Mistake blindness.

Toge · Post by **Toge** » Fri Jun 03, 2011 7:14 am

A great block to improvement is to not recognize mistakes one makes in game. If we could figure them out, we wouldn't make them in the first place. So I wanted to ask is there some systematic way to evaluating sequences in postgame review. How do professionals improve their game? Surely they can't depend on problems and books written by themselves.

Related question: is tesuji a solved sequence?

example: let's say the aim is to capture the marked stone to make eye for a group that is in danger. Playing at "a" makes shortage of liberties tesuji appear (white plays "b" and black "c"). Black could also capture straight away with "c", but ladder tesuji at "b" afterwards wouldn't work. So the correct way is to play "a".

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQcOb.... $$|..XXXa..... $$|..OOO......[/go]

edit:
note to self: never try to make illustrative example.

EdLee · Post by **EdLee** » Fri Jun 03, 2011 7:45 am

What about d,e,f and the rest of the board below :wc:

?

Click Here To Show Diagram Code: [go]$$ $$|........... $$|.OOOOOe.... $$|..XXXXd.... $$|..XO.O..... $$|..XXX...... $$|..WWWf.....[/go]

Bill Spight · Post by **Bill Spight** » Fri Jun 03, 2011 7:48 am

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ1O..... $$|..XXX...... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O3.... $$|..XXX1..... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O1.... $$|..XXX3..... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O3.... $$|..XXX...... $$|..OOO1.....[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX1.... $$|..XQ2O..... $$|..XXX.3.... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O3.... $$|..XXX...... $$|..OOO.1....[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O3.... $$|..XXX.1.... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O3.... $$|..XXX..1... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX..... $$|..XQ2O.1... $$|..XXX.3.... $$|..OOO......[/go]

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO..... $$|..XXXX.1... $$|..XQ2O..... $$|..XXX.3.... $$|..OOO......[/go]

Have I missed any?

Edit:

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO1.... $$|..XXXX..... $$|..XQ2O..... $$|..XXX.3.... $$|..OOO......[/go]

Thanks, Ed!

Click Here To Show Diagram Code: [go]$$c $$|........... $$|.OOOOO.1... $$|..XXXX..... $$|..XQ2O..... $$|..XXX.3.... $$|..OOO......[/go]

There are more, aren't there?

Toge · Post by **Toge** » Fri Jun 03, 2011 7:56 am

EdLee wrote:What about d,e,f and the rest of the board below ?
$$
$$|...........
$$|.OOOOOe....
$$|..XXXXd....
$$|..XO.O.....
$$|..XXX......
$$|..WWWf.....

Click Here To Show Diagram Code
[go]$$ $$|........... $$|.OOOOOe.... $$|..XXXXd.... $$|..XO.O..... $$|..XXX...... $$|..WWWf.....[/go]

- I made the example in a hurry, but the point is: is it better to play a move that enables a (working) tesuji, accomplishing the objective rather than some other move that accomplishes the same objective? Difference between proper playing and improper playing can be very small, like leaving behind a slightly bigger ko threat for opponent.

Okay here's a better one. Snapback versus no snapback:

Click Here To Show Diagram Code: [go]$$c $$......XO...| $$......XO...| $$......XO...| $$....X.XO...| $$.....XOXO..| $$.....XOab..| $$-----------|[/go]

Bill Spight · Post by **Bill Spight** » Fri Jun 03, 2011 8:11 am

I am put in mind of a statement by bridge champion, Terence Reese.

The man who plays up to himself is hard to beat.

The same goes for self-analysis. Do the best you can, you can't ask for more.

One thing that Ed's and my posts show is that it is good to consider all the plays you can that accomplish your goal. Unless you can read the play out to the bitter end, you have to fall back on your judgement. Post game review with your opponent can help both of you develop your judgement.

Bill Spight · Post by **Bill Spight** » Fri Jun 03, 2011 8:14 am

Toge wrote:Okay here's a better one. Snapback versus no snapback:

$$c
$$......XO...|
$$......XO...|
$$......XO...|
$$....X.XO...|
$$.....XOXO..|
$$.....XOab..|
$$-----------|

Click Here To Show Diagram Code
[go]$$c $$......XO...| $$......XO...| $$......XO...| $$....X.XO...| $$.....XOXO..| $$.....XOab..| $$-----------|[/go]

You know the answer.

