Any tips on doing self-reviews of games? I often look over my games, especially when I have lost, to see what I might do better. Sometimes I can identify a glaring mistake, but often I just wonder if I could have made a better move. The (free) AI will generally show me one or two spots where a different move should have been made. But I am looking for guidelines–check the cutting points, look for the biggest move. These I know. Any others?
Top two tips I have are
- Try to play more humans and less bots, you will learn better this way
- Try to have a plan for your moves, this way if the play doesn’t work out you can better identify why it failed, or if your plan succeeds and you lose anyway, you can learn to aim higher.
My advice is to get someone stronger to help you review first if you are just starting. Self-review is only effective when you reach a certain level, and if you do it wrongly you may end up in a vicious cycle and not be able to improve the game.
I think the AI usually points out the moves where the lead changes, which is not always helpful because it might be just a small 2-point mistake that swings the lead.
You should look at the moves where the AI shows a huge drop in points.
For example, in this game https://online-go.com/game/70154544, the AI shows that at moves 102 and 108, you went from losing by 3 points to losing by nearly 30 points. See if you can figure out why.
When there is such a big drop in points, usually it’s due to the life and death of a group, which I believe is the main thing you should focus on at 20k.
Don’t worry if you can’t figure out what the problem is. Just thinking about it is already helping you improve. And of course you can always ask someone stronger.
About self study, the essential point is that a stronger player will be able to reverse your judgement, when you think something is good but was wrong.
There are cases where you know you need to find a better way but there are cases too where you thought to be right and good but that was absolutely not the case.
AI is stronger as you but often too strong so you go in other difficulties where what’s happening is too subtle to get at any level besides the players of high levels and there are no warning about it.
The human player can use concepts, simplify a position, select what he thinks more crucial to look at, give you related “exercises” and so on. He can even try to point out your weaknesses when playing a game with you, just by generating similar situations, the bot won’t select his moves in an attempt to give you a better understanding.
So when you have no choice and use AI then you have to select on your own which position is relevant to your knowledge and stay relax when it’s not understandable so as to not waste your time and effort too much.
- Then try to find a better move and check if you’re right. On OGS this can be done by using the estimate score button in analysis mode (or by being a site supporter of course).
That’s one thing I do, but it’s not perfect. There’s always the danger of asking the AI for its opinion but learning nothing of it. Knowing which moves would have been better doesn’t help, if you don’t know why they are better.
When I want to review one of my games seriously, I first go through it backwards without AI trying to find where I went wrong, and what I could have done better. Then I’ll turn on the AI, and see if I was correct about where my mistakes were, if there were some ones I missed but realistically could have found, or if there were some moves I thought were mistakes but were actually fine
So rather than using the AI from the getgo, I use it as a source of immediate feedback to my own review, and secondarily see if I can learn anything else from it