In a game today giving 4H to a KGS 7k I killed a great deal of stones and then got lazy. I decided to do an ultra-deep, totally unreasonable invasion... basically just for the hell of it. If I had just reduced and worked to claim the center, I would have most likely won significantly anyways. Instead I go deep and end up with a large, quite dead group that is threatening to lose me the game.
So I abandon and work the other parts of the board. My opponent makes an endgame move that reduces him to two liberties. on an adjoining group and then plays tenuki with a fairly substantial endgame threat. Fairly substantial, but not nearly as big as the group. I use that opportunity to give myself an eye on my "dead" group, which has now sprung back to life. My opponent makes a series of forcing moves, ending with a 1 point gote endgame move that I don't have to answer, so I get back to his original threat and end up with both pieces, leading to me winning the game.
There have been several instances recently where trying to play "large scale" to either kill or live has gotten me in trouble. Sometimes I manage to pull myself out of hot water, sometimes I don't, but frequently I've found in my reviews that I don't need to invade or don't need to kill on such a broad scale in order to win.
In my game with the 1.9 dan at the tournament I won not because of any large scale kills, but by eating away at his edges and cutting off a few stones here and there.
Something to work on.
Friday, February 22, 2008
Fear, Luck, and Magic
One thing that is hard to conceptualize after a certain point is how the beginner sees the board. They seem to view groups living and dying as a matter of luck: that the opponent might sweep out of nowhere at any moment and sweep the stones from the board. I've commented before that "a ko started by a sufficiently stronger player is indistinguishable from magic."
Even as they get stronger, there are forms of magic that are just difficult to see even when they are happening, much less predict before hand. When I invade a three space knight's move, I check the ladders to make sure they work and, if they don't make sure that the result is still okay for me (or, alternatively, make sure that they do work with forcing moves). I know players who are so new, however, that they don't recognize that they are in a ladder until its already four steps into it. Others who only notice that its a ko for life when the group dies.
For a player at that level, whether that invasion works has a lot to do with luck. They fill a tiger's mouth with no apparent threat, or add extra stones to a shape--sometimes even killing what was previously a living shape--because of an unread, unrealized, and unrealizable fear. At one point I commented to player that the only way I could have cut there was if he played tenuki for six moves.
Eventually those problems may go away (or at least become less dramatic) but are instead replaced by underplays or hallucinations about whether a group is in danger.
This type of play has three consequences, the first two are obvious, the third is more subtle.
First, it gives up sente. This may be worth nothing, or it may be worth over a handicap stone in value.
Second, it is worth -1 point. I won a six stone handicap game the other day by half of a point. It takes just one -1 point gote move--even if there is nothing of value left on the board--to result in a loss under these circumstances.
Finally it keeps the player from seeing where a punishment can happen. If you lose by overplaying, the consequences of that overplay are quickly realized. A good teacher can then point out what happened and why it was an overplay. Underplays and filling your own territory like this will lead to loss after loss "for no specific reason." It keeps you from seeing if there actually was something that could happen there. If you do see such a sequence, then by all means protect but it should be as a consequence of specific reading and not of fear.
This is why review is so important, and one of the nice features of online games. In an online game I can go back over all of my decisions like that and see if I was hallucinating, but I have to review in some depth or with a stronger player to determine that. I may plays that are too small all of the time, and I work to ferret them out in the review and figure out where I should have played.
Even as they get stronger, there are forms of magic that are just difficult to see even when they are happening, much less predict before hand. When I invade a three space knight's move, I check the ladders to make sure they work and, if they don't make sure that the result is still okay for me (or, alternatively, make sure that they do work with forcing moves). I know players who are so new, however, that they don't recognize that they are in a ladder until its already four steps into it. Others who only notice that its a ko for life when the group dies.
For a player at that level, whether that invasion works has a lot to do with luck. They fill a tiger's mouth with no apparent threat, or add extra stones to a shape--sometimes even killing what was previously a living shape--because of an unread, unrealized, and unrealizable fear. At one point I commented to player that the only way I could have cut there was if he played tenuki for six moves.
