It's been a while since I checked in here.

My primary go activity continues to be Hwang In-seong's American Yunguseng Dojang; I just finished my sixth three-month season. I really can't recommend it highly enough for people who have the time, energy, and money for it. My day-to-day interest in go has waxed and waned but knowing that every week I will have a serious game reviewed by a top-flight teacher provides a really great baseline of activity and has kept my motivation flowing.

I still watch lectures and do problems from Guo Juan's Internet Go School. There's a ton of excellent content and the spaced repetition system is great. One small issue is that her recommendations, in joseki for example, sometimes conflict with In-seong's (he tends to be more on the cutting edge of theory), but I can generally handle the differences.

I also have my own spaced repetition database in Anki, which right now consists chiefly of flashcards of mistakes in my AYD games and Essential Life and Death (I think these problems are particularly well suited to spaced repetition; they tend to be more about techniques and patterns than reading).

Meanwhile my tsumego/tesuji practice has been going through these Chinese books (going through the 1d ones now) and "Speed Baduk" 13-15. The books have gotten harder so I've slowed down. I've been thinking of backing up and getting more of a running start.

I definitely feel like my competence has increased over the last year. According to my AYD results my strength is now about 2k and that fits with the friendly games that I've played with local friends. But of course that's not the same as real tournament play. I did play in one AGA tournament since the Go Congress; I entered as a 3k and went 3-1, beating two 4ks and a 2k and losing to 1k (by half a point!), all in even games, which was enough to boost my AGA rating to 3k, so that's what I plan to play at in this year's Go Congress. I hope I'll see some folks from here there, especially ones that weren't able to make it last year.

I didn't do a Go Congress diary this year but a lot of my experience can be found in Kirby's journal! It was very nice to meet him and others, especially since most of the crew I came with the previous year wasn't present this time, for various reasons. I did my best to overcome my usual introversion and was very happy to hear Kirby tell me "You seem pretty social, for a go player."

I went 3-3 in the tournament, which I suppose means that I met expectations, but I was planning to beat expectations so it felt like a bad result. But that's okay. One thing I have learned is that if I want to increase my rank, it's not enough to play evenly with 2ks, I also have to be able to crush 4ks. I did get all my games reviewed and learned some good lessons from them.

Another conclusion I have come to (I've probably made it here already if I look in the history!) is that the biggest thing holding me back is not my knowledge but my reading and practical decision-making. The most important things I can do to improve is not to study or learn more but to play more (always my weak point) and do tsumego to improve my reading. Although I made all sorts of mistakes in my Congress games, they were often due implicitly to my reading, because I wasn't willing to go for the sharpest (but best) line or even just because I used up lots of time early in the game painstakingly reading things out and then made suboptimal moves in byo-yomi. So I'm trying to play more online and diligently do tsumego, especially problems that really require precise reading rather than mostly recognizing and confirming patterns.

As a result of playing more online, I just reached 1d on Fox, having recognized that the most important thing one can do to achieve 1d is to choose the right server to do it on. I realize that it's not close to the same thing as an AGA or KGS 1d rank, but it's still nice to see it on the screen. It's also kind of nice that players there tend to be big fighters and not play the regular moves in the opening that I'm used to from AYD, so I'm thrown on my own resources more.

I found that getting from around 4k to 1k on KGS was exactly what you mentioned--being more practical. I often ask myself, "do I need to take this fight? Or can I just play simply and win in endgame?". Winning against a weaker player isn't about crushing them, through the moves you play, but rather crushing them by allowing them to make mistakes. What I am trying to say is: play patiently and capitalize on the errors of the weaker player, rather than trying to overplay. I look at getting stronger as making less huge errors in a game. So a 4k might make 5 large mistakes in a game. To get to 1D maybe you make only 2 large mistakes per game.

Congratulations on Fox 1d. I play on Tygem so I can see the 2D, even though I play at 1K on KGS.

Hmmm. My experience was the opposite. Early reading enabled me to play quickly later on, because when play resumed in a region of the board I already understood it fairly well. How soon do you go into byoyomi?

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Like, what stage of the game? Except for one game where I spent a ton of time ensuring I could kill a big dragon because the outcome was clear after that, I generally hit byo-yomi in the early endgame. It still would have been nice not to hit it at all, and I lost one game due to a shortage-of-liberties situation coming up that I didn't have time to read accurately. (Yes, of course I should have seen it coming.)

I thought you might hit byoyomi earlier. A lot of people are used to relatively short main time. That's common online, I think.

Well, everybody's different.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

The US Open time control is 90 minutes plus 5 periods of 30 seconds. I hit byo-yomi in every game, and only one of my opponents did; in fact, most of them used fewer than 30 minutes for the whole game. It made me feel a little silly to be sitting there thinking for so long, but they gave me the time, I'm going to use it! Coming from the chess world I'm used to having 90+ minutes to play all my moves.

