Life In 19x19
http://www.lifein19x19.com/

Knotwilg's practice
http://www.lifein19x19.com/viewtopic.php?f=48&t=12096
Page 1 of 19

Author:  Knotwilg [ Sat Jul 25, 2015 1:52 pm ]
Post subject:  Knotwilg's practice

All right. I'll try keeping my journal here. Those who've been around know that I'm concentrating on

- time management
- devoting overtime periods to key middle game positions
- remain thick
- devote time to evaluate the position before the endgame
- win in the endgame

Here's a funny game to start with. Any comments specifically on the abovementioned practice goals, are welcome and all others are too :)


Author:  Loons [ Sat Jul 25, 2015 6:26 pm ]
Post subject:  Re: Knotwilg's practice

I am pretty sure :w16: is considered even.

Not sure about the :b17: direction, I am unfamiliar with it.

I believe :b19: to be a mistake, at least after R13 (though I am unfamiliar with the joseki) because it strengthens white on top, weakens black and white seems a little hard to attack. (with that said I think white's :w18: approach was slightly questionable and that after Q17 black would be winning).

Ultimately white wanted to make small life many times and it was an easy game for you.

Author:  Shaddy [ Sat Jul 25, 2015 8:17 pm ]
Post subject:  Re: Knotwilg's practice

Close the game out as soon as you can; for example, with 51 you can add a move on the right and it'll be over. White has no target and no potential, so he shouldn't be able to catch up. I don't think it's useful to think of the game in terms of "I want to win in the endgame". Instead, keep your eyes peeled for opportunities and exploit them fully.

Author:  Knotwilg [ Thu Aug 06, 2015 2:34 am ]
Post subject:  Re: Knotwilg's practice

In between the project of moving into a new house, I played another game. I attempted to stick with the 25s/move pace and force myself into considering alternatives and reading them out. When I reviewed the game, I was surprised to notice how little I had read. I did consider alternatives on multiple occasions but then made a mainly intuitive choice. Only in a few cases (life & death scenarios) did I actually read a couple of moves deep. In some of those cases, I even took the worst decision after doing the reading.

I've tried to give a rendition of my thinking as honestly as possible. Key moments in the game:

:w39: a succesful net, using shortage of liberties, to connect all groups
:w93: the wrong choice, resulting in a ko which the opponent didn't play
:w105: I didn't consider a simple split of the top, which I would have definitevely chosen if I had considered it
:w177: and subsequent
:w191: the proper defense for endgame purposes, after reading out the L&D situation
:w197: it seems Black could still live if he descended here


Author:  Knotwilg [ Fri Aug 07, 2015 1:56 am ]
Post subject:  Re: Knotwilg's practice

My tsumego practice is centered around two of the best online collections on the web:

1. Hitachi Go Problems by Harada Minoru: http://www.hitachi.co.jp/Sp/tsumego/index-e.html
2. Go Game Guru's collection by David Ormerod and An Younggil: https://gogameguru.com/weekly-go-problems-week-1

I've gone through GGG's repetitively and I'm going through them again. They are currently at 136 triplets consisting of an easy, an intermediate and a hard problem. The easy ones are really basic tsumego. Last time I did them, I missed 1 or 2 out of negligence. The intermediate ones are at my level. Last time I had an 80% rate but more importantly I realized I'm not really reading them out completely. Rather I will scan the problem for an acceptable solution and then verify, as said with 80% success rate. Problem is that part of that 80% is luck. The hard ones are usually too hard for me and I'm basically browsing through the problem with the editor, sometimes "finding" the solution because I got the main idea right.

I'm going through the easy series again but I've raised the bar: I first look at the problem, then minimize the window and keep the problem it in my head to solve it there. I'm writing down the size of each path I'm traversing. I'm doing this to find another way of being active and postponing looking at the solution. In the meantime I'm building a metric for the complexity of these problems at my level and try understanding where (someone like) I "read" or "know". For example, I will prune any variation arriving at the L-group because I know it is dead.

My expectation is that my reading will become sharper and my "knowing" will enhance. One effect from keeping the problem in my head is that I'm much more sensitive to liberties. And that is something I can definitely use!

