinion is based on watching these same arguments happen in computer chess over the last 30 years. Almost every good player believed chess involved skills that could not be programmed or reasoned out given enough time.I have also observed (over the years) that even the weak slow computers of yesterday could beat very strong players at speed chess, but not stand any chance whatsoever at long time controls. And it's well known that Grandmasters play speed-chess several hundred points stronger than weaker players - lot's of anecdotes about strong players given 1 minute for the whole game and still crushing weaker players given 20 minutes for
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Sun, 2007-01-21 at 11:32 -0800, terry mcintyre wrote: From: Don Dailey [EMAIL PROTECTED] By the way, can I assume that in world champion GO matches they use fast time controls because long time controls don't help in Go? Don probably had his tongue in cheek when he typed that, but according to http://en.wikipedia.org/wiki/Kisei , games in the Kisei Tournament in Japan are played over two days, with each player given eight hours of clock time. Title matches in the Mejiin also use 8 hours clock time. No doubt, when games are adjourned for the day, both players ( and perhaps their assistants ) spend a considerable amount of time thinking off the clock. I don't understand why they don't just play quickly if the extra time doesn't actually improve the quality of the games. It seems like this would just make them tired - which might create weaker moves (due to human fatigue) later in the game. But I was being tongue in cheek. I guarantee, without knowing anything about championship go, that the players work hard over the board and I have no doubt that there are moves where they spend several minutes making their decisions. I just don't believe it is not in the nature of GO to be able to improve your moves by thinking out your decisions and these championship games make my point. - Don __ 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] an idea for a new measure of a computer go program's rank.
- Oorspronkelijk bericht - Van: Don Dailey [EMAIL PROTECTED] Datum: zondag, januari 21, 2007 7:02 pm Onderwerp: Re: [computer-go] an idea for a new measure of a computer go program's rank. By the way, can I assume that in world champion GO matches they use fast time controls because long time controls don't help in Go? Of course time helps. I guess the difference between 8 hours time and 1 hour time gives an advantage of about 13 points (1 amateur grade) at the top professional level, which will probably swing the winning percentage from 50% to 90% at that level. Is this about 200 ELO? I would also benefit from more time. However, i don't think that 8 folding the time limit once more will bring the same 200 ELO increase in winning probability. The human mind does not scale like this, i think. Also you have to train to use this much time effectively, to stretch you attention span as much as possible. In Europe time limits in tournaments are usually set to about 1.5 hours. Increasing it to 4 hours will surely improve my winning probability, because i can avoid a lot of (mostly tactical) mistakes. My guess is i may gain about 20 points (i guess that corresponds to 150 ELO at my level). But giving me 8 hours will not improve it very much more. I don't think any time limit will increase my level by more than 200 ELO (30 points?), because: 1- I would not have the stamina to use this extra time effectively. 2- Mark Boon pointed out the problem of conceptual barriers. I just lack some of the concepts that 7d players master and I can't master these concepts on my own by thinking very hard during the course of a game. Dave ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
A lot of this interesting discussion has been about whether humans can make use of extra time. Some participants ( such as Dave Devos ) believe that, after a certain point, humans cannot improve their rank, at least not linearly with respect to time alloted. Fair enough; we humans require sleep, and we are not particularly good at sustaining complex thought over long periods, especially when a great deal of memorization is required. But this is a computer go list. How about computers? Can a computer make effective use of long time controls? We can actually experiment with two computer opponents with different time controls and know that ( unless the programs are devised to ponder on their opponent's clock ), one program will have x time, the other 10x to work with. Given sufficient time and resources, we could give an approximate answer to the question for a given computer program, an x-fold increase in thinking time will yield y elo points improvement. Of course, we'll also have to consider memory limitations. Some programs will gain little from extra time; some will run into memory limits before the clock expires; others may be more scalable. As for my own human anecdote, I am slowly making some progress from about 8kyu AGA to 6kyu AGA or thereabouts. I've always been stronger at tactics than strategy, making the most progress in the middle game. One method which has helped me win quite a few games against dan-level players, at handicaps smaller than the difference in our ranks would indicate, is to deliberately muddy the waters, to make the games as complex as possible, with numerous battles which