Life In 19x19

I noticed that I have difficulties in chosing a weak group to attack (or defend) if there are many (usually there are many in the early midgame).

Since this strongly relates to chosing the right direction of play, I guess it might be useful to know how strong players decide.

The most obvious answer would be "chose according to what you can get from the attack (territory, power)". But in early mid-game it is usually very difficult to see which attack will gain the most. Or it is also possible that there are equivalent options.


There are several options I can think of:
1- The weakest one. But then which one is the weakest? Again several possibilities:
1.1- smallest base
1.2- fewest number of liberties
1.3- highest number of defects in the shape (too far extensions, cutting points, holes in the wall, etc)
1.4- minimum possibilities to connect to another group
1.5- minimum possibilities to make eyes

2- The heaviest one having the highest number of stones (because sacrificing it would be more painful for the opponent).

3- The one closest to your own weakest group(s)

4- There are no guidelines for this, just try to read 30 moves ahead.

5- The strongest of the weak groups (for demonstrating how bold you are)

6- Play backgammon

That's a pure strategy question of which the answer may depend on ones style but I am only interested if there are clearcut guidelines that strong players use for such decisions.

Thanks for eventual answers in advance...

Couple of thoughts from me:

1) If the group you want to attack is settled already, don't attack, it's pointless.
2) If the group can quickly settle against your groups / territory if you try and attack it, it is often better to take away that settling opportunity as your pre-emptive "attack" (normally gaining some points in the processes). If your opponent loses that privelege, they have to run, and everyone knows that you don't want to have to just run away
3) If the group can be killed, and you've read it, kill it. If you can't read the crazy aji (e.g. I can steal it's second eye, but then it runs into the middle of crazyland and I don't know who's going to come out with all their limbs), take away the running space and force it to live on the inside in gote.
4) If you can force the group to run in direction a) whilst you gain influence / thickness / territory in direction b), it's a good thing - provided that the influence / thickness you get is valuable globally, or provided the territory you build is more than you give you opponent.

1) is valueless
2) If you can pick up points in sente in this way, they are normally big enough to be worth playing early on. Sure, you'd like to attack more fiercely, but if you can't kill the thing you may as well force the opponent to take dame while picking up points.
3) If you are reading a group kill, you get 2 points per captured stone, 2 points per point of territory that would have been your opponents, and 1 point for each move you'd have to play to seal the thing in - that's valuable, _if_ you get it.
4) Hard to evaluate. Read out a loose rule of thumb estimate and ask yourself who looks better, and try and evaluate it if you can. This is very hard to do accurately.

Never forget sente. Normally kills are gote, but they are normally big enough for a half decent sized group to be worth it. If you are forcing to run ("2)"), this is normally sente, and should be played with that in mind. If it is sente (and the likelihood of this is directly proportional to the chance that the "attacking" followup you dismissed is now a killing move once you've taken away settling chances), this is only not worth playing asap if you think another way of attacking is likely to be more profitable (even in gote) later - rare, but possible. Always try to read 4) as if it will be gote, unless you are sure of a specific sequence that ends up in sente because of reason X - accidental loss of sente can be crippling in the midgame.

Once you've tried to assess the value of this for each opponent group that you are having homicidal urges towards, pick the one that looks biggest. Rule of thumb is sente points are worth twice gote points (true double sente always should be played instantly).

Just my 2c / 2p

One important thing: even if a group is weak, you don't always have to attack immediately. Sometimes there might be two different ways to play (ie. steal base or seal in) or various different directions; sometimes there are just no really good opportunities to attack while making profit. If you don't want to choose yet, or if you don't see a great way of attacking, why not tenuki? After all, if your opponent must spend an entire move on just solidifying a group, you're also profiting, right?

Also, letting your opponent's weakish group sit there for a while could even give you the possibility of setting up double threats later!

Of course, if attacking becomes urgent, go for it, and make your opponent suffer .

As attacking should only be done for some sort of "profit" carefully evaluating what can be gained from an attack is the first step.

