Log in

No account? Create an account
Measuring progress
It's important to have some way to measure progress when doing anything.  With this project, the question is how to measure progress in reading ability?

Game results are a poor metric.  First of all, they don't provide that much information until you have a lot of games.  Also, if you are following the cavego rules, you're playing very few if any games.  Subjective feelings don't really count.

What is needed as a systematic way of knowing that reading has improved.

Fortunately, there is goproblems.com and SmartGo (which contains problems from goproblems.com).  Although this is not the perfect problem set, one advantage it is that the problems are graded by the number of players who solve them correctly.  This is a better metric than an author's subjective feeling for the difficulty of a problem.

What I do when I want to measure my progress is the following:

1. I launch SmartGo Kifu and select problems at the level I was last working on, say, 3k.
2. If I solve 3 problems in a row correctly, I self-promote.  So, if I was working on 3k problems, I start working on 2k problems.
3. If I get 3 problems in a row wrong, I self-demote.  If I started at 3k, I will shift back to unsolved 4k problems.

I've noticed that if I do this for a long period of time, my results get worse due to fatigue, so I don't do that.  Also, I don't want to exhaust the problems at a given level, so I use this mostly for measurement and not for training.

Currently, I don't set any particular time limit, but I do impose the rule that I have to solve the problem in my head first rather than just clicking around.

For many years, I have found that the measured level is very close to my playing level and changes in my ability to solve problems correlate directly with improvements or degradation in my game results.

Blind spot in sacrifice
I spent more time on the following problem than makes sense.  This should be a very easy problem for me:

Black to play

The reason is that I rejected the correct first move too quickly, because I read it out this way, which fails for black:

I had this fixed idea that when white plays atari at 2, black has to capture at 3.  But the real purpose of black 1 here is to prevent white from playing atari at 3.  Black can give up those stones to save the corner ones:

So although I'm used to adding a stone to a sacrifice to get sente or make shape, I wasn't thinking of those three stones as being sacrifices after playing black 1.  Sacrificing more than two stones is sometimes really important in go problems and in real games.  Under-the-stones problems always involve larger sacrifices, and if this problem had looked like an under-the-stones problem then I may have considered the sacrifice instantly.

Missing approach kos
Because it's not extremely common in games, it's easy for me to miss approach kos in normal reading.  Consider this sequence:

Here, white 1 looks like a vital point, but playing it directly is a mistake.  The reason it's a mistake is that after white 5, black can play at C to make an approach ko.  The fact that white can make a real second eye if black makes the mistake of connecting at A or B is something that is easy to get attracted to, but of course one has to read the strongest moves for the other side.

If I were presented the diagram shown above with the caption "black to play" of course I would see the approach ko at C. But seeing that in my head after reading out 1-5 is another matter.  Stronger players are more used to that first-line shape would be much less likely to miss it.

(The correct solution is to exchange A for B in sente and then play white 1.)

While it is great that I saw white 1 as a vital point, it's too early to congratulate myself and just play it.  It's easy to get really happy about seeing that, but one has to take care to play it with the right timing and verify that it does indeed work.

Characteristic depth
Well, it's been a while.  As they say, I've been busy.  I'm trying to get back into playing go more.   One of the things I noticed in games I've had reviewed by professionals in the past month is the depth of reading required to resolve even the simplest-looking errors.    Somehow this is not as easy to notice when I am shown variations interactively.  In that case, it's too easy for me to get into head-nodding mode, and think I understand something that I don't.   I may believe what I'm being shown, but that's very far from being able to apply that knowledge in a real game.  Now I've taken to looking at the game tree in reviewed SGF files, and before I start looking at the content of the variations, I just expand it and look at the depth of the variation tree added by the pro.  It sounds silly, but it's sobering.  In order to answer simple questions like: "does this cut work?" or "can I capture this stone?", variations are shown that are anywhere from 9 to 15 plys or more. This isn't just being fancy.  I've looked at these variations and it's actually necessary to read them in order to start playing those lines.  There are a lot of missed opportunities hiding around that depth.