intersect with each other. Dan-level players outmanuever me strategically, and almost always out-read me on any individual battle - but when the waters are muddied enough, they'll lose focus and spend a lot more time per move. I do think on my opponents' time, and by the time he works out a move, I already have a good counterplay lined up. All I need is one lapse, one tesuji to yank 20 or 30 points from my opponents - enough turn the tables. Unfortunately, I lack the strategic depth; if my opponent can stay cool and not make errors, I'll not be able to upset him. Which leads me to wonder if, at some future date when enough processors and memory are available, go programs might be able to leverage the advantage of depth of reading and sufficient memory to handle complex interactions into winning strategies. By the way, for those of you developing in Java, Azul Systems has created a custom JVM with 48 processors on a single chip, and a few other tweaks which look real promising: http://www.azulsystems.com Never miss an email again! Yahoo! Toolbar alerts you the instant new Mail arrives. http://tools.search.yahoo.com/toolbar/features/mail/___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
If you guys are correct thinking the nature of the game is such that humans cannot improve with time, then the computers will pull ahead more and more at longer time controls. let's adjust this to avoid the strawman and say that the counter-argument is that humans cannot improve much with significantly more time. what is the proposed idea? how many doublings should equal at least one stone? i think that when we played around with your code we found that doubling worked up until about 8192/16384 -- when it seemed like it was starting to lose based on time fairly frequently -- perhaps it could be doubled quite a few more times if we had fast enough machines. i can't remember what the full ELO spread was, but from, say, 1024 to 8192 there were at least several 100-point jumps, right? What do you think will happen? Do you believe that computers are actually more effective at utilizing extra time in 19x19 go? I think you are wrong. i think that computers will tap out and no longer be able to gain ELO after some (unknown) amount of doubling of thinking time. :) Wouldn't that be crazy if it turned out that humans improve more in chess with time but are incapable of improving at go and that computers are actually superior in this regard for GO? only if it held true for important advances in ELO (i.e. proving that this is the case up until exactly the strength of existing non-scalable programs wouldn't be as exciting as proving that you could double a piece of code to be stronger than existing programs). because, frankly, a few doublings are quite easy to lay your hands on, if it's really a scalable (or in particular, parallelizable) program. s. Cheap talk? Check out Yahoo! Messenger's low PC-to-Phone call rates. http://voice.yahoo.com ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Sat, 2007-01-20 at 15:06 -0500, Don Dailey wrote: Years ago A player in the chess club kept beating me over the head with a non-standard opening move that was difficult to refute. I got sick of this, sat down in the privacy of my own home and didn't get back up until I discovered the correct response.In effect I consulted a much stronger player, myself, given a lot of extra time. I think I spent about 2 hours on this - so it was as if I consulted a player a few hundred ELO points stronger. I found a move I had no chance of finding in 20 or 30 seconds, even after repeated ad-hoc unstructured attempts. As soon as a started playing this move, my opponent stopped using it and he had to work harder to beat me. I forgot to mention an interesting addendum to this story.I was only about 1700 rated at the time and I later showed the position to a 2300 player - a good friend of mine. I had already figured out the correct response but out of curiosity I wanted to see how quickly the 2300 player would find the right response. I set up the position and he took a glance at it. He did a little analysis out loud and figured out the correct move but it took him about 30-60 seconds - it wasn't as quickly as I thought it would be. But it makes sense. He was some 600 ELO stronger than I was, so I would expect him to find it about 64 times faster if each doubling is worth 100 ELO. I don't know if you can apply the formula directly to a single move like this, but it was interesting nonetheless that it was roughly in the same ballpark. This very same master always analyzed his games and someone we analyzed them (and my games too) together. He often found moves he should have played that he didn't consider during the actual game. This was always after spending a great deal of time studying the position. You cannot tell me that thinking long and hard about a difficult move will not enable you to make a better one. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
