Life In 19x19

Hey guys,

I was wondering, what do you think the relationschip between being able to solve tsumego and rank ?
Or maybe more intresting, what was/is your experience with the relation of being able to solve tsumego and your rank ?

I'm a real tsumego addict... It would be exaggerated that I love it more than playing games but there is a thin line.
Howhever it seems that even though I'm able to solve quite some tsumego allready that my rank is always a lot lower.

For example right now I think I'm about 3k on kgs (I don't know for sure since I don't play a lot on kgs, I play more on tygem).
Howhever I can solve about 1d problems within reasonable timeframe.

For example I'm now running through graded go problems for dans 1 and it is quite doable (I'm getting about 85% correct).
I can only solve up to a part of the 2dan problems though. When the spaces in the problems become too big then my reading starts failing.
It is like I get analysis paralysis of big spaced problems but thats another issue

Would love to hear how it went for you guys,

Cheers,
Otenki

I can speak for myself: on goproblems.com I usually end up in the range of 4-6k, sometimes better if I'm in a good mood. However I'm not yet registered there, if the browser cookie gets lost, I start over at 30k but then it's fun to get back to the 4-6 k range as fast as possible. On KGS I'm currently 5k, so there seems to be some correlation.

I think books commonly exaggerate problem-rank comparisons -- one reason is to make the readers feel better. I would suggest taking the ranks they give you down 2-5 notches, depending upon the book.

As for being able to solve problems and having a solid playing rank, at one time (1-3d) I was doing so many problems that my problem solving skills outranked my play skills. So from the advice of a stronger friend I put aside most of my problem solving for a bit and played almost exclusively games. I discovered that my investment into problems had paid off as I was reading easier and farther than my similar level opponents, but that I hadn't yet integrated it into my overall game. How long it takes you to play catch-up in your games depends on how large of a gap you've made between your problem solving abilities (i.e. reading) and your playing.

I think there probably is some correlation between tsumego solving and rank but it can't be all that strong. Tsumego involves mostly local reading and your rank depends on overall strength. I know people who have learned the tsumego tricks very well but who are only sdk players, i.e. their tsumego rank is mid-dan level but their playing strength is much lower. And the opposite is also true. I know people who have a very good grasp of the opening and shape but aren't that good at local reading so they play at a mid-dan rank but have trouble with kyu level tsumego. I take all the tsumego grades with more than a grain of salt.

gowan wrote: I know people who have a very good grasp of the opening and shape but aren't that good at local reading so they play at a mid-dan rank but have trouble with kyu level tsumego.

I have a very hard time believing that someone who has problems with simple kyu reading can sustain a mid-dan rank.
He will not be able to read any life & death out on the board. This will result in either doing too many moves resulting loss in sente or getting killed. He will also not be able to read invasions well, his groups will be cut because he missed a connection problem etc.

Howhever I do agree that the other way is possible. If you look at some of these chinese 6 year olds who try to become pro, some are 2dan and have no idea about strategy or global possition at all.

Cheers,
Otenki

logan wrote:I think books commonly exaggerate problem-rank comparisons -- one reason is to make the readers feel better. I would suggest taking the ranks they give you down 2-5 notches, depending upon the book.

There is also the question of how fluently you solve the problems.

If a 1d problem means that the player must solve each problem in about 15s, or if it means he must solve them in about 5 minutes

otenki wrote:I have a very hard time believing that someone who has problems with simple kyu reading can sustain a mid-dan rank.
He will not be able to read any life & death out on the board. This will result in either doing too many moves resulting loss in sente or getting killed. He will also not be able to read invasions well, his groups will be cut because he missed a connection problem etc.

I'm going to go out on a limb and guess that by "good at local reading" gowan doesn't mean "good at reading", simply, but specifically good at the sort of intense, highly-branched reading with lots of tricky visualizations that make the hardest tsumego particularly interesting.

A DDK who has been cramming tsumego without playing is probably already bored by under-the-stones, but people who don't do tsumego find that sort of thing to be quite tricky.

otenki wrote: I was wondering, what do you think the relationschip between being able to solve tsumego and rank ?
Or maybe more intresting, what was/is your experience with the relation of being able to solve tsumego and your rank ?

