Is it better to review games from 'start to finish' or begin at the end and work backwards?

I don’t usually do reviews. I just enjoy playing :open_mouth: but I was about to start a review when it suddenly occurred to me that it might make more sense to work backwards.

What do others think/do? My reasoning is that having already examined the end/middle game it might be easier to see why moves are bad at the start.

1 Like

If you’re reviewing for someone else, ask what they were wondering about.

The rest doesn’t matter.


If you’re reviewing your own games… I think the important parts are where your opponent did something you didn’t expect.


… so from about Move 4 I have to look closely at every position … :wink:


(Assuming we are talking about reviewing for personal improvement)

In all serious, and taking into account I’m just a DDK, what do I know, it seems hard to me to imagine
reviewing backwards.

Things that occur to me about this are

  • It’s hard to hold all that state in your mind as you work backwards
  • Generally, my impression is that we are asking “was move x good given what we know at the time”
  • The final outcome is only one of an infinite set of possible outcomes from earlier in the game

Working backwards feels like it might place too much emphasis on that particular outcome, as opposed to figuring what is the best choice given what is known earlier in the game at the time.

1 Like

Valid points for sure. I think the value I would look for in reviewing backwards would be along the lines of

At this point I wanted to make move Y, but I couldn’t because my stone played at move X breaks this possibility.
So, I would then look at the position where I played X and ask “could I have gotten the same desired outcome that I got from move X with a differently placed stone that would also allow me to play move Y later”

I think it’s a similar vein to studying Tejun or Order of Play.

On a very basic level, it’s like discovering when to play a hanging connection and when to play a solid connection.


No. A whole-game self review. I definitely wouldn’t do a backward review for someone else without asking them what they were looking for from the review.

Absolutely YES and I suppose we should now add, ‘and where AI recommends something you didn’t expect.’

Not at all. A normal review involves going backwards and forwards through possibilities. We explore a possibility, find that it’s bad, backtrack and conclude that an earlier move is bad because it is refuted by a later move.

So the process is back and forth regardless of whether we go from start to finish or the reverse.

In truth the idea seemed like a silly thought bubble when I first thought of it but then it grew on me.

Sage advice. A backwards review could become contaminated by a kind of sophistry.

Glad you mentioned this. Sometimes, but not often enough, I find a good move by finding a good sequence and then ‘cutting to the chase’ by changing the order.

At this stage I’m thinking I’ll do a two stage review where I skim through forwards gathering rosebuds while I may and then work carefully back to the beginning.


Yea I doubt it matters at all which way you do it.

I mostly do backward analysis in order to find someone’s last chance to win a game.

1 Like

I cannot see the benefit myself. I like starting from the beginning because I can see how the board develops. It helps me to understand everything that comes after it. I find the idea of going backwards harder, simply because I gain a lot of knowledge from seeing how the game develops. What do you think the benefits of deconstruction are?


Do you find knowing this bit of information helps you in anyway personally? I’m curious :thinking:.

Well if someone says “I only needed x more points” and I have to go back 100 moves to find a mistake big enough that correcting it will put that player ahead,… yea. Also it means I only need to post 1 variation.

Personally, I think the main weakness of kyus is our depth of reading, which can also be expressed as our ability to visualise possible futures.

When you work backwards, the stones are already there, and so it’s easier to see the effect each stone has on how following stones were laid out.

Also, you already read the game forwards while you were playing it, so tactics like reviewing the game backwards (or even from 90 or 180 degree rotations) can help you see things you missed during play (and possibly also that your opponent missed).


This is so simple, yet it never occurred to me. Now I wish OGS had a feature where you could rotate the board! Because this makes truck loads of sense. Given that we strive to analyze and memorize patterns, seeing the same patterns from different angles would be very helpful. This idea really excites me. Thank you for sharing :nerd_face:.


@Eugene @anoek how hard do you think it would be to implement? I’ve lodged it as a request on GitHub.


I’m likely to want this for joseki dictionary anyhow, so there are multiple reasons to have it :slight_smile:


For reviewing one’s own games, professionals go through the game in reverse.
The reason is that they start from a known outcome and hard facts, and not a possibly flawed interpretation of a middle game position.

I don’t remember where I got this from; it might have been Myungwan Kim 9p in a game commentary video.

It makes sense though.
Game results will frequently be in the range of -1.5 ~ +1.5 points. If you discover a mistake towards the end, it will really have been game-deciding.
They can also skip easily over the forced end game moves and go to the last interesting position.
The opening is probably less interesting for a professional because all knowledge is vague. Even while they might be trying new things, it will be much harder to identify a mistake.

I believe the same principles hold nowadays even when they are using strong AI to help them.

Final note: reviewing your own games in any direction is hard. Get a stronger player to look at it.


This is exactly the kind of explanation I was hoping for. Thank you for sharing it. I found this to be extremely helpful :hugs:


AFTER doing it yourself first. That “hard” bit is the real value of the review, a stronger player should only help fill in what you missed, not replace the entire system of self review.


I dunno, most ddk self-reviews I’ve seen looked pretty useless.

If you want to self-review as ddk, you should probably focus on meta ideas and refrain from trying variations.

  • If you got surrounded and died somewhere, do more tsumego to more readily notice when a group is in danger. Also, try to find out what got you surrounded and try to avoid similar mistakes next time.
  • If you messed up a corner sequence, look up a simple relevant joseki and replay it a few times so you’ll remember it next time.
  • etc.

I mean, if your argument is they’re simply too weak to do an effective self review, then the same will be true of their attempts to do tsumego and joseki…

the point isn’t to “get the right answers” the point is to develop your reading and critical thinking… they’re new to the game, the review is allowed to be less detailed than what a dan would review, but I’ve also seen dans give DDKs completely stupid high level advice that could NEVER be applicable in an actual game… self review is important because it’s always at the right level for the player… AFTER self review, I think the ideal strength for an external review is 5-10 stones stronger than your own… personally I think there is very little a dan can teach a TPK because their worlds are just too far removed, they’re almost playing different games…


No, that’s not what I said. What I said was:

If you want to self-review as ddk, you should probably focus on meta ideas and refrain from trying variations.

If you can’t read 1 move ahead in an open situation, you should work on reading 1+ moves ahead in a constricted situation. DDK are very well able to say “oh look, there are lots of black stones on top, I probably should have put one in there earlier”.