I don't read that deeply in games.  I can't even read like that in problems.  Granted, my games also contain 3-move misreads that are worth preventing and probably can be categorized as blunders.

My losses recently are still mostly explainable by reading problems and time management, though some other errors in thinking exist and are significant.

Have I improved at all?   Maybe, maybe not.  As I was going through the Gokyo Shumyo again, I hit this problem again.  This time, I saw black 1 instantly.  In the past year, I'd seen this tesuji a few times in problem books.  However, when I read out the line shown, in my head, the marked white stone looked like an atari, so I rejected black 1 and wasted a lot of time on other ideas.

I'm happy I saw the key concept, but confidence in validation is key.  A tesuji without the correct follow up is just a mistake.   Sometimes just knowing when the image is fuzzy in my mind and taking a little time to clear it up will help.  However, in a timed game, it is easy to run down the clock that way and get into time trouble later.

Depending on how you look at it, this problem has a characteristic depth of anywhere between 5 and 8 moves.   The solution at Sensei's library stops at black 5 and if you are looking at the diagram at that point, maybe it's possible to just see that white can't do anything.  I think I would have to read (correctly) to 8 because the image at 5 would not be that clear in my mind.

I'm going to start paying more attention to characteristic depth in diagrams that I see, and try more to appreciate whether the 1st move is one that is probably okay even if I can't read it out (which is true for some cuts) or whether the 1st move is dangerous or bad if it doesn't work.

Small diagrams and reading

Hello!  I'm back in the cave again.  It's a little hard to remember where I left off in the Gokyo Shumyo, but actually it doesn't matter because after that much time it's still time for review.

Before I resume my postings on specific problems, I'd like to take some time to discuss something that I meant to discuss earlier.  Last year I noticed that problems books with small diagrams (stones approximately 3mm in diameter) are a litlte hard to read.  My optometrist would say  (and already has said) that I'm just getting old, but I think there's more to it than that: even with reading glasses I think larger diagrams make the same problems easier to visualize and remember.

I got to thinking about what the angular size of a go stone is when sitting at a goban, as compared to what is printed in books.  There is no need to make precise calculation, but if a go stone is 22mm in diameter and about 50cm away, that's like a 14cm stone at a reading distance of 35cm.  That's pretty big!  Diagrams like the ones in my copy of Davies' Life and Death (3.4mm diameter stones) are not necessarily hard to see, but when I'm visualizing move sequences, it just seems I'm working in a really tiny space.

In contrast, some Korean books have stones as large as 6.8mm, and that brings them a lot closer to the goban experience.  (Maybe you can call it the kibbitzer's view. )  A typical larger-diagram book like Janice Kim's Learn to Play Go series is about 5mm, and I find that comfortable.

Computer screens vary a lot, and it depends not just on the display but also on the software you are using.  (On of my main complaints about the GoGoD95 reader is that it doesn't scale the board when you resize the window.)  On a large display, you can play a 19x19 game with stones that are close the angular size you'd find on a goban, though.

So what do we do with small-diagram books?  Some people copy the problems on the goban, but that is time consuming, especially with simpler problems.  I suppose one could use a photocopier to enlarge the pages, since most of the small-diagram books are pocket-size.

But there is yet another strategy, and this one is interesting: you can glance at the diagram for a few seconds, then look away and see if you can reproduce the diagram in your mind.  This is akin to the technique of some chess players who stare at the ceiling or a blank wall rather than at the board.  For them, I guess the idea is that the existing pieces clutter the mind.  In go, the board doesn't change as much as it does in chess, though.  Maybe when you are doing under-the-stones problems it might be a good idea not to look at the diagram.

For some of the problems in the Gokyo Shumyo, I can hold the image in my mind.  Here is one that Janice Kim mentioned might be a good one to try:

Black to play

If you can hold a diagram like that in your head, you can think about it wherever you are.  You could be waiting in line for the bus or sitting through a boring meeting and still be able to improve your go.  The key seems to be to pick a diagram that's not too complex to remember, but has enough variations that you can't exhaust them in your head instantly.  Stronger players should probably just know the main line of the problem pictured, but then the exercise would be to find the best refutations for all the things white might do.  What if there are other stones on the board or some of the liberties pictured here aren't filled, for example?

