Hi all,
i am more a beginner than experienced player and mostly play against computer (doge-bot etc.). I moved from 12…11 up to 8 and back to 10 again and again - not really improving much. How can i improve faster? Are there special strategies, books, live discussions on games - what would you suggest?
Thanks a lot!
Rilo
Best thing is to play lots of different humans. Playing just bots leaves big weaknesses in your game.
I have never played against bots but I find that I have improved from working with tsumegos (e.g., the daily homework at 101weiqi.com and various books) and also from reading Janice Kim’s series Learn to Play Go. When it comes to playing, I participate in ladders, a couple of tournaments each month, and I have a couple of contacts on OGS who are a couple of kyus stronger than me with whom I play regularly.
Me too! My rank nearby 9k for a while. I play with GNU Go from zero after Alpha Go - Lee Sedol Event in 2016.
Now the suggestions I could offer are:
The idea is simple:
You can create a game and set ranks for it (no stronger than 9 kyu stronger than yourself if I remember correctly) so you simply set the rank so that only players who are stronger than you can accept it
.
It’s common to have a step in your progress at this level, so firstly you are not alone!
The way to pass it may vary with each player, There is no universal answer.
The fact that you mostly play with AI could obviously be a factor. Playing with humans may enlighten some neglected aspects of the game. Besides humans are very useful to review your failures.
Reading books or watching videos give opportunity to check your ideas. That’s important if your fundamentals are weak (like to play too agressive or to not make well the difference between a moyo and a territory or how to use influence etc… )
Last practicing reading by doing problems regularly and seriously will enhance your reading power and as a side effect, help you on the side of shapes.
Fresh idea to me! I will try it. But a bot model is better because it is fast and doesn’t need time to think like human. So I believe that:
One will save half of the hours to train his/her Go skills with models than with humen to reach the same rank.
But humans can conceptualize and explain which will save you much more time.
Playing just bots leaves big weaknesses…
Why? Shall you explain it in details?
For example a stronger player will tell you that you played too near the strength. And show you why with another more caricatural example. AI will not, you ll have to guess something similar on your own.
A human will have the ability to detect something wrong that you commit again and again. He will detect some distortion in the way you play. Then he will try to make it even more clear, so that you stop doing this same failure
Hi,
Try playing with really good players (1dan and above) or doing some of the OGS puzzles. That’s what I’d do.
I fully agree, i would still consider playing less strong players as they still have a lot to teach and may understand better at times the difficulties at your level. Besides to reach to win them some day is a motivating and interesting challenge !
“How to improve” is just one option. Another which I’m currently cultivating is an acceptance of my borderline ddk/sdk existence. I’m people. The unfiltered reality is that I lack the impulse control necessary for playing better.
Although I am not very strong so my opinion doesn’t matter that much, but I agree. The same goes for the AI review. The AI review can be good to note somethings you missed but most of the information that I get from the AI review, I really don’t find that meaningful although I and one of the persons with whom I often play usually discuss the AI review of our game which makes it a bit more useful.
My biggest improvement comes from reading books and watching training videos(specifically designed to train amateurs at my level, not the pro game commentary ones). Ofc, I have to play the games and practice the learnings.
Bots are not good at simulating human mistakes, so to get a bot at 10k level they kinda throw together some dan level play and some 20k mistakes to balance it out. Once you discover which 20k mistakes a certain bot makes you can exploit it reliably. It’s not the same as playing a human and usually teaches you the wrong lessons, that being how to exploit a particular bot’s weaknesses rather than fixing your own.
Iff you are stuck at a level, you could start to change things in your go style. This could be anything:
Don’t change too much variables in your go style at once, because this will only confuse you. Warning: this new approach might lead to a drop in rank first.
stop playing against computer
First i have to say i am overwhelmed to get soo much feedback that quick - thanks a lot!
I have to admit i like to play rather fast and mostly 9x9 games against bots - i just play for the fun of it not really too seriously. I recognized if i play a bit more slowly i also play much better in most cases. I also tried different bots but i tend to use the ones that play just a bit above my level so in most cases i loose - but i am happy to win against a stronger bot.
Maybe i should take the time to play more 13x13 or the real thing 19x19 games. 30 years ago i started in a Go-club in Kiel, north Germany, we had one 2 dan player there and he could teach you a lot for every move. Thats probably the best way to learn from a pro if he takes the time with you. When i play against human here there is hardly a comment at all, best is ‘hello’ 
cheers, Rilo
you can play vs human and then review with strong bot if you lose
you ever win against someone because they do mistakes
bots do completely different mistakes than humans
learning how to exploit bot mistakes may be completely unhelpful in learning how to exploit human mistakes