'Why are we still SDK?" -- some thoughts

For example for understanding the big picture of the whole board, you often need to judge how strong a group is. If it gets surrounded, does it have / can it make two eyes? Those are the questions to solve with tactic, and the task of the strategy is to use this information and shape it into a whole-board view.

Well said. And I agree with all your points.

I want to add that the reason I emphasize on tactics is I personally feel tactics are easier to learn. The method for solving tactical problems is there and there are plenty of metarials one can grind.

3 Likes

Wise words! At the end of the day, it’s just a game. It’s supposed to be a fun activity. I enjoy all my games. Even if I lose, so long as it was a fun game with some interesting fights, then I’m happy. Sure, it’s nice to improve, but that improvement just comes naturally and gradually over time. More study etc can increase the rate of improvement or help break through a plateau. But if, like me, you are limited on time, then don’t worry about improving. Just enjoy the game (or enjoy tsumego, or enjoy youtube, etc.)

For me with limited time, I find the most efficient activities to continue improving at a reasonable rate (one rank every few months, currently 6-7 kyu) are:

  • Play games. Focus on variety. Especially a variety of sizes (9x9, 13x13, 19x19), a variety of time settings (blitz, fast live, slow live, correspondence), games against a variety of opponent strengths (much stronger, a little stronger, about equal, a little weaker, much weaker), and a variety of game settings (handicap, no handicap, komi, no komi, etc.). I find the variety forces you to adapt your strategy / tactics to different situations and allows you to generalise your play more easily, e.g. playing high handicap games as white is good training for when you have to invade in an even game, playing against weaker opponents allows you to solidify fundamental ideas in your head without the distraction of difficult fights, etc.

  • Review all games unless they were extremely uneventful, e.g. handicap game with you as black where you just sailed straight to victory (maybe handicap was too high). Preferably review with opponent, but at least self-review. AI review might also help. Focus the review to a small number of things (1-3) to work on next game so as not to get overwhelmed or spend too long on the review.

  • Tsumego. Some strong players say you don’t need it. Maybe that is true for some people. However, I can notice a definite increase in my rate of improvement when I spend 5-10 mins per day doing tsumego compared to periods when I do not do tsumego. Plus, I enjoy tsumego so I will quite happily do that instead of playing a game if I am short on time.

  • Read (high quality) Go books. If it’s a good book, ideas will be explained in a way that you can understand and thus internalise more easily. Plus, this is a source of ideas beyond what you can get just from playing and reviewing your own games.

Things that I find do not help:

  • YouTube - I watch a lot of Go on YouTube but I don’t feel like I get much improvement from it. I find it’s more for entertainment.

  • Too much correspondence - takes already limited time away from other Go activities. Some correspondence is OK but not too much. Also, the feedback loop on correspondence games is too slow so not ideal for improving. as you forget your thinking by the time you get to review.

  • Reviewing pro games - if you don’t understand direction of play, then this can be a useful activity to get the idea. But beyond that, the level of play is too high to get much from it. There are other more efficient activities. I still do some of this because I find it interesting but I don’t feel like it helps me to improve much.

This one? Always read three moves ahead

Hehehe! :smiling_imp: Now I know your weakness. Watch out for my late comeback in our ongoing correspondence game!

5 Likes

Thanks for asking. I wasn’t aware the title doesn’t have any word like reading, calculation or tsumego.

3 Likes

Please elaborate on how you reached this conclusion with AI review. When I review with AI, I often find mistakes of strategic nature, and AI evaluation confirms it. For example in my game against drifterwolf yesterday I played a shoulder-hit intended as a reduction, but given nearby strength it was a poor choice. Sure enough AI indicated the mistake with a 4.4 points loss. I would call it a strategic mistake, not a tactical one, but the distinction is of course not clear-cut.

It is not necessary to find “the one and only correct strategy”, but rather to avoid strategic decisions that lead to a points loss in the long term. Principles such as “play away from thickness” are helpful, and it is important to understand when to apply them.

Putting the objective evaluation aside, good strategy can also include factors outside the board position, such as “Are you tired? → go for a compromise instead of risking a misread in this complicated fight.” or “Do you feel comfortable in games with competing territorial frameworks? No → Establish a foothold at various places in the opening. Yes → Keep developing a big framework around your two corners in the opening.” … etc.

5 Likes

Probably many studying methods work, as long as you enjoy them enough so that you are willing to put mental effort.

Retrospectively, I feel I did spend a lot of time playing serious tournament or online games, solving tsumegos, etc. but lost much more time reading superficially, or falling asleep in front of go videos or books.

