Challenge : 100 reviews


#1

Hello everybody,

I am a beginner (+/-15kyu ogs) and I do not have much time to learn and play go. I think, the best way to improve my level is play and review my games.

To do this, I challenge myself : Do a review of my next 100 serious games. I would like a review by a human when I lose my game and review myself my winning game with GoReviewPartner/LeelaZero.

What do you think about my challenge ? Is it a good way to study go ?
Is there someone to help me ?

Thanks in advance for your answers.


#2

Those who talk do not do.

Those who do,…


#3

Oops! A goal revealed:

:laughing::laughing::laughing::laughing:


#4

I know it’s in jest, but I’ll add that there’s a difference between wishes - something you don’t have control over (despite the wording suggesting intent - “want”) and action plans whose outcome you actually have plausible control over (“I want to bake a cake”).


#5

Yes, it is. Go ahead! :slight_smile:
Reviews done by stronger players are enlightening.
Also, I did some review with Leela (not Zero), and it was interesting too. Especially when you look into quick reversal of win rate: they’ll show you major mistakes.


#6

I think playing 100 serious games is one of the best ways ro learn anyways, certainly in the ddk range. Getting your games reviewed can only help.


#7

I would mitigate, not so good for all DDK but perfectly fitted at the OP level (15k).

I prefer free exploration and not that much burden of studies for a full beginner

@BenVsb i encourage you to go play in some real life tournaments, it’s one of the best way to get serious games


#8

I’d recommend the following:

  • Start with a self review without leela/goreviewpartner
  • Once you’re done with the self-review, do a leela review. Focus on moves that leela likes, and that seem sensible to you.
  • Get a good ‘human’ review, as well. At DDK, having someone point out “bad shape, bad direction of play” can add a lot that you won’t get from just “-15%”.

If you can do that for your next hundred games, that’s a GREAT goal. :slight_smile:


#9

Thanks to everyone for your answers.
I’ll do my best to become sdk in less than a year.


#10

Getting dozens of reviews from humans is quite ambitious. My recommendation: come up with specific questions for each game. This will help you, and show others you’re serious about it.


#11

Interesting. How does it works reviews with Leela? What can it say about a game?


#12

Basically you get how your win probability changes for each move.

Untitled

For your big mistakes you can try out different moves and see how the win probability changes.

Leela also suggests move she would play, but they tend to be not that usefull if you aren’t dan.

The 2 videos below are about what you cannot do with leela and what it’s good for.


#13

I think it is worthwhile to review both wins and losses with a similar approach. There are often lessons to learn and mistakes to realize whether we win or lose.


#14

Agree :100:% although it’s more easy to forgive yourself when you win ofc.


#15

A note on bot reviews: I use leela, not leela zero. It’s substantially weaker, but still amateur dan-level strength, which is generally more than enough. It tends to give more “human” looking moves than LZ, though it sometimes fails to read complicated fights. When I’m reviewing a game, I mostly look for things it identifies as outright blunders. Most of my moves are in the -2% to -5% range. Anything worth than that is something I should learn from.