dfan's quest for competence

[WriterJon]

My confidence is largely based on one postulate, which I have no proof of: 1d is not really all that good.

I don't mean that it's easy to get there! It's obviously quite hard! But the more I play and watch and talk to dan-level players, the more I think that 1d play consists largely of good fundamentals, good reading, the lack of blunders, and the willpower to play at one's best all game (all things I need to get much better at), rather than, say, superhuman talent.

This is more or less my thinking in my own goals (I'm a lot further off than you). How does this compare to your previous idea that a western chess USCF 2000-rating is a similar sort of thing to a KGS 1 dan rating?

[dfan]

My confidence is largely based on one postulate, which I have no proof of: 1d is not really all that good.

I don't mean that it's easy to get there! It's obviously quite hard! But the more I play and watch and talk to dan-level players, the more I think that 1d play consists largely of good fundamentals, good reading, the lack of blunders, and the willpower to play at one's best all game (all things I need to get much better at), rather than, say, superhuman talent.

This is more or less my thinking in my own goals (I'm a lot further off than you). How does this compare to your previous idea that a western chess USCF 2000-rating is a similar sort of thing to a KGS 1 dan rating?

I think the statements are very compatible! I think of my USCF 2000 rating as indicating fundamental competency in the field of chess, rather than brilliance or genius. So in that respect I don't think 2000 is "really all that good". On the other hand, that doesn't mean that achieving that level is easy; it takes a lot of work and practice to achieve competency in a complicated domain.

[dfan]

I spent the last couple of years concentrating on chess rather than go. Chess has always been the game that I'm most fluent at, and I had burned out a bit with go. It's interesting to reread my update from December 2020 because honestly I didn't really remember that playing go had become more stressful and less pleasurable, but looking back, that was definitely the case. Anyway, concentrating on chess again for a while was fun; I found a supportive community where I enjoyed helping others improve. I did put some real elbow grease into it but I know that my best days are behind me, and that's fine.

One interesting thing that happened during that time was that I discovered that the International Master David Pruess has aphantasia (the inability to form visual mental images), like I do, but can still play multiple blindfold games of chess simultaneously. Previously, I had always assumed that blindfold chess was just inherently impossible for me. So I spent some serious time over a couple of months diligently improving my "visualization" skills, and by the end of it I really could follow an entire game from beginning to end just by reading the notation. Actually playing a game blindfold is more difficult but I can stagger through it. It was really rewarding to develop a new mental skill at my relatively advanced age (I'm now over 50), especially since I had always been sure that it was beyond me.

Now I'm back playing go again, because I always eventually return, and I have a new attitude about reading. My reading in both chess and go has always been mostly a matter of "flow"; I know how positions transform into each other so I'll start a sequence knowing that it has got to end up with a good result rather than doing the hard work of tracking every piece/stone. You know how it is with a throw-in/shortage-of-liberties problem or smothered mate where you don't need to laboriously read it out, your neural net evaluation function just sees that it must work. I am "reading", not just playing with a pure policy network; I do actually follow the sequence of moves, going "doot doot doot doot doot", but I don't really keep track of all the stones, I just know from eyeballing the shape that the usual pattern of moves will work.

It's nice to have this intuition but at some point these games come down to reading, and actually knowing where everything is N moves into a variation. As with chess, I always assumed this was basically impossible for me in go past a certain depth; now I think I just haven't tried hard enough yet. So in addition to playing, I'm doing a lot of tsumego and really concentrating on actually tracking everything precisely.

So far it's been rewarding. I just finished a run through 1001 Life & Death Problems and it really felt a lot less punishing than last time. Of course those problems only go up to 5 moves, but still, you have to look at multiple variations that branch, count liberties of future strings, etc., and I felt like I was doing that more diligently than before.

In the meantime I've been playing on Fox again and doing well there, although as usual after a break it's hard to tell if the ranks have shifted. I'm not really trying to improve my rank anymore - I feel like 2016-2019 was my chance to do that, and was only a limited success - but I still feel like I can do productive work on honing my skills, which brings me pleasure. So I'm explicitly avoiding having any rank goals again but I am trying to accomplish things that are totally under my control, like finishing tsumego books and playing games.

[dfan]

One of the things I've been doing is to keep careful track of my tsumego performance. It's motivating to keep checking off problems (slightly less motivating are the Xs for failed ones), and I'll be able to compare my performance in the future, or specifically come back to problems that I've missed in the past. I might as well put those stats here:

Weiqi Rapid Drill 800 Problems (review): 728/800 = 91.0%

1001 Life-and-Death Problems:

• 1-move Black to live: 169/200 (84.5%)
• 1-move Black to kill: 193/200 (96.5%)
• 3-move Black to live: 175/200 (87.5%)
• 3-move Black to kill: 99/100 (99.0%)
• 5-move Black to live: 84/100 (84.0%)
• 5-move Black to kill: 161/201 (80.1%)
• Total: 881/1001 (88.0%)

I remembered finding the 1-move problems kind of annoying in the past, and they were again. In my opinion a lot of them aren't really 1-move problems, they're longer problems where you're only required to present the first move of your line. As you can see, I actually did better in the 3-move problems. The 5-move problems are where the fun starts, and I would hope to get 90% the next time I go through them. Honestly, there was only one problem where I threw my hands up (it turns out the correct solution ended in a double ko); when I failed, I'd say it was about 50% of the time due to carelessness (forgetting the capturability of stones in the corner was a common one), 25% due to lack of rigor in my reading (e.g., I had the right first move but didn't consider an important variation - I marked these wrong), and 25% due to actual reading failure. This felt a lot better than the first time I went through the book, when I found a lot of the problems pretty brain-busting.