If the opponents has weak groups it means at least, he has to invest at some point, moves to strenghten his group. Depending on the board, this can mean he has to play small gote-moves during the middle game - which is in itself a profit for you.

If you can easily make points and reduce potential opponents territory at the same time, do it. Especially when the weakness of his group means he has to defend and you remain sente. (ex. would be a exentension on the side)

In most other cases i would have a look at the board at large and determine "where is my potential territory/where can i make points?" and "where can i possibly reduce my opponents territory/where are good invading spots?".
Normally i need for all these things power. Then i look how i can get this power by attacking his weak groups.

Attacking is a mean to achieve a specific goal, not a goal in itself. So deciding on the goal comes first, deciding how and where to attack to achieve this goal comes after that.

entropi wrote:I noticed that I have difficulties in chosing a weak group to attack (or defend) if there are many (usually there are many in the early midgame).

If you have a number of weak groups to attack, why attack one of them? (I should have that problem. ) Attack two at once (more if you can).

If you have many weak groups, well, that is a problem, isn't it?

If there were straight forward answers to questions like this, the game would get really borring
There are however some guidelines that might prove useful when making a decision.

here my thoughts:

first the obvious: having weak groups puts you at a disadvantage (so strong groups are positive as well as weak groups of your opponent)
to put it into perspective, as a general rule you can count a weak group as something like -20 points.

there are basically 3 kinds of moves:
1: defending
2: attacking
3: expanding
(combining them into multi-purpose moves is sometimes also possible.)

To decide which one you can ask yourself 3 basic questions:

1: do I have a weak group? if yes, defend; if no, go to next question.
2: does my opponent have a weak group? if yes, attack; if no, go to next question.
3: expand. (play big open area of the board (keeping in mind corners first, side second, center third)

once you decided what you want to do, you need to find the best move to accomplish this goal.

some examples:
1: If you have a weak group you want to defend it. If you can attack your opponent at the same time as defending your own group this would most likely be one of the biggest moves possible.
2: from this we learn that the best way to attack is to play from the weakest side, pushing your opponent to your strong stones. allowing you to keep pressure on your opponents stones while getting stronger in the proses.
3: when playing big empty areas keep in mind that if you create a new group you will most likely end up having to defend it while your opponent attacks you. therefore extending or reducing might not look as big but are often a safe and good alternative that should be considered.

couple other things to keep in mind are:
strong a weak is relative: groups are strong or weak with respect to the surrounding position. sometimes a shape can be strong one tome and weak another time.
sente: sente is very important as it lets you steer the game in a direction you want.
who is ahead?: you need to know this to know how much risk to take. (invade or reduce, complicate or simplify)
attacking is for profit not for the kill. don't expect to kill, usually a kill only happens when your opponent really messes things up.
edge of moyo. this is also often vey urgent in middle game.
splitting: if you can split your opponent into multiple groups this is good, you opponent will need to spend time defending both.
there is a lot more...


attacking when you still have weak groups will usually backfire, because you will be forced into defending, thus giving your opponent the upper hand. he will then defend his weakness while exploiting your weaknesses. I would therefore pick #3 from your list as most important.

This thread has been very helpful. Especially in making me realize that an attack should have a greater purpose than just attempting to kill an enemy group.

entropi wrote:(usually there are many in the early midgame)

I think I may have found your problem...

Thunkd wrote:This thread has been very helpful. Especially in making me realize that an attack should have a greater purpose than just attempting to kill an enemy group.

Attack and Defense, by James Davies, is a book that really made me aware of this theme. It's probably aimed at slightly more advanced players, but I was a DDK when I read it, and it was eye-opening.

daniel_the_smith wrote:

entropi wrote:(usually there are many in the early midgame)

I think I may have found your problem...

This is funny, but it also makes a good point. When I play stronger players, I'm often aghast at how quickly we arrive at a situation in which I have nothing to attack and the only trouble my opponent has is deciding which of my groups to torture first.