- Oorspronkelijk bericht - Van: Don Dailey [EMAIL PROTECTED] Datum: zaterdag, januari 20, 2007 9:06 pm Onderwerp: Re: [computer-go] an idea for a new measure of a computer go program's rank. Years ago A player in the chess club kept beating me over the head with a non-standard opening move that was difficult to refute. I got sick of this, sat down in the privacy of my own home and didn't get back up until I discovered the correct response.In effect I consulted a much stronger player, myself, given a lot of extra time. I think I spent about 2 hours on this - so it was as if I consulted a player a few hundred ELO points stronger. I found a move I had no chance of finding in 20 or 30 seconds, even after repeated ad-hoc unstructured attempts. As soon as a started playing this move, my opponent stopped using it and he had to work harder to beat me. It seems really odd to me that you are incapable of doing this in GO, or that the games are too different. If that's the case, then I prefer Chess, it is a far deeper game. I would find any game boring if it was so limited that there is nothing to think about that can't be seen in just a few moments. In my opinion in Go a game leaves the standard opening book very quickly, usually early in the opening. There are so many ways to play in the opening. If you opponent is trying to manipulate you into his favourite joseki(the taisha joseki for instance, with its proverbial 1000 variations), you have so many options to avoid it. But usually you just don't know what my opponent will play, so preparing for a particulal opponent is usually a waste of time. In my opinion, the difference is that in Go the possibity of variation is so great that a player is forced to rely on his own strength much earlier in the game than in Chess (in relation to the full length of a game). My level is 4d. For me the way to improve my results is studying professional games and Go problems. The aim is to get a very wide and general knowledge, more than a very deep knowledge of particular situations, because the level of variation in Go is so great. By improving you general knowledge of the game, you improve you ability to handle all those unique situations for which you cannot prepare in particular. Dave ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Sat, 2007-01-20 at 21:55 +0100, [EMAIL PROTECTED] wrote: In my opinion in Go a game leaves the standard opening book very quickly, usually early in the opening. There are so many ways to play in the opening. If you opponent is trying to manipulate you into his favourite joseki(the taisha joseki for instance, with its proverbial 1000 variations), you have so many options to avoid it. But usually you just don't know what my opponent will play, so preparing for a particulal opponent is usually a waste of time. Yes, there are not volumes of exact memorized opening moves in Go and so you can't prepare against an opponent with specific memorized variations. Of course this has nothing to do with the point I was making about the relationship between thinking time and move quality. In my opinion, the difference is that in Go the possibity of variation is so great that a player is forced to rely on his own strength much earlier in the game than in Chess (in relation to the full length of a game). Yes, I would agree with this. Even Bobby Fischer noticed this and came up with a chess variant to render opening knowledge moot. My level is 4d. For me the way to improve my results is studying professional games and Go problems. The aim is to get a very wide and general knowledge, more than a very deep knowledge of particular situations, because the level of variation in Go is so great. By improving you general knowledge of the game, you improve you ability to handle all those unique situations for which you cannot prepare in particular. All interesting games require this - Go is not unique in this regard. I did not intend for anyone to think I was making a statement about the importance of memorizing openings or preparing for specific opponents. I get the feeling that you believed I was talking about this. I was responding to Ray Tayek who believes that he cannot produce higher quality moves no matter how much time he is given. That's not how it works for me. - Don Dave ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Sat, 2007-01-20 at 15:34 -0700, Arend Bayer wrote: Hi Don, On 1/20/07, Don Dailey [EMAIL PROTECTED] wrote: If what you are saying is true, this is a waste of time. They should not be able to produce better quality moves than what they produce over the board. This has little to do with the question of whether you can improve a single move a lot by spending a lot of hours on it, but more with the fact that Go has many more reasonable moves in every opening position, so the game will leave your opening book preparation much quicker, especially compared to the overall length of the game. But I'm not talking about opening preparation. My point is all about just a few critical moves, not the majority of them. If you are given twice as much thinking time, there is bound to be 2 or 3 moves in a 300 move game where it makes a difference in the quality of those 2 or 3 moves. And that is worth 1 or more ranks of strength. snip: on improving a move in chess by spending many hours on it It seems really odd to me that you are incapable of doing this in GO, or that the games are too different. If that's the case, then I prefer Chess, it is a far deeper game. I would find any game boring if it was so limited that there is nothing to think about that can't be seen in just a few moments. I think of that in the opposite way. Go is such a deep game that in any position, there is a lot I will never be able to understand just by spending many hours on it. There are some things I may always misjudge that a professional will see immediately. If