2 Likes

I didn’t mean to say that strategical mistakes don’t exist.

I meant that in many cases (particularly in the early game when I feel that strategy has more relevance than in the endgame) you can choose between almost opposite strategies, for example territory vs outside influence, or aggression vs prophylaxis. And while the AI may prefer one strategy over the other in a specific position, the difference in evaluation can be tiny with (nearly) perfect execution (less than 0.5 points difference) from the AI’s point of view, which is mostly insignificant compared to most tactical inaccuracies.

Indeed, so it can be hard to claim general objective truths on strategic decisions. Strategy has much more to do with personal psychology/emotion/style/preference than tactics.

There are a few things I think I especially need to work on.

I feel my reading is too weak, especially in life and death. So I think I will benefit from tsumego. I found Cho Chikun’s Elementary Encyclopaedia of Life and Death got too easy, but that the Intermediate was uncomfortably difficult. SL’s “kyu exercises” are pretty limited. I could grind through Attorante’s problem thread but the ratings of those puzzles are of the vague “Easy–Medium–Hard” kind. I used to use TsumegoHero, but I eventually reached a point at which I’d already done all the tsumego I could do and the more advanced sets seemed too hard. I suppose I should go to 101Weiqi but the Chinese interface seems like a bother.

My joseki fundamentals are perhaps lacking in the realm of post-AI Go. I am still being surprised by new moves that players have researched at home with bots, that secure them some 90pc ok result but will not be on Waltheri or OJE. I’ve never studied with a bot and it doesn’t appeal to me, but practically every 5k-up player seems to do so, so maybe I’m putting myself at a disadvantage.

I don’t count the board and I tend to play eyeballing territory and going by a feel of who is ahead. Some people advise counting in pairs, but I can’t see why it’s any easier to count pairs of points than the points themselves. I would like to develop my counting skill, but in a real game there never seems to be the time. I’d really like to pause for ten minutes, heh.

I may not own the right books. I have a variety of books, including the classics Attack and Defense and The Breakthrough to Shodan, but I don’t think they’ve helped me. I’ve never read Lessons in the Fundamental Principles of Go which is so popular.

I’ve never really bought teaching, only a couple of lessons from an amateur back in the day. I probably should, as Nibby advised, but I don’t have a ton of cash for it. Especially I can’t afford to join an 80 euro / month payleague like some other improving SDKs. gennan and a few other dans and other SDKs do give me pointers on GoKibitz, which I’m grateful for.

I found the comment to “change your technology” interesting. I used to play quite territorially as Black and I’d often cede too much centre and lose that way, so around the start of this year I began playing much more influentially, whilst staying territorial with White.

I can’t say I don’t review enough, but I wonder whether I’m reviewing in the most efficient way. It’s easy to review for a long time but not so easy to make that time worthwhile. When reviewing with a stronger player, the stronger can ask the weaker questions about the position and ask them to propose candidates and ideas; but the weaker player doesn’t know what questions create the most revealing dialogue.

At Durham I definitely suffered from not having a good idea of when I was ahead. It’s hard for me to blunder twenty points and say calmly “I still have five points; the position isn’t bad, I will play on.” Especially if there isn’t enough time to even try to count.

So, tl;dr

  • where are the tsumego?
  • how do I improve my counting and positional judgement of who is ahead?
  • is it necessary to study with a bot?

I also remember now that I discussed whether commercial teaching is necessary to reach shodan in my thread Discussion of the relationship of student rank to commercial teaching (or a similar title).

As a longtime OSR and GoKibitz member, I can’t say that these free communities have brought me far closer to shodan, despite much work by GoDave on the former and Gennan on the latter. It would be a shame if there is an “improvement paywall” through which sufficient funds are needed to advance, but I can understand that that may be the reality of growing at this rank. Strong players are within their rights to charge for their time, although I’m thankful to the ones like Martin, gennan, GoDave, Mark etc. who don’t.

1 Like

Whether strategy or tactic is more important, I have no opinion on it. In the first place the distinction between them is not well-defined and so it depends on the objective view of everybody individually. But whenever somebody claims one is more important then the other, I get skeptical.

It is easier to conclude a review by “I lost this game due to these reading errors.” than by “I lost this game due to these strategical mistakes.”, and so some people (imo prematurely) come to believe that tactic is more important.

I’m hesitant to get into this topic to be honest, so I’ll just try be helpful.

I followed this by @aesalon and it was a good guide. I don’t use 101weiqi that often. I think it still has some potential usual issues with user submitted content like goproblems etc. In any case if you like timed puzzles or sets graded by rank, (which of course probably doesn’t correlate to any particular server or association) it’s fine. The comment about google translate well is true, with chrome translating the page works mostly fine, and you can get used to just what button does what without translating after a while.