Author:  Knotwilg [ Fri Aug 07, 2015 11:11 am ]
Post subject:  Re: Knotwilg's practice

Some conclusions after doing 30 intermediate GGG problems through absolute reading, without looking at the solution and 30 easy ones without looking at the position but visualize them in my head.

1. you understand solutions better if you read all failed variations
2. you get a special eye for shortage of liberties by visualizing problems in your head; I often went back to check if a stone was in place at the outside when it

mattered
3. the solving practice tends to be more oriented along patterns of reading than patterns of shape
4. the reading tends to be more systematic, like trying hane to reduce eyespace first and only later find the vital point
5. reading deep becomes easy very soon; for example the 13 moves in the solution of intermediate 21 were not such a big deal

What makes a problem hard?

1. the hardest problems are those with an undefined purpose, like middle game attacking exercises where one wants to set the opponents' stones afloat
2. how much you "know" is definitely contributing to the level of a problem; pruning can be quite aggressive when you find an L-group, a door shape or a comb shape, or

bent four in the corner
3. in L&D exercises, sometimes the outcome is ko; this is harder than pure life or death; also, it can matter who takes the ko first
4. problems with lots of short alternatives are harder than those with a few deep ones, because you need to keep track of the tree and the outcomes

I went on to tackle the Hitachi problems from 1 as well - a recipe for burn out, as there are already more than 1000 duplets of them. Problem 3 was a real cracker and I worked very hard on it, during a bike ride home:


Author:  Knotwilg [ Mon Aug 10, 2015 9:05 am ]
Post subject:  Re: Knotwilg's practice

One of the best things about reading a problem completely is to look for failure modes after having found the solution. One possible effect is that you find "another solution" which leads to denouncing the first or the second, but the more beneficial effect is that you get many problems in one and an overall better comprehension of the problem and its topology:

Click Here To Show Diagram Code
[go]$$B corner
$$ -----------------
$$ | . a . . . . . .
$$ | 1 . X . . . . .
$$ | . X O O . O . .
$$ | b X O , . . . .
$$ | . O . . . . . .
$$ | . . O . . . . .
$$ | . . . . . . . .
$$ | . . . . . . . .[/go]


For example, in this basic problem, it's almost trivial to see that :b1: is the solution, making miai of a and b. This time, I went on to see why it is the *only* solution:

Click Here To Show Diagram Code
[go]$$B corner
$$ ----------------
$$ | . b . . . . . .
$$ | a . X 1 2 . . .
$$ | . X O O . O . .
$$ | . X O , . . . .
$$ | . O . . . . . .
$$ | . . O . . . . .
$$ | . . . . . . . .
$$ | . . . . . . . .[/go]


After exchanging :b1: for :w2:, which follows the principle of trying to expand eyespace first and then play the vital point, the original move doesn't work anymore. In fact the position has become symmetrical and this time around a and b have become miai for White to kill, due to shortage of liberties at both ends (exercise left to the reader).

This pattern is known to me from the tripod group but it was reinforced after investigating the failure, rather than leading me to the problem's solution, which was trivial to begin with.

At this stage of my intuition in problem solving, investigating failure modes still leads to (re)discoveries like these. After sufficient rounds of similar problems, the whole problem will eventually become trivial, not only the solution but also the failure modes, and full reading of the problem will no longer be required, in a more profound way.

As such, the discipline to read all variations not only improves the overall reading discipline but also and maybe more importantly, the overall understanding of shapes.

Author:  Knotwilg [ Thu Aug 13, 2015 11:51 am ]
Post subject:  Re: Knotwilg's practice

I have been battling the pathetic reports about blunders and playing without thinking, and then I do this:



Oh, I won the game all right.

Author:  ez4u [ Fri Aug 14, 2015 5:07 am ]
Post subject:  Re: Knotwilg's practice

After E19 it is seki, no? Black just does nothing. How does White play to remove Black's stones?

Author:  Knotwilg [ Fri Aug 14, 2015 9:17 am ]
Post subject:  Re: Knotwilg's practice

OK. Even that I got wrong. Butterfly lives.

Time to rewind.

Thx Dave (I think)

Author:  Knotwilg [ Fri Aug 14, 2015 5:40 pm ]
Post subject:  Re: Knotwilg's practice

After the (victorious) debacle against Ayabot, here's a better game against Hirabot. It's still full of mistakes but they do not vary as wildly in size and in the end it's a convincing win against a 2d bot.