I think a group is weak and needs strengthening, but a pro just sees that it can never be attacked profitably, then that's not something where I can correct my mistaken thinking by spending many hours on the position. I believe this is all part of the strength/time relationship curve. If there is a huge disparity in playing strength, giving you a thousand times more thinking time won't be nearly enough to make up the gap. For instance ... Even when you double the speed of a chess playing computer, you add only a tiny amount of strength - so small it's not easily measured statistically. It's the same, I believe, with humans and probably why everyone here seems to believe what I'm saying is wrong, they think that I am implying that you can spend a few minutes on a move and play champion level. But if you are given twice as much thinking time, it's not going to turn you games from idiotic to brilliant. It will improve the (average) quality of your moves, but barely enough to notice. Having said that, I believe it's a lot more in GO based on some experiments I did with Steve Uurtamo in trying to get 19x19 CGOS ready. There is an ENORMOUS strength difference between programs that think twice as long - do 2X more monte carlo play-outs.Someone on this group (I can't remember who) correctly pointed out that a 19x19 has a lot of moves in it and so just a slight improvement in skill translates to a large winning percentage against even a slightly weaker opponent.This appears to be quite true. To put another perspective on it: If I had an hour for every move in a tournament game, I might play good EGF 5d level instead of average EGF 4d. That's a big difference from my perspective, but a small one when you compare it with the strength difference between me and a Korean who just became pro. This is understood. See what I said above.I don't really know how much 1 extra dan represents at this level - I think it translates to 200 or more ELO points. We can figure this out - what is the win expectancy of 5 dan over 4 dan without handicap? You said an hour per move - what are you comparing this against? 10 seconds per move? 1 minute per move? Arend ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
Le dimanche 21 janvier 2007 01:23, Don Dailey a écrit : On Sat, 2007-01-20 at 15:34 -0700, Arend Bayer wrote: Hi Don, To put another perspective on it: If I had an hour for every move in a tournament game, I might play good EGF 5d level instead of average EGF 4d. That's a big difference from my perspective, but a small one when you compare it with the strength difference between me and a Korean who just became pro. This is understood. See what I said above.I don't really know how much 1 extra dan represents at this level - I think it translates to 200 or more ELO points. We can figure this out - what is the win expectancy of 5 dan over 4 dan without handicap? http://gemma.ujf.cas.cz/~cieply/GO/statev.html 4D 30.6% (out of 4000 games) Alain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
On Fri, 2007-01-19 at 14:04 +0900, Darren Cook wrote: My point being that a top pro will find a high quality move in the time it takes him to move the mouse from one side of the board to the other. But still it's *WAY* below his normal tournament playing strength to play so quickly... Everything I know about the way top pros play says the opposite: quickly diminishing returns from extra time. The first move they think of is often the one they will choose even after 10 minutes of study. Hi Darren, What do you mean by diminishing returns?You don't get 10 ELO stronger for every second of thinking time, it's not linear like this. I think you are suffering from a human perception issue which I will explain in a moment, but first ... With computer chess it's well documented that you get an enormous strength improvement with each doubling of speed. Over the years this has fallen off - it used to be close to 100 ELO but I think it's more like 50 ELO now. However, doubling the thinking time USUALLY didn't make it play a different move. In fact, today the top programs play a great move instantly most of the time. Even the older much weaker programs found the best move very quickly and never changed their minds.And yet they still improved hundreds of ELO points when hardware continued to get faster. How can this be? The answer is that only a few moves make the difference. Even weak players play the same moves that grandmasters play - it's only the occasional move that makes all the difference. So your intuition is correct that strong players play great moves quickly, but that has little to do with what is required to bring this up a notch. Now, about the time perception issue. I can explain this best with an anecdote: Years ago, I marketed a chess program and I received a lot of correspondence and feedback from my customers. Do you know what the most cliched comment was? It went like this: I really like your program but I kept beating it at 5 seconds so I doubled the level and it didn't play any stronger. Of course they were quite wrong - it DID play stronger - perhaps 60 ELO rating points. But they had unrealistic expectations about what a doubling of time really means. 