Another option is apps:

  • BadukPop has similarly timed puzzles, or puzzles adapted to you level (rating in app) etc.
  • Tsumego pro has daily puzzles and sets you can pay for.

There’s always tsumego books too. I recently got Cho Chikun’s “all about life and death” and I have a bunch of others. Sometimes the books focus on more realistic (compared to actual games) puzzles, whereas sometimes puzzles on Blacktoplay or other sites/apps focus on making an unintuitive puzzle to make it hard (either a very arbitrary looking arrangement of stones, or forcefully making the 1-1 point the only right answer, or a sneaky throw-in etc – something to try catch people out.)

I think there’s an improvement paywall in most things if one wants to get strong in a reasonable time frame, if one isn’t already efficient with their learning. Even if it’s not like a pay-to-win video game, top professionals in a lot of sports will try to get the best coaches, equipment etc to help them gain an edge. Similarly players that want to become professional will invest lots of time and likely money (their own, their parents etc unless there’s some sponsorship/scholarship) to get to the goal.

In theory there’s no reason anyone can’t get to near professional strength by just playing against the strong bots like Katago or LeelaZero hosted on OGS for example and that wouldn’t cost anything extra than what you’re already paying to just be online (internet, electricity etc), on the forums etc. Similarly with all of the freely available apps, servers to play on, Go lectures, videos etc.

It’s just I think playing against such a strong player (bot/AI) is probably quite draining, and demoralising. It’d be hard to notice an improvement without mixing in games vs people, and people will play suboptimal moves and variations that Katago won’t, which one has to extrapolate the “lessons” from Katago themselves to deal with. I guess similarly one has to somehow know how to efficiently learn from others study material (lectures, reviews etc).

[As an aside, I feel this way with Tumbleweed at the moment. There’s not a lot of resources to learn from. There’s no theory books, or definite methods on how to improve. One has to either learn from stronger players, their games etc, or by playing oneself or against the bot and reviewing. It’s just a slow chug with not much indications of progress.]

2 Likes

I have never been able to spend enough time on tsumego practice. I just don’t enjoy it at all.
I find tesuji practice a bit less boring, but even that I rarely do. So I can’t give you any advice on this.

If you cannot find those hypermodern variations, maybe you can avoid them by playing a slightly more classical opening? The Chinese opening for example should still be fine up to pro level, even though it’s not very fashionable.
But if you really want to dive deeply into all this hypermodern stuff, I think you need to review your games with AI somehow.

Counting is important. Knowing where you are is important in deciding where to go next. The only way to get better and faster at counting is doing it more often.
If you lack the time for that in the later stages of your live games, you should manage your time better.
I used to spend too much time in the opening, ending up in time trouble in later stages of the game. Try to avoid that situation. Be prepared to not go into that interesting and sharp, but very complicated fight, if that saves you valuable time to count and play good endgame later in the game. I have come to believe that a strong endgame is more important than a strong opening (even though it may not be as interesting for most players).
A way to manage your time is to ensure using less than 25-30% of your time in the opening (choosing a familiar opening can save you time, for example a Chinese opening). Also, check your opponent’s time regularly in the midgame and endgame and try to avoid falling behind in time there.

I don’t think regular payed lessons are really needed where you are.
My greatest improvement occured in a period where I had a couple of sparring partners of similar level IRL. We sort of “grew up” together and went from strong kyu to mid dan together, stimulating each other through friendly rivalry, springing new joseki variants and weird openings that we picked up somewhere, going to tournaments together, studying pro games together etcetera.
We had some group lessons from a pro, but I think this was far less important than being immersed in this overall group effort.
Also, studying and training together is much more fun than doing it alone. Only if that is out of reach, and if your time is more valuable to you than your money, I would consider getting payed lessons.

3 Likes

You know, even if i have said multiple times that i want to reach 1dan someday, the truth is that i havent put much/any effort on studying for years now. I think that my go skills did pretty much plateaued even before alphago and other modern bots came along, like maybe i’ve learned some new josekis and such, but my rank has been stagnated for 5-6 years now. And since i’m not in my teens or twenties anymore, i dont feel like starting to study again >___>

Now i play just because i love the game and its fun activity for my spare time (and because i like chatting with go players but it would be weird to hang around ogs on daily basis without playing go xD)
And i think that’s just fine, i am hedonistic person so i rather spend my time on doing things i enjoy, instead struggling for something which might or might not benefit me in the long run.