One thing I do need to improve at is concentrating at the end of the game and continuing to look at opportunities for both sides over the whole board. In a chess game, usually by the time you've gotten to the end of the game, there is just one narrative thread going on, but it's quite the opposite in go. That damezumari game I lost might have gone differently if I had looked more carefully at that area during the preceding play.

I did somewhat poorly this year in managing my time. One thing that caused me problems, I think, was that I was a bit antsy. Since I was nervous, I made several trips to the restroom, and got up often to get a refill on water or coffee, even if I still had some. In one game, I even went down to the cafe and bought a coffee while on the clock.

I rationalized this by telling myself that I was helping myself to calm down. But in reality, I think I was wasting time.

_________________
be immersed

Interesting point.

One thing that struck me in the write-up by the Elf team is that they pointed out that Elf started off by getting good at the endgame. That makes sense, because with self play it did not have any opponent who played the opening or middle game well to imitate, unlike humans. However, that was still a surprise, since the endgame is typically the worst phase of the game for bots. And while Elf avoids mistakes of weaker bots in the endgame, it still makes strange suggestions in review.

Elf, and other bots, do not take advantage of the fact that go endgames break up into a number of independent games, each of which can be analyzed separately. Even when the games are not independent, often when you read them out, at some point they break up into independent regions. The main problem for humans, then, is when to switch between independent regions. Play evaluation helps a lot, but only provides a heuristic. Bots use whole board evaluation, and, as a result, get good at switching between local regions of play, even when they do not know best play in each region.

If a human player has analyzed a number of different regions of the board before entering byoymomi in the endgame, that player will be pretty surefooted in the endgame, unless a ko breaks out, which perforce is a whole board contest, or if the opponent plays tenuki enough to reach some unfamiliar local positions.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Chess GM Ben Finegold ( https://www.youtube.com/watch?v=u0ai_3hqZ60&frags=wn ) tells of a game in which David Bronstein, in his later years, took 15 minutes on his first move, and later lost on time in a superior position. Finegold, who was in the same tournament, asked Bronstein why he had taken so long to play his first move. Bronstein replied, "I was calming down."

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I had a friend who would do this in long online tournament matches. Say he had 45 minutes thinking time. He would go to the store and buy food. But he said he would do it to psych out his opponents. He would often leave in complex positions.

2020 wasn't much of a go year for me. Even before the pandemic turned everything upside down, I had been feeling a little frustrated with go; I was at a plateau, and every game I played felt very fraught, like it was a make-or-break chance to prove I could actually play well. This was particularly true for my Yunguseng Dojang games, where I would both feel bad for "letting In-seong down" and bristle internally when he made some general comment about my play that I didn't feel was fully deserved.

So I took a break from go and went back to chess for a while, where I'm a better player and I don't get stressed out playing games, online at least; I know that I already got as good as I'm going to get and I can just have fun exercising those skills, winning some and losing some. During this time I was still doing some go activity: daily spaced repetition, watching videos (mostly Baduk Doctor), occasional 9x9 GoQuest games for fun.

Then a few weeks ago the itch to play some more go came back, I'm not even sure why. So I brought a tsumego book back to my nightstand, and played some just-for-fun games on Fox... and won my first three games in blowouts, ranking up to 2d. So, great! Another notch on my belt. But I want to just be glad I got that achievement and not worry about keeping it.

One thing that was interesting in all these games is that my opponents made some mildly weird moves in the opening (nothing crazy, just direction of play that I knew wasn't optimal), I lost some respect for them and assumed I could push them around... and I did! In-seong's main criticism of my play has been that I play too safely and softly, but against these opponents whose sense of direction of play I didn't trust, I was happy to tenuki all over the place and attack without mercy where I was strong. So I think that assuming my opponent is weak (or at least assuming my opponent's play is weaker than the play I am capable of ) really helps my game. Of course overconfidence is not a good thing, but I need to swing back from underconfidence first.

(Or maybe Fox ranks just got softer during my break. Shhh, don't spoil my dreams.)

I'm also hoping that the "take a break, forget bad habits" rule has worked for me a bit too.

So my goals for 2021 are simply:

• Have fun playing go. It's a game! If you lose, maybe you'll win the next one.
• Play as if I'm having fun! Will this attack work? I dunno, let's find out!
• Assume my opponents' moves are bad unless proven otherwise.

Not that I want to spoil anyone's dream but I also mentioned Fox ranks recently:

https://lifein19x19.com/viewtopic.php?p=261939#p261939

Yeah, I've only played five games since I started playing on Fox again but I do feel like something weird is going on with the ranks. I'm 2-0 at 2d now and both opponents misread simple ladders. I've been playing all my games with the new 5:00 + 3x30 automatch button; maybe the pool of players who play that mode is different, or people take it less seriously?

I am also using the Automatch button, but have changed the settings (by using the button "choice") to 20 minutes + 3x60 seconds because I don't like to be under time pressure, even if both players are far from using all their time.

I haven't played yet at 2d level, but at 1k level my opponents felt several stones weaker than before the appearance of the new button.

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.

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.

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.

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.