Well, on servers like Tygem I count...









...on the in-game SE to do it for me.

- 30'+, except for last few minutes
- 30'+ + byoyomi
- 1' + 5*30"+, as long as I still have 5*30"+
- for shorter time limits, it depends much on the game development (e.g., if I waste my time on LD, I can then lose it for judgement) and whether I am tired (during the night, I might simply be too tired for doing judgement)

I counted more a few years ago when I was more serious about trying to improve. I count several times in almost all tournament games (except for crazy life and death fighting).

In day to day games, I sometimes do and I sometimes don't, rarely more than once (I can count in games with 20:00 + 5x30). I do have a bad habit of only counting when the question is "resign, play the endgame as well as possible, or cruise to victory?" I know I should be counting when it's time to invade/reduce/play safe, but it's hard to motivate myself.

Oddly, I rarely count in turn based games until it's too late.

_________________
Occupy Babel!

Yes, in games where I have the time. I am not so fast, though. I wonder if there are players who can count in, say, 30 seconds byoyomi by doing partial counts in each period and then adding them up. I.e., spend 4 periods counting, 1 quarter of the board each period, but actually play some move each period. I've tried this while watching games, but I find it hard as the count changes as players play. Of course, if I am watching other players, they may also be playing faster than 1 move per 30 seconds even when they have 30 seconds and that makes it harder.

I don't know if partial counting is an actual skill people try to develop or if they just not bother if they don't have the minimum amount of time to do a complete count before their next move.

It's been a while since I read Ogawa and Davies' "The Endgame" but I think there is a diagram where Ogawa says she can count some full board position in a few seconds. By twos. (At least I assume that's Ogawa's claim and not Davies'.) And I'm thinking...okay...I'm not cut out for this game. It was like learning that the standard for basketball players was to be able to jump nine miles into the air. Since then, I've seen some pros who are not that fast and some who just might be. I think Curtis Tang said one of the skills that he learned while studying at go school in China was how to count quickly.

Forget about partial counting, but learn incrementally updated counting etc. Counting can be slow, partial counting saves only 75% of the time, but incremental updates can save most of the time.

It could be somewhat similar to chess and how pros just need a glance at the board to memorise it if strong players are playing. They effectively see shapes and large scale structures and can replicate the board position with far less time than one would need to memorise each piece's position individually. It's similar for go players early in a game if there's just four josekis on the board, they can replicate the board position without needing more than a few seconds to glance at the board. Equally if the pieces are placed randomly in either game or if very weak players are playing, strong players can't memorise the position that fast, or at all in go depending on how far into the game it is.

I suspect a lot of the counting speed of pros comes from them knowing the precise count many relatively common shapes give on the board reducing the amount of manual counting they have to do in a pro game by quite a substantial margin. But if you presented them with a 30k game I suspect they would be substantially slower giving you an accurate count if they could do it at all because as said earlier on the forum "too many stones will be in the wrong places."

I am able to estimate stuff near yose, if the difference is larger than 10pts.

Can you guys point out some counting techniques ?

There's not a lot of technique to it, just counting 1-2-3. Some people count in groups of 2 or 5 to make it faster. Depending on the point in the game, you can just drop imaginary boundaries from stones to the edge of the board, or you can try and figure out the average of yose if each side plays, which is more involved. One of the keys is to make sure that what you are counting is actually territory, though, and not a moyo that can still be invaded. This comes with experience, but generally 4th line stones do not secure territory against the edge unless there are third line stones backing them up or they are very very thick.

As Robert suggested, it's also good to not get a total count, but a count for each territory region, so that you just have to change the mental count for one region when something changes, and then add them up. Also, don't forget to add 2 for dead stones on the board (1 pt territory + 1 prisoner) and to add the komi to white's score. Don't try to get too accurate a count in the middle game, because things can change a great deal; Just estimating each group to a multiple of 5 is close enough, generally.

The advantage of counting in pairs is that a dead stone counts as one. The advantage of doing a full-board count is that you don't need to have more than one number in your memory at once.

