This topic is about how to conduct a review of a Go game. Not practically, like how do I open the OGS review tools - that is already covered here:
No, no! I mean how does one make a game review an effective improvement tool.
I will share my limited knowledge and experience. But I would very much like to hear what the rest of you have to say about this topic as I’m not sure that my review methods are necessarily that great.
This topic covers three types of reviews:
Reviewing your own games
Reviewing other people’s games
Reviewing pro games
In each case, aspects to consider might include (but should not be limited to) the following. But really, any tips or guidelines about how to review could be useful:
What is the aim of the review?
How does one perform the review? Alone or with opponent? Move by move or just focussing on key moves? What are you looking for as you go through the game?
How should the review be recorded, if at all? OGS review tools, paper and pencil, in-person discussion?
Who should conduct the review? In a post-game review, should the winner lead the review? What is a good rank difference for reviewing other people’s games? At what level is reviewing pro games relevant and how does the approach to pro reviews change with level?
How long should one spend reviewing a game?
What about AI review?
I will put my initial thoughts in a separate post. Thank you for reading this far and I look forward to an interesting discussion.
Here’s my thoughts on reviewing so far. I’m not too sure about some of these ideas so please share yours too.
Reviewing your own - to see what you / your opponent did good / bad so you can do better next time.
Reviewing others - give them advice as above
Reviewing pros - get ideas for direction of play in the opening
I’m really not too sure here. I’ve tried discussions with the opponent by OGS chat or in person - fun and sociable but I’m not sure that it does much for improving. I’ve tried looking over the AI review myself - useful for direction in the opening but some of the recommendations are too complicated. And I’ve tried OGS review tools but going through the game move by move and adding comments and variations takes too long. Anyone got any ideas here?
The only other useful thing I can think of to share right now is something that I learned from some basic training I did in how to teach - the most effective feedback is depersonalised feedback on what went wrong and how to put it right. Praise doesn’t make much difference. And by depersonalised, I mean “the cut here is unplayable because blah blah blah” rather than “you should not have cut here because blah blah blah”. The former allows the feedback recipient to be more objective about the criticism, and so it is more effective. This isn’t just hearsay - there’s research to back it up (I can’t remember the references - look them up if you’re interested).
I recently reviewed a professional game from a “group thinking” standpoint – when & why were the different groups formed, when & why did they connect, and when & how did they influence each other?
Recently I’ve been interested in this kind of review, the “group thinking”, “haengma”, or “narrative” style – in this way of reviewing, one focuses mainly on the directions that the groups are travelling in, whether they are free or enclosed, whether they appear weak or strong, etc.
Even @Trevoke at 8k (I don’t mean to use an insulting “even” ) produces reviews of 17th- and 18th-century professional games that are, imo, very good.
I’ve made some reviews I’m happy with: the Takagawa one, Lessons from 1914, and the Classical-era reviews I did for The Tearoom group.
A lot of amateurs, even SDKs, review Kitani games as well. The streamer Jr4ya springs to mind… he might be a dan actually, not sure.
In general, my feeling is that SDKs usually like to review games played before about 2000 (Cho Chikun, Kobayashi Koichi, Takemiya Masaki etc., segueing into Lee Changho) and especially games played before about 1970 (Kitani, Go Seigen, Takagawa, the elder and younger Fujisawa etc.) and from there back into Classical Japanese and Chinese games.
On a second thought though, when one watches pro games with pro commentation, it is a hobby or entertainment, like watch UCL or EPL. When one studies the pro games alone, it becomes a chore or work, not fun.
At 6k, the first game I memorized was one of Shusaku (vs Shuwa). I had the goal to review it on my own which gave me a reason to do it. I enjoyed the deep commentaries from the book (Invincible) but when I started to play with variation on my own, it became quickly a endless headache (vertigo?). I enjoyed mostly a part of the yose, more reachable but still astonishing.
I am a lowly DDK, still chasing my first 100 losses. Still, I’d like to mention a tip I got from In Sente. In an early video, she suggests reviewing every game you play, and taking one strategic lesson and one tactical lesson from every review. For me, this has been extremely useful. Sometimes I find I can’t properly find two lessons to put one in each category. One lesson is fine. But for me, so far, there is always a lesson.
I keep a note on my phone, with a few lines about each review, and the lesson in bold type. Reviewing this, I see which lessons I’ve internalized, and which mistakes I’m still making. Also, some of my early notes were wrong. And I think this is the most useful part - continue to refine your review process. Here is the meta-review! Continue to question how and why you’re reviewing your games. Are you getting something useful from your reviews? If not, you should probably change what you’re doing.
I was reminded of a self-review strategy first used, or at least popularised, by @GoDave89.
He would take screenshots of his mistakes and store them in a file database, with the nature of the mistake expressed in the filename. This idea caught on in OSR, where it became known as “davebasing” (Dave databasing) and was given its own channel in the discord.
I tried the process several times but I encountered a seemingly insurmountable issue, in that most of the “mistakes” I entered to my davebase were found, on inspection by stronger players, to have not been mistakes at all. Figuring that it was a waste of time feeding myself false information, I gave up.
I mean - many have used it successfully, so I’d like to think I have cleared some paths for others.
I’m not that special… It’s not like this idea is new or anything and I am the only one who can ‘handle it’ :'D
Have you actually tried it yourself so you can at least speak from your own experience?
What are my ‘ways’ anyway? Doing lots of tsumego? Playing lots of fast games? Reviewing my games with AI? Very uncommon… :'D
(this is GoDave btw, in case that hasn’t become clear ^^)
This is essentially what I do with my database idea - I have a few specific focus topics and try to find mistakes in those categories to learn from to get a bigger picture - trying to take a few key elements away from each game. Ofc the stronger you get, the more specific those takeaways get, but I still have a lot of basic things mixed in there as well like “don’t play too close too thickness” - just in a bit more advanced way of thinking (I’d like to think lol)
Here is a recent example - I was more concerned with that single black stone when in fact my own invasion stone (C12) was just as weak (especially with black so strong on the other side) - so essentially I started a fight that was equal-ish but black had the first move to get an advantage.
This is the kind of conceptional things I try to take away from my own games (with the help of AI often enough)
This is what an excerpt of my database looks like btw., in case anyone is interested ^^
It’s an idea I’ve been using for 3 years now and it helped me and a bunch of other people improve - but as you can see it’s also not for everyone (just like with everything that is related to learning something - everyone learns a (slightly) different way )
I guess for me, I’m not sure I would need to make a database separate from OGS.
While the screenshots pinpoint a particular moment in a game so too would having a review with a comment.
Similarly the screenshots allow you categorise them with names relatively easily, but maybe one could just sgfs in folders in the sgf library with similar folder names, and the review will still be linked with the game.
Of course it’s not something one wants to change with a 3 year backlog, and the ai review and interactive ai review are quite new features to Ogs, but it may be an idea for others to try this method.
Sure - I just like that I can sort them by category which I find hard to do with sgf, as there is usually different kind of lessons in one game.
I don’t have a 3 year backlog btw - I renew my database every few months (every time my main focus shifts) - I am still improving a lot after all, so it wouldn’t make much sense to look at mistakes from 3 or even 1 year ago imo ^^
And I’m not married to the idea either
I think it is important to check on a regular basis if what you are currently doing/think to know is valid and adjust accordingly (without questioning yourself TOO often ofc, you still have to try something out for a while to see how it works out (for you) ^^)
) produces reviews of 17th- and 18th-century professional games that are, imo, very good.
)