Eventually those problems may go away (or at least become less dramatic) but are instead replaced by underplays or hallucinations about whether a group is in danger.
This type of play has three consequences, the first two are obvious, the third is more subtle.
First, it gives up sente. This may be worth nothing, or it may be worth over a handicap stone in value.
Second, it is worth -1 point. I won a six stone handicap game the other day by half of a point. It takes just one -1 point gote move--even if there is nothing of value left on the board--to result in a loss under these circumstances.
Finally it keeps the player from seeing where a punishment can happen. If you lose by overplaying, the consequences of that overplay are quickly realized. A good teacher can then point out what happened and why it was an overplay. Underplays and filling your own territory like this will lead to loss after loss "for no specific reason." It keeps you from seeing if there actually was something that could happen there. If you do see such a sequence, then by all means protect but it should be as a consequence of specific reading and not of fear.
This is why review is so important, and one of the nice features of online games. In an online game I can go back over all of my decisions like that and see if I was hallucinating, but I have to review in some depth or with a stronger player to determine that. I may plays that are too small all of the time, and I work to ferret them out in the review and figure out where I should have played.
Monday, February 18, 2008
17 February at Fiery Rain
Played five games (six, if you count a game against the club's founder in the morning).
My morning game was online, giving 3 stones to madamecp. I was ahead by a significant amount and then she played a 1/3rd of a point ko. I ignored the fact that it put a massive number of my stones into atari, and played elsewhere to fill another 1/3rd of a point ko across the board (whoops).
Right after that it went "erm" and "resign." ^_^
So later when we played at club she tried to make a forcing move she thought I would reply to before connecting her group. I read out that she couldn't live even with the followup to the forcing move, and so I went ahead and cut if off. She said "I thought you would reply to that!" so I joked "I didn't respond this morning when I had a 25 stone group in atari, why would you think I would reply to that!"
Played another game against wolvie at 6H, I won by 0.5. So my next game is kadoban against her to go to 7 x_x We were joking that I would end up throwing the game ^^;;
Played a 1k at 1H, taking W. He plays san-ren-sei against my dual 3-3s, and never seems quite sure how to handle the low, territory-seeking style of the 3-3s. We were both a little pressed for time, and I ended up winning by resignation.
Played an AGA 2.6 dan at 2H, kadoban game to go to 3. I ended up killing things in this game (for a change) and so managed to eek out a win, but it was a tough game. Keeping him dead is difficult, and his L&D skills are beyond mine. Hopefully this is the start of me actually overcoming that 2 stone handicap, since so far I've been horribly unsuccessful at keeping connected in the face of his ability to invade.
Played an AGA 3k or so even, taking white. It was a fast game, and I ended up winning.
So I had a good day for my play, I'm attributing it to still coming down off of the high from the tournament.
My morning game was online, giving 3 stones to madamecp. I was ahead by a significant amount and then she played a 1/3rd of a point ko. I ignored the fact that it put a massive number of my stones into atari, and played elsewhere to fill another 1/3rd of a point ko across the board (whoops).
Right after that it went "erm" and "resign." ^_^
So later when we played at club she tried to make a forcing move she thought I would reply to before connecting her group. I read out that she couldn't live even with the followup to the forcing move, and so I went ahead and cut if off. She said "I thought you would reply to that!" so I joked "I didn't respond this morning when I had a 25 stone group in atari, why would you think I would reply to that!"
Played another game against wolvie at 6H, I won by 0.5. So my next game is kadoban against her to go to 7 x_x We were joking that I would end up throwing the game ^^;;
Played a 1k at 1H, taking W. He plays san-ren-sei against my dual 3-3s, and never seems quite sure how to handle the low, territory-seeking style of the 3-3s. We were both a little pressed for time, and I ended up winning by resignation.
Played an AGA 2.6 dan at 2H, kadoban game to go to 3. I ended up killing things in this game (for a change) and so managed to eek out a win, but it was a tough game. Keeping him dead is difficult, and his L&D skills are beyond mine. Hopefully this is the start of me actually overcoming that 2 stone handicap, since so far I've been horribly unsuccessful at keeping connected in the face of his ability to invade.