But back to my original topic, when I first started trying to do this, I found that the diagrams I was visualizing in my head were about the same size as the ones I saw on paper.  If I initially glance at a big problem, it's easier to visualize.  If I start with the small diagram, enlarging it in my head seems to take little extra effort.

Emerging from the cave..
...well, at least for a week or two, while I enjoy actually playing some games.

I don't really feel I've gotten anywhere.  Maybe I solved 200 of the 520 problems of the Gokyo Shumyo, but I certainly haven't been doing it every day, and some breaks have been too long for me to believe that I'm getting any benefit.  It's a lonely business.

I've grown to really like some of these problems, though.  I look at some of them, and think, "ah, that's impossible, there's nothing there."  And I then look again, and again, and still nothing seems good.  Then suddenly, something new, something magical appears in a slightly suspicious shape, and then, there's that rush, that feeling of "maybe, just maybe."  It's a flash, a fuzzy specter of a variation, a chill of something lurking out in the darkness, patiently waiting---devastating, peculiar and barely perceived.  It exists outside of the normal proverbs, invisible to the clarity of forced sequences and vulgar habit.

I will have to come back and try again...and again and again.

Learning from mistakes

One of the things I'm doing differently while studying these problems is that I'm trying to learn more from my mistakes.

Take the following problem, for instance:

White to live

I keep track of my "first feeling," which is the move that springs to mind without any thought or analysis.  For this problem, I wanted to play this move:

It's just locally a basic instinct.  However, it's quickly refuted.

The next thing I thought is to look at connecting other the triangled or squared stones in the following diagram:

There are some interesting threats here, at least.  Starting at the bottom didn't seem to get anywhere, so using the threat to connect to the top must be tried.

The following diagram is the main line of the solution:

Notice that after white 11, black cannot push through at the bottom, because he doesn't have enough liberties.  This is important to see  in the problem, and it's one of the hardest things for me to visualize.

Here's a wrong variation that I considered, and you can see how this way, black can push through and kill three white stones (and therefore make life), all because of one extra liberty:

So at least in this line, white 1 here doesn't work in part because of that extra liberty.  (There may be other reasons, but my point is that seeing these liberties is important.)

What did I learn from this mistake?  I'm not sure, but the quick take might be:

Fill a liberty

Fill a liberty!

Don't be fancy!

Priority and urgency
First, the main line of the solution from my previous post.  White 1 is the key point.  If black plays 2 to prevent white from easily living on the left, white 3 is a tesuji to connect.

But that's not the point of my post today.  Obviously I haven't posted in a while, and so the question at hand is why that happened.  I have good news and bad news.  The good news is that I didn't fall "off the wagon" and break the cave go rules.  The bad news is that really I didn't post because I didn't work on go at all recently.

I have various excuses of course, the little lies one tells oneself at times of failure.  I can make up as many as necessary:

"I had a lot of work to do at my job and that's more important."
"My family commitments got in the way."
"I couldn't find the time."
"I got interested in other things temporarily and used my free time for that."

You'll note that almost any of these statements can be used to justify almost any kind of failure in a goal of action (i.e, one that requires doing something rather not doing something.)  I call these "universal excuses."   Probably everyone has their favorites.  If you know the people around you well, you may even know which ones work best to get sympathy with a particular person.

That's all very well and good, but one can't escape the feeling that this is kind of a loser mentality.  Universal excuses are really dangerous.  Probably they should be banned.  They sometimes work because they are nominally true or at least plausible, and they tend to dissuade further inquiry.  But for someone who is trying to achieve a goal, they aren't helpful.

What is needed is to get to the bottom of the real issue.

A lot of people don't achieve their personal goals because they confuse priority and urgency.  This is because urgent situations usually manifest themselves as external pressures.  Someone is asking you to do something now.  It can't wait.  There also the time-sensitivity that adds to the pressure.  There is an imminent deadline for some commitment you've made.  The consequences of missing it are obvious and immediate.