In a situation when my opponent has more than say two weak groups, I try to find sente moves against the strong groups, and by them build up strength in my own groups. Sometimes a weak group dies by itself, sometime I get enough stones in place close to one so I can kill it...
If I don't see how to get profit from attacking the weak groups (gain territory, kill, gain strength to kill another weak group) I don't attack.

/Mats

Thanks for all the answers.

In the light of the answers, trying to reformulate my question, I think it comes down to "how do you determine the weakest group", which indeed cannot be answered easily.

But trying to put it simpler by an extreme example. Suppose white has symetrical open positions on left and right sides. There is a single invading black stone on the right side of which the future is not yet clear.
There is a 5 stone black group on the left side which is also not clear.

5 stone group is obviously stronger and more difficult to attack. The opponent has more possibilities to escape and ruin your moyo. But it is for the opponent more difficult to sacrifice (let's say it is heavier) and if you manage to surround it (force it to live small) or even kill it, the profit is big.

Single stone on the right side is easier to attack, it cannot escape and is likely to be captured or be forced to live small. But the opponent can handle it lightly, meaning to sacrifice it and use the aji for invading somewhere else.

Which attack is likely to give more profit? Attacking the single stone and gaining power to attack the other group, or the other way round?

Similar situations occur very often in games and I always get puzzled what to do. Go after the elephant, or try to catch the bird (if I may make some japanistic analogy )

But please don't say "you have to read". Even my 3 year old daughter knows that you have to read but what if you cannot? There are positions even Cho Chikun cannot read. Where does the intuition come from?

A useful guideline could for example be (if the opponent group has that many stones and/or that many liberties and/or a potential eye and/or has an X stone wall facing your moyo, don't attack yet). A simple and practical hint like that is what I am looking for, if it exists of course.

Thanks again for all the answers. Even if I cannot get the exact answer I am looking for, the discussion still helps a lot for understanding the basic strategies a bit better.

But please don't say "you have to read".

Well you have to read You have to read "Attack&Defense" i would say.

Attacking a single stone and probably capturing it (if the opponent choose to sacrifice it) gives you territory/points, while attacking the heavy group, that is unlikely to die will give you power.
Now, are you in need of territory or in need of power? This depends on the overall position, in Attack&Defense it is called "Balance of Power" and "Balance of territory".

But maybe there isn't any good reason to prefer one over the other? Sometimes both options are perfectly valid but lead to different sort of games and which one you choose comes down to personal preference.

Personally though, i would attack the smaller group. Maybe not a single isolated stone, but something small like 2-3 stones.
Attacking something big should only be done if you are strong enough in the sorroundings. Because the attacked group will be strengthed, either by tactical plays to create good shape and eyespace, or by escaping towards the center. You can't profit with 3 moves attacking a 10-stone-formation however heavy it might look, you will be left with a weak 3-stone group facing a strong(er) 13-stone group.
However it is possible to profit from attacking a 2-stone group on the side by a few moves, giving you a solid base for further fights.

Don't be tempted by the big profit you may get from a big heavy group. You don't have to win the game with one single attack, all you have to do is make more points than your opponents - with each single move. So if you can catch a small 10-point profit from a small attack in which your opponent gained nothing (he defended what was already his), than thats good enough in my opinion.
Some people like to take risks, i like to keep it safe. So if i have these two options:
1. Attack something small and gain a sure profit of around 10 points while keeping sente
2. Attack something big which may give me 30 points, unsure wether i can keep sente
I would choose the first option. Other players might choose the second one, because they like the fight, which this may start.
(You can change the 30 to any number you whish, if i think the 10 points in the first are enough to keep/gain the lead, i will choose the first one - or will curse myself after the game if i took the second )

MagicMagor wrote:Well you have to read You have to read "Attack&Defense" i would say.

QFT - Add Kato Masao's "attack and kill" to that too

entropi wrote:Where does the intuition come from?

Intuition comes from experience, and from having read and played out similar situations previously. The rest is reading.