10 Likes

My rivals have a habit of surpassing me eventually~

1 Like

Obviously you don’t know: 100 post likes =.1 rank point.

4 Likes

A wise 5d once told me that you cannot win in the opening so no need to spend too long on it. Essentially, his tip was to play the opening fast just making sure to avoid blundering but not trying to eke out each small advantage and spend your time in the middle and endgame.

This tallies with that suggestion to spend absolutely no more than 25% of time on the opening.

My other thought @bugcat is that you should maybe try thinking less while playing. How are you with blitz? Sounds like you like to overthink things. Similarly with reviews, I’ve enjoyed your detailed reviews but maybe more beneficial to focus on the one or two key issues in a given game rather than how each and every move could have been improved or not.

Anyway not the I have any useful advice of course being a fellow SDK water treader!

4 Likes

Obviously not, before the existence of bots people were able to reach dan level without them. If I remember correctly, Ryan Li doesn’t study with bots but studies games by pros who have studied with bots.

I personally don’t like researching joseki variations with bots either, but think that bots can be used to show you better moves, so can be of great help. OTOH, studying pro games could also do the job.

2 Likes

Unfortunately I can’t give detailed advice on studying go, because I never really spent much thought about how to study go most efficiently - I just did what I enjoyed, which often happened to be taking in various kinds of go content, like the discussions on this lovely forum :slightly_smiling_face:

But there is one thing I feel confident in saying. No matter the level, I recommend playing games with players of similar rating (optimally opponent is a little bit higher rated), review / talk to players about their opinion on the game, experience and experiment with all kinds of strategies and not try to fall into the habit of playing the same moves over and over again. Try not to be narrow-minded / dogmatic about proverbs, that blocks your development.

6 Likes

If my memory is correct, we played a tournament game a few months ago and you said you like to drink wine while playing? That’s indeed quite hedonistic!

1 Like

I think you are too harsh on yourself and maybe over think a little. Ofc, we have our weakness, but at the same time, we are not aiming to be pros.

  1. Joseki: I have been watching a Joseki series of a 7p. He always emphasizes the importance of learning the thought process instead of memorization of the joseki, and also the choice of the joseki based on the board situation. For example, do I want the corner, left, right or outside etc. Our regular opponents are SDKs and I don’t think they know that much of a difference between an AI and classic joseki. I used to be so scared, when my opponent started the corner with 3-5 etc, but I had to play. Guess what I found out, they don’t necessarily know how to play that either. After all a 1k is still a 1k, that he/she starts the game with a joseki unknown to me, does not necessarily make him/her a 4d. Just pick a simple variance to control the “loss”, it may not end up with a loss anyway.

  2. Counting: I thought counting is easy, counting fast is not. I am curious if your definition of counting includes valuing unsettled area. I don’t think we need that level of understanding to achieve our current goals. I only count settled areas. When the board has too many unsettled areas, I just don’t bother to count.

I typically count on my opponent’s clock.

I also think we need to count or evaluate the status when it affects how we play. Many times it does not. A recent game I observed is all stones are live, one side is behind in territory, but with a big moyo. So the tactic is simply how much it can enclose in the center. Now instead of reading how to maxmize the center territory safely, the player used 5 minutes to count and concluded it was doable to enclose enough points to win the game. In other words, the player spent 5 minutes to consider resignation.

I do read the board often to see if I am ahead or behind and find out more than half of the time, I get it wrong. It looks very bad, but not really. Keep in mind, my opponents are at the same rank. When I get it wrong, they don’t necessarily get it correct. So don’t over read, knowing that I don’t get it right most of the time, I will just play the game and don’t let the negative feeling affect my performance. We have to fake confidence during the game.

4 Likes

Taking 5 minutes for a single count is too slow for a live game IMO. Better to use about 1 minute, 2 minutes max (even if it’s not as accurate that way) and do it more often during the game.

And yes, don’t count unsettled areas. Only count secure points and use that to inform you how many points you or your opponent needs to gain in those unsettled areas.

3 Likes

A simple but brilliant idea! And one that I had not thought of. Counting takes time, even with techniques to make it faster. During opponent’s turn is an efficient time to count as it’s less efficient to read then as we don’t know their next move yet. I will definitely try this out next game :slight_smile:

And another! I waste a lot of time estimating the score of unsettled areas and this leads to inaccuracy. For example, I recently had a game where I counted I was ahead by about 8 points in the late middle game including some unsettled areas, so I played safe moves. Turns out my counting was off - AI said it was much closer - and my safe moves were too safe. I lost by 0.5 :cry:

3 Likes