The funny thing is that "urgent" items don't often match up with priorities.  Personal health is a common priority people have, but it doesn't become urgent until some dramatic event occurs, like a heart attack.  So people ignore it and make excuses like, "well I don't have to exercise today, but I do have to complete this report today."  That may be true, but that's a universal excuse, isn't it?  You can use that every day.  But if you use it frequently enough, sooner or later you have to admit you're acting like health isn't a priority.

The truth for me is that I fell into this trap.  If I'm honest with myself, becoming stronger at go is more important to me than many of the things I've actually been spending my time on in the past couple of weeks.  But no one external is pressuring me on it.  It clearly isn't urgent.

Yes, work is more important to me than go.  I certainly don't want to lose my job in a bad economy.  But not everything at work is important.  Sometimes I create work for myself by taking on ill-defined projects with unclear charters or sponsorship. I say "yes" to requests I could legitimately reject or pass on to others.  A clear focus can buy time.

Family is a little harder for me to analyze.  It's literally "too close to home," I guess.  I definitely have a feeling that if I do everything that "needs to be done," then I would never be able to do anything else.  Also, I would never finish it all, anyway.  It's still difficult to say no.

There is a proverb in go: "play urgent moves before big moves."   Some players say it is in fact the most important proverb.  If one takes this as a metaphor for life, it would seem to contradict what I am saying here.  But I don't think it does.  The proverb sounds simple, but the problem is that it's actually very hard to see the urgent moves.  To some extent, the ability to see urgent moves is go strength. In life, too, things sometimes appear urgent that are not.  And things sometimes seem big that are not, as well.  A misunderstanding of these things in go leads to a pathetic game, playing small moves in gote.  A misunderstanding of these things in life can be more serious.  On one hand, there are truly urgent situations that are life-threatening.  But there are also demands that masquerade as important that are not.  If one gives in them, then that leads to a passive, unfulfilling life.

Anyway, I'm far behind in the Gokyo Shumyo goal and will fail to complete it.   But I will keep at it, anyway.

Seeking a realizable plan..

White to play

There are many problems in the Gokyo Shumyo that require one to find one or more dual-purpose moves.  The above problem is one example.  It seems white cannot make two eyes in place.  That's true.   And white cannot connect to the two stones on the right, either.  If your goal is to do one or the other independently, then you will fail.  But if you look for a move which aims to achieve both options, you can easily spot the tesuji.

There is always a tendency to read with some sort of single-minded objective, but when I listen to stronger players discuss their games, the don't seem to think that way at all.  Some very complex reading that a weaker player would find discouraging because it fails to achieve immediate success is still useful to the stronger player, who is able to chunk that information.  So they will say things like, "well, this move is sente against the corner, so that's useful, but first, white has to make this exchange, etc."  That's a totally different level of thinking.  This sequence that a weaker player (who is still able to read) dismisses as "doesn't work," is just a piece of a larger puzzle for the stronger player.  Stronger players are able to chunk at a higher level.

If it is not complete heresy to quote a chess player:

"It is not a move, even the best move that you must seek, but a realizable plan” (Eugene Znosko-Borovsky)

I have to agree.  Some of the problems in this collection I can solve on sight, but others torture me with endless branching.  There is no way for me to make sense of them by extensive search or by randomly trying out cool-looking moves in vain hope that one is a tesuji.  I have to try to chunk, and look and the position and try to see what's really going on.   Then maybe I start to see things. The moves have wishes of their own, it seems.  "If only I could be played in sente," they seem to call out.  I don't know if they can, but what if they could be?

Knowledge is power

My original post seems to disparage knowledge a bit, but I have to concede that some knowledge is useful.  In particular, knowledge of basic shapes is a big time-saver.  Consider this problem (which I consider to be quite beautiful.)

White to play

This is a problem I was able to solve more quickly than some others because I immediately recognized the possibility of creating a basic under the stones shape.  If I didn't know those shapes, the key idea in this problem wouldn't spring to mind.

Other problems in the collection are made easier if one knows something about the door group, the comb formation, the L group, etc.  These are all covered in Davies' "Life and Death", a book I appreciate more and more as time goes on.

Anyway, it is July 4th, and I have to get back to spending time with my family.