Toge · Post by **Toge** » Fri Jun 03, 2011 8:33 am

Bill Spight wrote:You know the answer.

- Answer to that question, yes. I wanted to know does it work because snapback is a tesuji, the only tesuji available in the position. Should I always make a snapback when possible, instead of capturing the stones right away?

daniel_the_smith · Post by **daniel_the_smith** » Fri Jun 03, 2011 9:00 am

Toge wrote:...Should I always...

This is go. You know the answer to that is always "no", no matter what comes next.

Usually.

Bill Spight · Post by **Bill Spight** » Fri Jun 03, 2011 9:28 am

Toge wrote:
Bill Spight wrote:You know the answer.

- Answer to that question, yes. I wanted to know does it work because snapback is a tesuji, the only tesuji available in the position.

No. A play is tesuji because it works.

Should I always make a snapback when possible, instead of capturing the stones right away?

Given what? Obviously there are positions where you should capture immediately, but they are rather different from the diagram.

On rare occasions you will see in a pro commentary, such-and-such a move is tesuji, but it was a mistake.

If you are really in doubt, would you rather say, I lost the game because I played tesuji, or, I lost the game because I saw the tesuji but didn't play it.

Toge · Post by **Toge** » Fri Jun 03, 2011 10:06 am

daniel_the_smith wrote:
Toge wrote:...Should I always...

This is go. You know the answer to that is always "no", no matter what comes next.

Usually.

- There is a certain type of person who likes all uncertainty and dislikes absolute truths. But saying that there are no guidelines, principles and rules of thumb is taking it too far. There exists absolute knowledge and that knowledge is represented in mathematics and logic. Go bots operate on logical inferences. The better their playing logic, the stronger the bot.

In that snapback example we 'know' the answer. How can we be sure? Did we go over all possible moves starting with a1->a2 ... a1->a3? No. Our attention was fixated on the stones that were in atari. There was a tesuji that allowed black to preserve the stones while taking additional territory of white's. Had we not understood the tesuji, our result would have been worse. So there seems to be something inherently good about playing the snapback. If one were to design a bot that could detect all snapback situations and suggest a move that makes the snapback happen in a local context, would it be in this local context the best move? Not always. If there's a ko fight, making snapback instead of capturing creates additional threat for opponent. For that matter any two stone shape that is not nobi can be cut with two successive moves. What if there is no ko?

I created this thread because I thought this was common knowledge for strong players and I wanted to be sure I understood the nature of tesuji correctly. I don't believe that person is a good Go player because their brains operate at higher rate, but rather because they have better heuristics what to look for.

Website · Post by **Chew Terr** » Fri Jun 03, 2011 10:26 am

For your first question: I have trouble with this, too. The best thing I can do is to look backwards: check where territory developed and what died, then see if anything directly or indirectly led to that. Also, I try to play through while watching the groups as wholes, seeing which are strong and which are weak. A lot of times, I don't notice my groups get weak in games, or I play stones to help groups that are already strong, but hindsight lets me catch it. That's really all I've got, but I think practicing reviews (for yourself, or weaker players, or even stronger players) does seem to help you notice flaws in your own games.

In response to your second question, there seem to be a lot of times when the 'normal' answer is better than the 'fancy' response. In your example picture, if capturing is enough to live, it may be best, because it leaves the least aji. There are certainly times when the tricky responses are best.

Using the example of snapbacks (which seem to be one of the tesujis that is more often worth doing), I can see only two reasons not to play a snapback: ko threats, and shortages of liberties. It's often easy to tell if a group might have liberty shortages later (especially if you're reviewing the game afterwards). For ko threats, you can look to see if a big ko already exists, but I feel like it's often worth moving slightly faster (which snapbacks tend to do) in exchange for granting a ko threat.

EdLee · Post by **EdLee** » Fri Jun 03, 2011 10:38 am

Toge wrote:I made the example in a hurry

Haste is not good. In Go, or in general. Usually.

Toge wrote:How do professionals improve their game?

See John's post 22: viewtopic.php?f=48&t=3844&start=20

Life In 19x19

Mistake blindness.

Mistake blindness.

Re: Mistake blindness.

Re:

Re: Mistake blindness.

Re: Re:

Re: Re:

Re: Re:

Re: Re:

Re: Re:

Re: Mistake blindness.