Played an AGA 3k or so even, taking white. It was a fast game, and I ended up winning.
So I had a good day for my play, I'm attributing it to still coming down off of the high from the tournament.
Rocky Mountain Winter Tournament 2008
The tournament took place last weekend and I had a blast.
Four rounds, all even games (unless you are below around 15k, all games are even).
First Game
First I played someone who is a weak 3k on KGS and entered as a 1.0 dan--no previous tournament experience. He took the center and I took the edges/corners. In the middle game I had invaded one of sides, created a ko for life, and then ignored a fairly large threat so that it would live. I think I turned out slightly better from the threat, but only slightly.
I was beginning to work my way into the center, and hooked two stones into a ladder to give me access. The ladder was broken, then I fired a peep across the board threatening to split two of his groups which worked as a ladder breaker. He ignored it to secure the ladder, and I split his group into two separate groups. I then managed to kill both groups (one of them cleanly, the other because he played tenuki).
Second Game
I do not know who this was. He had no previous tournament experience, was very intense, and had entered as a 1.5 dan. He was there with some friends. His play was... weird. He played the "Go Seigen" style three star points on a diagonal, and then proceeded to duke it out with me for territory. Color me confused. There was a large group of his that would collapse to three liberties, and I had a tesuji to give me three liberties against it, but couldn't see a way to get 4 or to get those 3 with sente, so I left the position alone.
He was extremely confident in the position, since while I read it out he seemed to get bored. He played very very quickly, and I got the impression that he probably plays faster games online and wasn't used to dealing with the 45 minutes (longer than most Korean professional tournament games) of thinking time afforded us in the tournament.
Then we hit endgame. I basically won by luck. He played endgame with the large group and then didn't secure it, so I could now collapse it to two liberties vs. my three. When I began to collapse it, he didn't sacrifice any stones. Then I pulled my tesuji and captured his large group, and basically swung the game in my favor. If he had abandoned a few stones at any point during that sequence, I would have probably lost. Instead, he just passively let me collapse it to 2 liberties and then pull my tesuji to give me 3. Whoops.
Game Three
Game three was against the tournament organizer, an AGA 3.2 dan. This is the only game I took black in, and I lost by about a stone or two worth of points. I felt like I played well, and managed to keep the game fairly close. His comment was that I should work on not responding to moves--basically endgame/macroendgame--and that was the main portion of the difference in our strengths.
The game itself was a lot of fun for me. I managed to take the center without giving up too much on the sides.
Game Four
For this game I played a Korean 1.8 dan with a low sigma (uncertainty). He was so far 3-0 in the tournament. I made an early mistake and lost a side, but managed to take the center in exchange. He worked his way into my center, and for a while it really looked like I was going to lose the game--then I managed to start cutting off the edges of his invasions. There were a few cases where his invasions were connected by knight's moves, which I could cut through, or in one case where I could split his stones with a hane into a one space jump. His western side was enormous, and I really thought I had lost at the end of the game, but it turns out I was actually ahead.
Overall I had a lot of fun--it probably helped that I had nothing on the line--and fought some very strong opponents. It was great to play so many people around my strength in person, rather than just online. The intensity is different and I think it was good for my play.
Four rounds, all even games (unless you are below around 15k, all games are even).
First Game
First I played someone who is a weak 3k on KGS and entered as a 1.0 dan--no previous tournament experience. He took the center and I took the edges/corners. In the middle game I had invaded one of sides, created a ko for life, and then ignored a fairly large threat so that it would live. I think I turned out slightly better from the threat, but only slightly.
I was beginning to work my way into the center, and hooked two stones into a ladder to give me access. The ladder was broken, then I fired a peep across the board threatening to split two of his groups which worked as a ladder breaker. He ignored it to secure the ladder, and I split his group into two separate groups. I then managed to kill both groups (one of them cleanly, the other because he played tenuki).
Second Game
I do not know who this was. He had no previous tournament experience, was very intense, and had entered as a 1.5 dan. He was there with some friends. His play was... weird. He played the "Go Seigen" style three star points on a diagonal, and then proceeded to duke it out with me for territory. Color me confused. There was a large group of his that would collapse to three liberties, and I had a tesuji to give me three liberties against it, but couldn't see a way to get 4 or to get those 3 with sente, so I left the position alone.