I also remember being somewhat put off by the artificial nature of the problems the first time I went through the book, but it didn't bother me this time. I'm not sure why that is. Maybe I'm learning to enjoy reading more.

[EricBackus]

1001 Life-and-Death Problems:
...
I also remember being somewhat put off by the artificial nature of the problems the first time I went through the book, but it didn't bother me this time. I'm not sure why that is. Maybe I'm learning to enjoy reading more.

For me, the fact that the problems are artificial means that I have to read out each problem, rather than just "know" the answer. It's reading practice rather than "recognize common shape" practice. I credit this book with helping me out of a rut at around 5k.

[dfan]

Indeed, I appreciated this at the time, and went through the book because it was good for me, but I still wasn't happy about it. This time the artificiality somehow stood out less to me. Maybe I've just done a lot of other tsumego in the meantime.

Meanwhile I was just going through a Jump Level Up book for some easy practice, and I think that the themed nature of the book, which is rightfully generally seen as a feature, is actually a bit of a negative point for me. If I see that a chapter is all about belly attachments, say, it's too tempting to glance at every problem and see what the answer (or first move, at least) must be. I still try to check my reading, but there's a lot less incentive to be diligent in the way I was describing above. So I think that these books that are organized by pattern are not good for me right now. I know all the fundamental patterns up to a certain level; what I need to work on is applying them accurately, reading out when a tempting move fails by one liberty, etc.

[Kirby]

I'm not really trying to improve my rank anymore - I feel like 2016-2019 was my chance to do that, and was only a limited success - but I still feel like I can do productive work on honing my skills, which brings me pleasure.

Hey, dfan! It's nice to hear from you. I am glad you are enjoying reading more.

The quoted statement above made me curious: while it is perfectly cool not to focus on rank, what was significant about 2016 through 2019 that's different than now?

[dfan]

Hey, dfan! It's nice to hear from you. I am glad you are enjoying reading more.

The quoted statement above made me curious: while it is perfectly cool not to focus on rank, what was significant about 2016 through 2019 that's different than now?

Oh, it's just that that was a period when for various reasons I made a big concerted push to make a quantitative improvement to AGA 1d or so. And I did improve a lot, although my AGA rating doesn't reflect it - my online ranks are a lot stronger, I beat 1ds occasionally in casual games, and I know my knowledge and judgment have gone way up. But my relationship with go by 2019 or so became mostly about trying to prove that I could be as strong as I should be (whether that was rank or playing a praiseworthy game on AYD), which meant I was too invested in my results, which made me tense and frustrated. So now I am making an explicit effort to just enjoy the game. Enjoying still means thinking hard and learning and studying, but I'm trying hard not to grade myself by my results.

[Kirby]

Enjoying still means thinking hard and learning and studying, but I'm trying hard not to grade myself by my results.

Gotcha. That is difficult for me to do. I am glad you are finding a way. These days, I am having a difficult time deciding how much to focus on work vs. go. Doing well at work is important for having an income, but I find more happiness studying go.

[dfan]

I did a new pass through the Jump Level Up books as a refresher. I didn't keep track of how well I did when I first did the books, but this time my scores were 95.5%, 94.6%, 95.3%, 94.8%, and 91.9%. So only the last book had appreciably harder problems (or maybe I was paying a little less attention).

I like books like this in principle (a prescribed graded syllabus in which each topic is presented and then drilled before moving on to the next one) but they're the wrong kind of book for me, at least right now. I already know all these SDK-level techniques (at least in theory), although getting a refresher was nice. And because the problems are grouped by technique, you usually know what the correct key move is likely to be before even bothering to do things like count liberties, and although of course you should still read things out, I lose a lot of motivation to do so. I could feel my reading skills atrophying as I went through these ~3000 problems, and it was kind of a relief to return to a regular tsumego book where I really have to work at the variations.

Meanwhile I've been playing on Fox. I had dropped down to 1d during the last year of playing an occasional game but not thinking much about Go, so that's where I started, but I went 16-1 at that level and double promoted to 3d. That can't all be due to my immense skill; I think Fox ranks have gotten softer. But one thing that I did feel like I was doing better this time was reading, and having confidence in my reading. If I think I've read out a local situation correctly, I trust it. This can occasionally lead to disaster but you learn more from that and getting actual negative feedback than from overprotecting things.

I also have acquired a new strategy of physically putting my hands in my pockets until the 10-second byo-yomi countdown starts (these are 30-second games), no matter how obvious my next move is. There's always something to look at. I'm sure it's annoying to some opponents but I definitely am noticing some opportunities that I would typically miss.