60 ELO is significant, but statistically it would take a really long match to measure it accurately. You are not going to play a couple of games and say, WOW, this is a LOT stronger! A commercial GO program must have a mode where it plays moves very quickly. Why? Because nobody will buy it otherwise. It's biological, we get impatient waiting for a move after a few seconds. Double the thinking time and it plays significantly stronger, but not enough to be immediately noticed.What we WILL notice is that most of the moves are the same - and human perception is better at making quick binary judgments, i.e. it still stinks, it's not any better! I'm suggesting that it's no different with us humans, in fact I'm absolutely convinced of it.If you ignore human frailties, such as attention span and ability to focus for long periods of time - us humans will play MUCH stronger given a few hundred percent extra time. It won't make us play EVERY move better - a reasonably good player probably plays a lot of moves correctly. But that's not what makes the difference between a good player and a better player. It's those difficult moves that we require a lot of time to work out. I'll give you a hypothetical chess example to show you the limitations of this. Let's take an 1800 ELO chess player. He has almost NO chance of beating a top grandmaster. A tiny fraction of a percent per game.Let's give him an acceleration potion that speeds up his metabolism so that he can think 64 times faster. Instead of 3 minutes he is thinking over an hour per move on average. I assert that this 1800 player is now playing at least 300 points stronger - about 2100 points.But guess what, he still has almost no chance of beating the 2800 player. The skeptics will look at the game, laugh, and say, see, the extra time didn't help a bit. Do you, or anyone, have studies that deal with this, for go? (I saw your other post on chess, but I think this may be somewhere chess and go differ: perhaps due the emphasis in go on good shape?) It's really hard to believe that GO cannot be studied but chess can. I contend that this applies to any field of endeavor. Put a man on the moon? You need a LOT of brainpower. It took a team of men several years to solve the problem.Playing games is just a set of problems to be solved, some very easy some very difficult. For a strong player most of the moves are easy - the difference between 6 dan and 7 dan has nothing to do with most of the moves, it's just a very few highly critical difficult ones, perhaps just one or two decisions. Present a player with a set of difficult to solve problems and ask
Re: [computer-go] an idea for a new measure of a computer go program's rank.
for what it's worth, strong players often spend enormous amounts of time on moves. professional tournament games are not generally of the 2-second-per-move variety. historically, they have taken days, but i'm not sure what the standard is now. perhaps someone who has seen a web simulcast of a recent tournament game can comment here. keep in mind that lots of money is riding on the result of just a few games, and that these tournaments are played over weeks and weeks of time. look at the schedule for a professional go tournament sometime -- it's not a weekend affair. the reason that a pro would need at most 1s/move to beat the top go playing program is simply that any of his top 10 move choices will be vastly better than any of the computer's top 10 move choices with nonzero probability. that means that even if they overlap on the top move 90% of the time (and this is highly unlikely), any of the pro's top 10 move choices will be better the rest of the time, and this slight move/evaluation difference will magnify itself into huge board-changing trauma over the course of the game. if it takes a 9-stone stronger player 100 moves to undo your 9-stone advantage in a correctly-handicapped game at 5s/move, how many moves at 1s/move do you think it would take a professional to undo your zero-stone advantage if you were a computer player? my guess is 3 or 4, since computers tend to choose fairly random opening moves from a small safe set, but without full-board knowledge. s. Never Miss an Email Stay connected with Yahoo! Mail on your mobile. Get started! http://mobile.yahoo.com/services?promote=mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
At 08:45 PM 1/18/2007, you wrote: On Thu, 2007-01-18 at 20:05 -0800, Ray Tayek wrote: yes. i would easily give my opponent *much* more time than a few handicap stones. the effect of time making someone (or thing) play better (or worse) is non-linear and probably only effective over some small range of time and talent. I think the formula is probably similar to UCT or Chess, but even more so for humans. sorry, no clue about the formula. Double the amount of time you have, and significantly increase the quality of the move. I don't think this is a limited effect over a narrow range of time. i suspect that it is in humans. i am only a 1-dan player. but during most of the game: if i have a reasonable amount if time (say 1 hour or so), doubling or trippling the time to think about one move (or for the whole game) does not make any difference (but i have been playing for 40 years). i tend to reach my limit of reading (look ahead). i would cut my time to 40 minutes for 2 stones and play for money. 