You need to learn positional judgement from scratch. There is a lot of technique, and multiples of 0.5 are good enough.



The topic has been mentioned a few times during the previous months.

Also worth mentioning here is that there is a very good book available which deals with this subject, if you're willing to spend some money. PDF file is cheaper.
You can find the link here: http://home.snafu.de/jasiek/

Highly recommended (by me, if this is of any worth.)

PS>
Fair is fair.

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

I don't count at all. And I'm just getting to the point where I realize what a handicap that is.

Hm, I want to say no. But basically I just try to judge by how big the areas look. If they seem to be about equal, cool. If not, and I'm behind, time to try and do something crazy.

So I'm not sure if you'd call that counting or simply positional judging.

_________________
Slava Ukraini!

One makes an average error per "counted" region of, say, +-X points. For several, say, R regions on the board, the errors might accidentally cancel each other or accumulate. Therefore, the total counting error is roughly between 0 and R*X points.

If X = 0.5, then R*X is still small enough to make good strategic decisions most of the time. If X = 5, then R*X almost always is so big that many strategic decisions must be wrong.

Therefore, just guessing counts or counting imprecisely is fundamentally wrong.

Even if R = 2 (each player has only 1 region), X = 5 lets be R*X = 10, and the error can be between 0 and 10, i.e., somewhere near 5 points. Already then, one cannot make good strategic decisions between enlarging one's moyo or reducing / invading the opponent's moyo.

There is only one way to meaningful strategic planning: rely it on accurate judgement!

So why is there so much talk about counting very imprecisely or by guessing? To improve, learn precise judgement incl. precise territory counting!

My guess is that this conversation is mostly about kyu-level counting (no offense.) Assuming that is correct, knowing you might be 5 points ahead (or behind) still does not necessarily bring you any closer to choosing good strategy. More often the game is tilted one way or another my much more than 5-10 points, and what's more - this tilt changes hands as games goes by, sometimes multiple times. I would say that in such situations guesstimate is as good as solid counting.

I think that even in dan-level games, knowing that you might be 2 points ahead by early middlegame means squat - because of the many subsequent mistakes, most of them of a magnitude much larger than this point difference. These mistakes might cancel or accumulate, depending on luck.

Your mileage might vary.

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

Yilun Yang's advice to kyu (and up to 2 dan, I think) was, "Don't count. Just play the best move." If my paraphrasing is wrong, blame me, not him.

Certainly, I (and he) understand that if you know whether you are behind or ahead, you might change your aggressiveness. But, in amateur games, we all make so many exploitable mistakes, that one doesn't need to overplay to stage a comeback, just exploit mistakes.

Finally, if you are so far behind that you can't catch up through "proper" play, you are probably far enough behind that you don't need to count.

That said, I am sure high-dans should count. Someday I hope it's a problem for me.

_________________
- Brady
Want to see videos of low-dan mistakes and what to learn from them? Brady's Blunders

Positional judgement (and its counting) is not only for high dans but is also for SDKs and low dans. It is one of the important fundamental skills for strength and improvement, besides strategy, tactics, reading and psychology.

Positional judgement is important for
- knowing who is ahead or whether the count is equal, so that strategy can be chosen well accordingly,
- planning and comparing the effects and resulting positions of different possible moves and sequences, to choose the next move well.

Therefore it is essential for everybody - not something to be postponed for later.

Reading is the ability to imagine sequences and construct follow-up positions. Strategy and tactics choose among the latter. Positional judgement lets such choices be meaningful, whenever trivial characterisations (such as "kills" versus "does not kill") do not suffice for distinguishing well qualities of the choices.

I find that counting helps me find the biggest move more easily, so I count regularly during my games. It helps you look away from your local position and see where most of the potential is. There have been games where I didn't count at all, and suddenly my opponent has a big framework which is really difficult to reduce. If I had counted during that game, I would have noticed that there was a framework forming and could have played accordingly.