Most importantly, I'm enjoying playing.

[jlt]

It's possible that Fox ranks softened a bit. I got promoted to 3d last month after being stuck at 2d for 21 months (and still stuck at 2k+ on IGS for 18 months).

[dfan]

My sense of what activities are effective for me personally continues to improve. As I started to relearn some of the knowledge I had let slip during my break (such as life and death statuses and joseki) I found my play slipping a bit - playing by shape and knowledge instead of by reading. So I went back to 1001 Life-and-Death Problems to remind myself that it's fundamentally all about reading in the end. My results were a little better than a few months ago, for what it's worth:

Code: Select all

           9/2022  2/2023
Live in 1   84.5%   88.0%
Kill in 1   96.5%   96.0%
Live in 3   87.5%   93.0%
Kill in 3   99.0%   97.0%
Live in 5   84.0%   83.0%
Kill in 5   80.1%   87.1%
Total       88.0%   90.8%

I managed to hold on to Fox 3d in my first period of games by the skin of my teeth, but it's not unlikely that I'll drop down to 2d at some point, which is fine. I may not totally deserve to be a Fox 3d yet, but I deserve to play games with Fox 3ds, if that makes sense.

[hl782]

It's fascinating to me how your accuracy rate is rather consistent throughout the book. I remember clocking in around 10%+ accuracy higher on the 5 moves to kill than the 5 moves to live problems.

[John Fairbairn]

As I started to relearn some of the knowledge I had let slip during my break (such as life and death statuses and joseki)

I wonder, I wonder. Do we learn them properly?

I was reading a Chinese book last night on calligraphy. It is not just in go that there are startling differences between the words or the mindsets that Chinese compared use to us. In the case of calligraphy, western books tend to talk about beautiful Chinese calligraphy in terms of the final shape, with often an accent on symmetry. Like urban architecture of the tower-block variety. Any deviation from symmetry is then deemed strange (though possibly wonderful). But symmetry is a side issue to a Chinese calligrapher, and is usually rather rare.

What this book talked about instead was the beauty of characters coming from the "shi" (power, or what we call influence in go) of the various components - which obviously varies according to position. These "power" points then have to be joined up by "techniques" (fa) of writing and combining strokes. This instantly put me in mind of the similar common discussion of shi and fa (in the form of zhaofa) in old Chinese go.

The calligraphy book went on to explain that attention has to be given (under the heading of technique) to the beginning and middle and end of each stroke (once the various shi elements have been mapped into the writer's consciousness).

You could dismiss that as sheer coincidence, but it resonated with me because I have been lucky enough to stand beside famous calligraphers as they have written large characters, and when you do that you can literally feel they are writing the beginning, the middle and the end of each stroke - not quite separately but with a change of pace over the whole stroke. That change of pace shows up in the final stroke. This process makes the writing of each stroke ever so slightly - but noticeably - slower than the way a western writer writes characters, where a quick, showy slash is deemed skilful. Furthermore, when the Chinese writer writes in cursive or running hand, which looks on the page as if it has been dashed off at high speed, he actually writes rather slowly with a perceptible b m and e.

When you read well-written script you can follow the writer's movements in each stroke and so deduce the sort of feeling he is trying to imbue.

I spent some time after waking up this morning thinking about whether we should be thinking about zhaofa (or Japanese suji or Korean haengma) in these same BME terms. And I'd be quite confident in asserting that next to nobody here does. I certainly don't. Most people just go straight to the middle part and try to find the tesuji. If they can't they MAY go back to the beginning and look harder at the problem. But even when they do find a tesuji that works, do they do the end bit, too? Check for weird variations and/or think how this sequence can combine with other sequences (or power bases)?

But if we did do problems (life & death, tesujis, josekis) on that "full English breakfast" basis, what would the process look like? I suppose the beginning part would be recognising the various shapes and weak points. The middle is the actual tactical sequence. The end part might be thinking about "where do we go from here?" as described just above.

My own thoughts haven't settled down yet, but I'd hazard a strong guess that the apparently slow BME process would end up a lot faster (rather like the alpha-beta algorithm) because it is more efficient, pruning out lots of unnecessary lines.

But also (and here I revert to the quote above) this step-by-step approach would surely make it much more likely that we learn the sequence more securely. Just as a full English breakfast keeps you going a lot longer than a bowl of cornflakes. And since you would largely be understanding it better, this understanding would make other techniques much easier to learn.

Your thoughts?

[xela]

Interesting thoughts as always John. I remember going into an art gallery in Melbourne where they had not just a display of calligraphy, but also a video of a calligrapher at work. It was quite an education.

Remember that this forum is a gathering place for go-playing amateurs! I think what you're seeing in our approach is not an intrinsic weakness of "the western mind", just the dallying of dillettantes. I'm glad you haven't given up on us, and appreciate the continued nudging.

There's also the question of how to work efficiently as a part-timer. If you're not studying go eight hours a day, but struggle to make even eight hours a month, is it better to spend an hour studying five shapes in depth, or to get a superficial exposure to 50-100 shapes in the same time?

Did Behind the scenes of some endgame tesuji move us in the direction of a full English breakfast at all?