He was extremely confident in the position, since while I read it out he seemed to get bored. He played very very quickly, and I got the impression that he probably plays faster games online and wasn't used to dealing with the 45 minutes (longer than most Korean professional tournament games) of thinking time afforded us in the tournament.
Then we hit endgame. I basically won by luck. He played endgame with the large group and then didn't secure it, so I could now collapse it to two liberties vs. my three. When I began to collapse it, he didn't sacrifice any stones. Then I pulled my tesuji and captured his large group, and basically swung the game in my favor. If he had abandoned a few stones at any point during that sequence, I would have probably lost. Instead, he just passively let me collapse it to 2 liberties and then pull my tesuji to give me 3. Whoops.
Game Three
Game three was against the tournament organizer, an AGA 3.2 dan. This is the only game I took black in, and I lost by about a stone or two worth of points. I felt like I played well, and managed to keep the game fairly close. His comment was that I should work on not responding to moves--basically endgame/macroendgame--and that was the main portion of the difference in our strengths.
The game itself was a lot of fun for me. I managed to take the center without giving up too much on the sides.
Game Four
For this game I played a Korean 1.8 dan with a low sigma (uncertainty). He was so far 3-0 in the tournament. I made an early mistake and lost a side, but managed to take the center in exchange. He worked his way into my center, and for a while it really looked like I was going to lose the game--then I managed to start cutting off the edges of his invasions. There were a few cases where his invasions were connected by knight's moves, which I could cut through, or in one case where I could split his stones with a hane into a one space jump. His western side was enormous, and I really thought I had lost at the end of the game, but it turns out I was actually ahead.
Overall I had a lot of fun--it probably helped that I had nothing on the line--and fought some very strong opponents. It was great to play so many people around my strength in person, rather than just online. The intensity is different and I think it was good for my play.
Tuesday, January 29, 2008
Winter Tournament
There's a rated tournament on February 16th out here I'll be attending. It will be my first real tournament. It looks like a lot of fun: there will be four rounds and all of my games will be even. So if I land in a game with a 5 dan, it will still be an even game. I find that idea fun, but I'll have to work on my "even skills" to prepare for the tournament.
Last Time at Fiery Rain!
Played 6 games:
Giving 9H to firebee, a beginner who is also in my Tuesday group.
Even against an AGA 1k (entering the next tournament as a half dan).
Giving 2H to a "SDK".
Giving 3H to an AGA 4k (entering the next tournament as a 3k).
Giving 4H to an AGA 6k (entering the next tournament as a 5k).
Taking 2H from an AGA 3d.
I won the first five games, but lost the last game after an intense fight ^_^ It was a lot of fun.
Giving 9H to firebee, a beginner who is also in my Tuesday group.
Even against an AGA 1k (entering the next tournament as a half dan).
Giving 2H to a "SDK".
Giving 3H to an AGA 4k (entering the next tournament as a 3k).
Giving 4H to an AGA 6k (entering the next tournament as a 5k).
Taking 2H from an AGA 3d.
I won the first five games, but lost the last game after an intense fight ^_^ It was a lot of fun.
Thursday, January 24, 2008
Self Review
When I used to review my own games--back at 18k--I saw a world of potential in things I could work on. It was clear, obvious, and direct.
It keeps coming progressively harder. I'm not sure if that's because I am out of practice, or if it is because I am strong enough that the self-review requires a more subtle grasp of what to look for.
It keeps coming progressively harder. I'm not sure if that's because I am out of practice, or if it is because I am strong enough that the self-review requires a more subtle grasp of what to look for.
Tuesday, January 22, 2008
Play by Heuristics
Yesterday I played a game against an unranked player. It was a strange game, where the other person's play depended on trying to incite mistakes. I won by resignation. I played someone else who's play was similarly strange, and lost by resignation.
It occurs to me that, at my level, many of my games are won and lost on what basically comes down to luck. In the first of those games, I made a dangerous cut that paid out because I was running on a rough heuristic involving eyespace and a count of liberties--not because of any deep reading. I read it out far enough to know that it looked promising--and no further.