30 minutes for 3 stones, 25 minutes for 4 stones, 20 minutes for 5 stones. giving most 1-dans more than an hour is not going to help their game that much. we only play so well. pro's can probably defeat this since they can make the game complicated. I understand chess better than go, I used to be a tournament player. Give me time to think and I can produce moves of enormously higher quality over tournament time-controls. I know this for a fact. I seriously doubt it is different for go. i don't play chess. but it seems different to me in go. ... It probably is non-linear like you say - even in the more limited game of Chess, the curve was amazingly linear (every doubling in time seemed to give a fixed amount of ELO strength improvement) ... well, chess is close to 1+ battles. more look ahead should help in some linear way perhaps. go goes off the rails fast when you consider interactions of say the corner josekis to other corners. As far as talent is concerned, some chess experiments seem to indicate ... I think it might work the same with humans - ... ... don' t know enough to comment. So I think strength in humans is very much the same - perhaps even more scalable than with computers - subject of course to human frailties of attention span, sleep time, ability to focus for long periods of time, etc. i play 20-25 minute games on yahoo sometimes when i am bored. these are moderately fast. some people play insanely fast (too me). like 10 minutes (this is total time. no byo-yomi). ignoring what a group of people might be able to do, i suspect that having more than two hours of time per game for amateurs is the limit of usefulness. a pro could probably benefit from a much larger increase in time. thanks --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] an idea for a new measure of a computer go program's rank.
Hi, The challenge to write a go playing program that could beat a professional was issued before the wide availability of Internet Go Servers, and broadband access. Under these new conditions, it is trivial to write such a program, provided the game takes place on a server, and at time limits chosen by the program. For example a random point playing program could choose time limits of half a second per move, sudden death. Therefore I suggest that a program's strength can (if needed) be expressed as the shortest time limits that a player of a standard strength (eg Pro. 1 dan) would be willing to play the program at, given an equal reward/loss regime (ie the chance of either winning would be 0.5). The format of time limits for such games would need to be standardised, for example - it could be decided that only limits of the type 'sudden death, x number of seconds per move' were allowed. In that case, 'x' could be used as a measure of the program's strength (as an abreviation for 'would beat a standard strength player half the time at x seconds per move') Of course the strength of a 'standard strength' Go player varies, and professional one dans would likely be unwilling to be beaten in ultra blitz games for the benefit of computer go programmers, so 'amateur 1 dan' is a realistic idea for a standard strength go player. dan ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
Hi Dan, Your suggestions hits at what I consider a basic truth or an axiom for game playing entities, humans or computers - that strength is a function of time and memory.Skill can be viewed as time. The skillful player is just making his time count more by being more efficient, sometimes many orders of magnitude more efficient. It works like running. The fast runner gets to a destination in a certain amount of time - but someone on crutches can still get there if you give him more time.We view the faster runner as more skilled as a runner even though they both can cover the same ground. Of course in computing - and in humans - memory can be traded off for speed. Intensive knowledge based programs is one way to make that trade-off.Every child learns this when he memorizes the multiplication table - it makes it possible to multiply much larger numbers faster. And sometimes people confuse knowledge with IQ, and in a sense I think they are right. - Don On Thu, 2007-01-18 at 21:18 +, dan wrote: Hi, The challenge to write a go playing program that could beat a professional was issued before the wide availability of Internet Go Servers, and broadband access. Under these new conditions, it is trivial to write such a program, provided the game takes place on a server, and at time limits chosen by the program. For example a random point playing program could choose time limits of half a second per move, sudden death. Therefore I suggest that a program's strength can (if needed) be expressed as the shortest time limits that a player of a standard strength (eg Pro. 1 dan) would be willing to play the program at, given an equal reward/loss regime (ie the chance of either winning would be 0.5). The format of time limits for such games would need to be standardised, for example - it could be decided that only limits of the type 'sudden death, x number of seconds per move' were allowed. In that case, 'x' could be used as a measure of the program's strength (as an abreviation for 'would beat a standard strength player half the time at x seconds per move') Of course the strength of a 'standard strength' Go player varies, and professional one dans would likely be unwilling to be beaten in ultra blitz games for the benefit of computer go programmers, so 'amateur 1 dan' is a realistic idea for a standard strength go player. dan ___ 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] an idea for a new measure of a computer go program's rank.
