Playing consistently, but getting worse (instead of plateauing)

I’ve tried to play regularly and consistently (across all 3 board sizes) ever since joining OGS 5 months ago. Up until about a month ago, I was seeing the typical (and dare I say “expected”) pattern of “improve a bit, plateau and struggle at “x” rank, break through plateau after a bunch more practice, repeat” but for the past few weeks, I seem to be getting worse, and it never feels like I’m at my best when I’m playing anymore. I find myself making mistakes that I got out of the habit of making (or thought I got out of the habit of making, since clearly I haven’t :pensive:) and instead of staying at the same spot and plateauing (which, as I mentioned above, is the expected case), I seem to be only decreasing and getting worse and I can’t figure out why. The only thing that comes to mind is that this downward trend first started around the same time as my final exams, but if stress from finals was what was causing me to not play at my best, then it shouldn’t have lasted after I finished my finals. (My only other guess is that trying to play equally on all 3 board sizes is having a “jack of all trades, master of none” sort of effect on my playing, but that doesn’t sound right either.)

1 Like

Like, Go, learning needs a strategy.

I had a look at your record, and it seems like your game choices are “fast and all over the place”. You are playing all sizes of board, and various time settings, including up to blitz-like live, and vs bot.

I think this is very sub-optimal way to improve: a plateau is not surprising.

The first question I would ask is “what kind of Go game do you want to get good at?” There are two dimensions with at least three points on them, and you need to chose one: size and speed of play.

For improving, even if your ultimate goal is to be a blitz player, I think you’ll improve better with games where you think about each move. Blitz speed from instinct play comes from building good instinct by thinking.

Then, get a review of each game that you lost. Make sure that there is one concrete “lightglobe” kind of “I can do better at that” that you take away from each review.

The last general thing I can think of is that if your goal is to be good at 19x19, then get a few lessons from someone who undertakes to teach you “the basics”. In my mind, this means “the kind of thing that Dwyrin teaches in his basics series” - at DDK, for sure there are a set of “explicit things you can be told about how to think” that will improve your play. Watch some videos from that series, and or find an alternative that works for you.

Here in the forum there are no shortage of teachers, so reach out and find someone that suits you.

Good luck.

PS: Oh - don’t try to learn from playing against bots. It’s too hard to understand what good bots are doing to learn from them, and low level bots can only be worse: they make hard to understand _worse_moves. There are only two good reasons I can think of to play a DDK bot: to get a game quickly just for fun (no learning intent) or to try a specific strategy against a bad player. The important thing is to realise that you aren’t seeing good responses, so you aren’t learning from a good player when you play against a DDK bot. Losing your first 100 games against one of those would be an awful way to go: you’d be learning how to play badly while losing.


Just looking at your rating graph I can’t see any downward trend. Just a bit of fluctuation, which is just normal and you’ve had on other levels before as well. Maybe you are just becoming more aware of mistakes you are making? If so, that would be good rather than bad. :slight_smile:


I’ve come to expect, in myself and others, a deterioration in learning, right after one grasps something important. It’s different for anyone, and different for different learning subjects, but I sense it is just our brain reassessing information under a different light, a calibration of sorts.

If I may, I have (very recently) discovered another aspect of playing with bots: I’m very new and very bad at Go, but my understanding of the game is that “it’s all there” for everyone to see. So, I frequently find myself “hoping” that my opponent will “miss it” when I’m trying to play creatively (not that I’m usually capable of such moves, just saying). But playing against a bot, which by nature will calculate all possible outcomes (according to its programmed level, of course), takes misdirection out of the equation. I just play the move I think best, and my opponent’s reaction is irrelevant to my ability (although I don’t really plan to reach that level, like, ever :slight_smile:) .

If you want to know how to get better, play higher ranked people and review their games / get reviewed. Take that knowledge and translate it into skill by playing opponents similar to your rank.

Also try to stick to 19x19 live or blitz games.


Stop playing amybot. Make a new account where you only play live 19x19 seriously and with the time setting of your choice… Don’t resign games where you are not sure you have lost. That’ll make you find your actual current ogs rank which’ll help your understanding of your own strength, to some extent. The way you’re going at it (from a look through your game history), it might be that you are improving in strength but can’t project it onto your rank.

If that’s not the case and you’re actually stuck, then look for a teacher/mentor to give you weekly advice on what to do to improve. The advice they’ll give -more or less- will be doing tsumego(hero/pro), using your time/trying to read as much as you can in your games, browsing through pro games and watching nick sibicky/dwyrin/yoon’sbadukcafe videos.


just play those games like you mean it, think and think hard before you play a move

another thing is doing tsumegos/tesugi problems, it’ll help you improve your overall reading

as for opening and such, you can get games reviewed on the forums, or get reviewed by an AI whichever works best for you

i think most of us is stuck with reading + shapes instead of opening, because western go communities emphasize a lot on opening (ie. settling groups then attack etc.)

1 Like