In the second game I mentioned (it was actually my first of that day) I did something similar and it turned out I was wrong.
A lot of games--particularly more aggressive games--at my level seem to come down to one liberty, one L&D problem, etc. Sometimes this is not enough to swing the game, other times it makes all of the difference if both players bet on the same thing. This is one of the reasons a single game means very little: even a difference in several stones can be erased by just one or two mistakes of this nature.
While in the aggregate this will work out against a weaker/stronger player, against someone of roughly equal strength a heuristic going wrong can make all of the difference in the world.
To conclude this rambling, I think the underlying point is:
Heuristics are good, reading is better.
It occurs to me that, at my level, many of my games are won and lost on what basically comes down to luck. In the first of those games, I made a dangerous cut that paid out because I was running on a rough heuristic involving eyespace and a count of liberties--not because of any deep reading. I read it out far enough to know that it looked promising--and no further.
In the second game I mentioned (it was actually my first of that day) I did something similar and it turned out I was wrong.
A lot of games--particularly more aggressive games--at my level seem to come down to one liberty, one L&D problem, etc. Sometimes this is not enough to swing the game, other times it makes all of the difference if both players bet on the same thing. This is one of the reasons a single game means very little: even a difference in several stones can be erased by just one or two mistakes of this nature.
While in the aggregate this will work out against a weaker/stronger player, against someone of roughly equal strength a heuristic going wrong can make all of the difference in the world.
To conclude this rambling, I think the underlying point is:
Heuristics are good, reading is better.
Wednesday, January 09, 2008
The Benefits and Hazards of Computers
When you first start out--particularly if you are an introvert-computers seem like they are ideal opponents. They do not judge you, are not as intimidating, and they play in an apparently consistent manner without the nerve-wracking insanity that humans seem to carry. You can play a lot of games in a hurry and not feel to bad about losing or wasting someone else's time, and you can easily experiment with different strategies or varied handicaps.
Indeed, at the very start of playing many people will advise to play either IGOWIN or gnugo on a 9x9 until you can beat it consistently when taking black, just to get some of the fundamentals down pat.
The problem is that, as you get stronger, computers can ingrain bad habits. Computers are rarely good at strategy and seem to have--at best--fuzzy notions of attack and defense. They tend not to invade locations that should be invaded, and will get confused by even simple things that do more than one thing at once.
Most critically, they lack variability. In Hapki Do we practice against people of different heights, builds, and genders because what works on one person may not work on another, or may need modification. A technique that brings someone's arm across your body may not work the same way if the person is substantially taller or shorter. Or girl who is shorter and heavier may have an easier time with some techniques than a guy who is tall and lean, and vice-versa.
So its good to play them occasionally--to screw around or to get used to the basics--but after that we need to get to playing real people. Playing people online, playing people in person: just playing people and not becoming dependent on playing against any one individual or computer's "style."
Indeed, at the very start of playing many people will advise to play either IGOWIN or gnugo on a 9x9 until you can beat it consistently when taking black, just to get some of the fundamentals down pat.
The problem is that, as you get stronger, computers can ingrain bad habits. Computers are rarely good at strategy and seem to have--at best--fuzzy notions of attack and defense. They tend not to invade locations that should be invaded, and will get confused by even simple things that do more than one thing at once.
Most critically, they lack variability. In Hapki Do we practice against people of different heights, builds, and genders because what works on one person may not work on another, or may need modification. A technique that brings someone's arm across your body may not work the same way if the person is substantially taller or shorter. Or girl who is shorter and heavier may have an easier time with some techniques than a guy who is tall and lean, and vice-versa.
So its good to play them occasionally--to screw around or to get used to the basics--but after that we need to get to playing real people. Playing people online, playing people in person: just playing people and not becoming dependent on playing against any one individual or computer's "style."
Monday, January 07, 2008
Tactical and Strategic Bubbles
A long time ago back on SL I blogged about the concept of tactical and strategic bubbles. This, to me, is a useful set of terminology for modeling how players of various strengths see the board.