especially because computers don't have to click the relevent move with a mouse. They just think it and its done. Make a computer go program move the mouse and click like the human or make a computer go program physically place the stone on the board and if a computer can win in speed go, i'll be impressed then. Although that is a somewhat different task. On 1/18/07, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: I would not consider it very impressing nor interesting if a fast 10k program beats strong players on time. It think the stronger player will win with 10 seconds per move, but lowering the time limit until the stronger player loses on time is just silly. Dave - Oorspronkelijk bericht - Van: dan [EMAIL PROTECTED] Datum: donderdag, januari 18, 2007 10:18 pm Onderwerp: [computer-go] an idea for a new measure of a computer go program's rank. Hi, The challenge to write a go playing program that could beat a professional was issued before the wide availability of Internet Go Servers, and broadband access. Under these new conditions, it is trivial to write such a program, provided the game takes place on a server, and at time limits chosen by the program. For example a random point playing program could choose time limits of half a second per move, sudden death. Therefore I suggest that a program's strength can (if needed) be expressed as the shortest time limits that a player of a standard strength (eg Pro. 1 dan) would be willing to play the program at, given an equal reward/loss regime (ie the chance of either winning would be 0.5). The format of time limits for such games would need to be standardised, for example - it could be decided that only limits of the type 'sudden death, x number of seconds per move' were allowed. In that case, 'x' could be used as a measure of the program's strength (as an abreviation for 'would beat a standard strength player half the time at x seconds per move') Of course the strength of a 'standard strength' Go player varies, and professional one dans would likely be unwilling to be beaten in ultra blitz games for the benefit of computer go programmers, so 'amateur 1 dan' is a realistic idea for a standard strength go player. dan ___ 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] an idea for a new measure of a computer go program's rank.
On Thu, 2007-01-18 at 23:02 +0100, [EMAIL PROTECTED] wrote: I would not consider it very impressing nor interesting if a fast 10k program beats strong players on time. It think the stronger player will win with 10 seconds per move, but lowering the time limit until the stronger player loses on time is just silly. It's not silly at all. It's just there is a point in which the mechanics of making a move add too much noise - you can win or lose because you dropped a stone on the floor for instance. I know that even humans vary considerably in strength with time, and I don't mean just blitz moves.The problem is that this is generally not observed because BOTH players usually get the same time so both are handicapped by the clock. We used to play Chess by subtracting or adding 1 minute every time you win or lose a game. This works pretty well if one player isn't too much stronger and it would probably work great if it were not for the fact that the stronger player isn't handicapped as much as you would think - he uses the weaker players extra time.If you could isolate that factor by slowing down time for one of the players, you would have an excellent handicap system. In Chess, humans seem to benefit from time even more than computers, although most peoples intuition is just the opposite. You don't want to play speed chess against a computer and a human would be the favorite at postal chess, say 24 hours per move despite the enormous strength increase this gives a program. Most people are irrational and illogical about all of this, they think humans play about the same but it's just a serious perception issue. That's not how it really works. - Don Dave - Oorspronkelijk bericht - Van: dan [EMAIL PROTECTED] Datum: donderdag, januari 18, 2007 10:18 pm Onderwerp: [computer-go] an idea for a new measure of a computer go program's rank. Hi, The challenge to write a go playing program that could beat a professional was issued before the wide availability of Internet Go Servers, and broadband access. Under these new conditions, it is trivial to write such a program, provided the game takes place on a server, and at time limits chosen by the program. For example a random point playing program could choose time limits of half a second per move, sudden death. Therefore I suggest that a program's strength can (if needed) be expressed as the shortest time limits that a player of a standard strength (eg Pro. 1 dan) would be willing to play the program at, given an equal reward/loss regime (ie the chance of either winning would be 0.5). The format of time limits for such games would need to be standardised, for example - it could be decided that only limits of the type 'sudden death, x number of seconds per move' were allowed. In that case, 'x' could be used as a measure of the program's strength (as an abreviation for 'would beat a standard strength player half the time at x seconds per move') Of course the strength of a 'standard strength' Go player varies, and professional one dans would likely be unwilling to be beaten in ultra blitz games for the benefit of computer go programmers, so 'amateur 1 dan' is a realistic idea for a standard strength go player. dan ___ 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] an idea for a new measure of a computer go program's rank.