One of the main lessons here is that it is hard to detach from a fight that previously was important for the overall balance but isn't anymore after a capture elsewhere.

So, after some stones have been captured, probably the essence of the game has shifted elsewhere.


Author:  Knotwilg [ Wed Aug 19, 2015 2:04 pm ]
Post subject:  Re: Knotwilg's practice

A human player this time.

Opening and early middle game not too great. Fortunately the opponent made a big mistake at :w58:

If you look at the :b57: - :w58: exchange you get an idea what I mean by stating that middle game techniques are all important. A pro would not see the difference between these two dan players and a beginner.


Author:  Knotwilg [ Thu Aug 27, 2015 8:28 am ]
Post subject:  Re: Knotwilg's practice

Looking back at the games I've played over the past months, most of the mistakes I made boiled down to some kind of emotion take over from proper analysis. Fear, desire, sloth (!), abound.

Somehow it is difficult to resolve to fundamentals.
Let me take this situation as an example.


FF to 163 (except for Bill, who can comment on my opening questions, well anybody can, even the kyu players (inside L19 joke))



The game is difficult to assess but White has 4 major territories, while Black (me) has taken a big group, a decent side and a big corner. In the upper right Black has several ways to reduce the corner, which is otherwise very big.

At this point I (think I) remember that I started realizing the game was not as favorable as I had been thinking, after the capture in the left of the centre. In fact, White had astutely sacrificed his group to conquer the lower side, and solidified his corners.

I didn't really count the position (sloth) but I reset my judgment of it and now had to deal with the possibility of losing this game (fear) which I had thought was in the bag (vanity).

I calculated what would happen if I tried to capture the marked stones (good). But I wasn't able to conclude on the liberty race. Instead of taking the safe path, which would anyhow reduce the corner, I chose the risky path (desire).

In the end I lost by a small margin but I shouldn't have lost this battle, especially since I had not calculated it to be beneficial for me and there being a safer path with smaller profit.

Author:  Knotwilg [ Sat Oct 03, 2015 1:53 pm ]
Post subject:  Re: Knotwilg's practice

After a long hiatus, let me post another game for analysis. It all went well until the atari blindness fueled by greed in the end. This sadly proves my point: even at 1d, one loses games by blunders. And one should use a period to calculate the score, which I failed to do, and which would otherwise mute my greed for sure.



Edit: looking again at the game in this browser, I could have killed the bottom group.

Author:  Shaddy [ Sat Oct 03, 2015 3:08 pm ]
Post subject:  Re: Knotwilg's practice

You would win if you played E14 instead of resigning.

Author:  Knotwilg [ Sun Oct 04, 2015 8:14 am ]
Post subject:  Re: Knotwilg's practice

Hmph ... Okay that takes the cake

Author:  Knotwilg [ Tue Oct 27, 2015 8:34 am ]
Post subject:  Re: Knotwilg's practice

Goals for November:

- play 10 games with time settings 5+5x30s
- don't resign, don't lose on time
- for all moves, choose between 3 alternatives and read a few moves deep
- slow & thick rather than fast & thin
- when surrounded/ing read out L&D situation and consume up to 3 periods
- if ko, think clearly about the ko
- consume 1 but last period at start of endgame to identify 5-10 biggest points
- try estimating the score
- review the game

Study:

- 1-5 pro games: end game study
- 30 Hitachi problems

Author:  Loons [ Tue Oct 27, 2015 11:32 am ]
Post subject:  Re: Knotwilg's practice

Why stipulate wanting to err on the side of playing slow? Not that it's a problem, that's mostly what I like to do.

Also, I like those time settings. Want to play? (This is definitely not just because I lost ridiculously in an automatch against you ages ago).

Author:  Knotwilg [ Wed Oct 28, 2015 10:06 am ]
Post subject:  Re: Knotwilg's practice

Thanks Loons.

I made an endgame study of a Gu Li - Lee Sedol game. It was really very interesting. Comments welcome.

Analysis starts at move 155. In the beginning all alternatives are my blind attempts.


Author:  Knotwilg [ Thu Oct 29, 2015 10:27 am ]
Post subject:  Re: Knotwilg's practice

Another Gu Li Lee Sedol game, analyzed for the endgame from move 143


Page 1 of 19 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/