When a brand new player starts out, the board is huge and intimidating. So the player begins to try and break it down, isolating the various components into a series of smaller "tactical bubbles." This gives her a tunnel-like focus, frequently following wherever the opponent has played last. At the very beginning, this focus is extremely narrow to the point where things that have a tremendous influence on the fight and are right next door to that bubble are ignored.
As the player gets stronger three things happen:
Eventually the player realizes that she is kicking ass tactically but still losing games to stronger players. She begins to focus on strategy. This "strategic bubble" operates in the exact reverse of the tactical bubbles: the initial strategic bubble encompasses the entire board. Issues such as broad direction of play, ala Otake Hideo's Opening Theory Made Easy, start to become important. The emphasis at the beginning is on picking the correct side and on what can be considered "big plays."
As the player gets stronger the tactical bubbles get larger. "Will the ladder work" is considered more moves in advance, for example. Meanwhile the strategic bubbles get smaller and more refined: will developing this group end up hurting that group, etc.
Eventually these concepts also start to blur and blend together.
Just random musings from a long time ago that I've been revisiting.
When a brand new player starts out, the board is huge and intimidating. So the player begins to try and break it down, isolating the various components into a series of smaller "tactical bubbles." This gives her a tunnel-like focus, frequently following wherever the opponent has played last. At the very beginning, this focus is extremely narrow to the point where things that have a tremendous influence on the fight and are right next door to that bubble are ignored.
As the player gets stronger three things happen:
The player begins to be able to track multiple tactical bubbles at the same time.
These bubbles can overlap and move around, but do not jump as much with where the opponent has last moved.
The tactical bubbles grow larger. They begin to encompass adjacent stones, adjacent groups, surrounding groups, and groups that affect ladders.
Eventually the player realizes that she is kicking ass tactically but still losing games to stronger players. She begins to focus on strategy. This "strategic bubble" operates in the exact reverse of the tactical bubbles: the initial strategic bubble encompasses the entire board. Issues such as broad direction of play, ala Otake Hideo's Opening Theory Made Easy, start to become important. The emphasis at the beginning is on picking the correct side and on what can be considered "big plays."
As the player gets stronger the tactical bubbles get larger. "Will the ladder work" is considered more moves in advance, for example. Meanwhile the strategic bubbles get smaller and more refined: will developing this group end up hurting that group, etc.
Eventually these concepts also start to blur and blend together.
Just random musings from a long time ago that I've been revisiting.
Fiery Rain of Stones
Attended a new go club yesterday. It was my first time at the Fiery Rain of Stones go club and I had a great time. I played four people: giving 2H to two of them, playing one of them (a 1k?) even, and taking 3 stones from a 3 dan. I won all of the games, and they estimate my strength at around 1d in local terms. The 3 dan commented that my tactics were "very strong" and that the biggest thing to work on was being a little more willing to sacrifice stones--just a few moves sooner. I also gave up too much influence.
On the plus side, I managed two major kills (well, one major kill and keeping something from living).
It helps that I was playing at what I felt like was my best. I've always been a few stones stronger in person, and KGS rankings are evidently slightly stronger than local rankings.
I think I especially need to work on my endgame and my attack and defense, especially the uses and prevention of influence, and start working on my strategic view of the board again.
Experiments in L&D show that my L&D is not great, but is suitable to someone at my level. I'll need to work on it 'soon' though. My grip on maek is coming back to me, but isn't where I want it to be. Other parts of my game can wait until my endgame, attack/defense, and global strategy improve.
On the plus side, I managed two major kills (well, one major kill and keeping something from living).
It helps that I was playing at what I felt like was my best. I've always been a few stones stronger in person, and KGS rankings are evidently slightly stronger than local rankings.
I think I especially need to work on my endgame and my attack and defense, especially the uses and prevention of influence, and start working on my strategic view of the board again.
Experiments in L&D show that my L&D is not great, but is suitable to someone at my level. I'll need to work on it 'soon' though. My grip on maek is coming back to me, but isn't where I want it to be. Other parts of my game can wait until my endgame, attack/defense, and global strategy improve.
Subscribe to:
Posts (Atom)