There is one way to attempt to adjust for this - give the computer a 1 or 2 second penalty for each move. - Don On Thu, 2007-01-18 at 16:06 -0600, Nick Apperson wrote: especially because computers don't have to click the relevent move with a mouse. They just think it and its done. Make a computer go program move the mouse and click like the human or make a computer go program physically place the stone on the board and if a computer can win in speed go, i'll be impressed then. Although that is a somewhat different task ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
In my opinion lowering the time limit just forces players (human and computer) towards random play. I am sure there exists a time limit where a random playing program can beat Lee Chang-Ho 50% of the time. But what is the use of that? To me it sounds like an invention to be able to show some progress in computer go, even if programs don't become very much stronger over the years: at least they will become quicker :) Dave - Oorspronkelijk bericht - Van: Don Dailey [EMAIL PROTECTED] Datum: donderdag, januari 18, 2007 11:19 pm Onderwerp: Re: [computer-go] an idea for a new measure of a computer go program's rank. There is one way to attempt to adjust for this - give the computer a 1 or 2 second penalty for each move. - Don On Thu, 2007-01-18 at 16:06 -0600, Nick Apperson wrote: especially because computers don't have to click the relevent move with a mouse. They just think it and its done. Make a computer go program move the mouse and click like the human or make a computer go program physically place the stone on the board and if a computer can win in speed go, i'll be impressed then. Although that is a somewhat different task ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] an idea for a new measure of a computer go program's rank.
Why not go the other way? Granted that postal players put more thought into their moves than those on a 30-minute clock, but they surely do not think about their move for an entire 24 hours - but a computer can actually allocate a full 24 hours per move. Considering the benefits Mogo observed with multiple processes, and the recent results on memory-efficient monte carlo algorithms, perhaps this tradeoff would work to the computer's advantage. Terry McIntyre - Original Message From: [EMAIL PROTECTED] [EMAIL PROTECTED] To: [EMAIL PROTECTED] Cc: computer-go@computer-go.org Sent: Thursday, January 18, 2007 4:12:58 PM Subject: Re: [computer-go] an idea for a new measure of a computer go program's rank. In my opinion lowering the time limit just forces players (human and computer) towards random play. I am sure there exists a time limit where a random playing program can beat Lee Chang-Ho 50% of the time. But what is the use of that? To me it sounds like an invention to be able to show some progress in computer go, even if programs don't become very much stronger over the years: at least they will become quicker :) Dave - Oorspronkelijk bericht - Van: Don Dailey [EMAIL PROTECTED] Datum: donderdag, januari 18, 2007 11:19 pm Onderwerp: Re: [computer-go] an idea for a new measure of a computer go program's rank. There is one way to attempt to adjust for this - give the computer a 1 or 2 second penalty for each move. - Don On Thu, 2007-01-18 at 16:06 -0600, Nick Apperson wrote: especially because computers don't have to click the relevent move with a mouse. They just think it and its done. Make a computer go program move the mouse and click like the human or make a computer go program physically place the stone on the board and if a computer can win in speed go, i'll be impressed then. Although that is a somewhat different task ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ Never Miss an Email Stay connected with Yahoo! Mail on your mobile. Get started! http://mobile.yahoo.com/services?promote=mail___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