I think that there is a fairly high correlation between tsumego ability and general go ability. I would be surprised, however, if tsumego ability amounted to more that 10% of general go ability on average. For one thing, there are a lot of different skills at go. For another, from my experience and observation, crucial life and death decisions occur before the usual tsumego stage (some very advanced problems aside).

As for my own tsumego experience, I reached 4 kyu (now maybe AGA 3 kyu nowadays?) with hardly any knowledge of life and death. But I bought the Maeda three book tsumego set and worked problems on the subway. By the time I was 2 kyu the Maeda 4 kyu problems were about right. I. e., I could solve about half of them in a minute or so. By the time I was shodan the Maeda 1-3 kyu problems were about right. So I almost caught up pretty quickly. At least as far as problem books are concerned. Over the board is another question.

I feel that the metric to measure "rank" and "tsumego rank" is not derived in the same way, so it's kind of difficult to compare.

Your "rank" in go is based on when you win games. Various ways are proposed to do this, but if you have a consistent mechanism, you can have some accurate predictions on who will win what games.

Problems feel different to me. I think problem "ranks" are given based on a pre-existing go rank. So if you are 1d because you win against 1ds, then you solve this problem that another 1k can't solve and a 2d can solve... maybe someone will call that problem a 1d problem.

But it's not really a fair way to measure problems:
1.) There may be a certain shape a 5k has seen that a 1d has not seen, and the 5k happens to know the correct answer.
2.) Less directly, a given 5k may be more familiar with a certain type of problem that another 1d is not.
3.) If you have someone that is 1d in go... maybe he's 1d because he is good at fighting and he can out-read others. Another 1d in go may suck at fighting but may be pretty good at positional judgment. It'd be likely that a more difficult reading problem can be solved by the first person than the second. So in that case... What's the rank of the problem?

I feel that the "rank" for a problem is simply a reflection of the ranks of the players that solve those problems. So you could say that your tsumego rank is a function F where F(X, Y, Z, ...) takes, as input, the ability for player X, player Y, player Z, ..., to solve the problem in a given amount of time. How else do you get a "grade" for the problem?

With this mindset, I think that it comes down to this:

Given that I am rank X in go based on my ability to beat other players of various ranks, my strength in reading localized problems is greater/lesser than average.

To give an analogy, let's say you are filling a cup with 100mL of liquids. You could fill it with water or with milk or with coffee. Maybe some person fills their cup entirely with water, and another fills 43% water, 17% milk, and 40% coffee. Both people have 100mL of liquid, but one guy has more coffee. Either way, it all contributes to your volume of liquid. To bring it back to go, maybe some 1d have that skill based 30% on reading, 20% on joseki, and 50% on good looks that psych out the opponent. And maybe others are more like 80% reading and 20% fuseki. Both 1d are still 1d.

This is all theoretically speaking now. Practically speaking, how do we measure that someone has a skill of 30% reading vs. 50% reading? I don't think it's possible. The only measurement we have is the ability to win games. And how you do that... Well, that's complicated.

I'd weigh on the side of "meaningful correlation."

The absolute rank estimates mentioned in books of course differ from each other in the same way that rating systems in different countries can have skew. Graded Go Problems for Dans, IMHO, is a little bit generous, compared to, say, some estimates in Yilun Yang's books. So probably pay that no mind.

That being said, from a relative point of few, for any given collection that has been sorted via difficulty, the correlation can be almost painfully clear. If reading ability is your bottleneck---and I'd wager it is for many adult players---then the deltas show. When you can start solving harder problems than before, the results should reflect in your games. If they do not, rejoice! It means you are super lucky. It means a teacher can likely help you improve because something else is holding you back and almost everything else is much easier to improve than reading ability. If you're in that situation, you have a good chance of making sudden improvement with some external help. But if your reading is stuck and your rank is stuck---I can't say the prognosis is good. No one likes to hear the grim prescription of "just work harder."

Remember also that your results in problems vary from day to day depending on fatigue, etc. just like results in games. So you're looking at trends over time, not problem to problem variability. I think earlier comments about just knowing certain techniques have some validity.

I have one more comment on how you solve problems. If you are placing the stones on a physical or virtual board and working through them by trying out moves that way, that's very different from doing them entirely in your head. In the former case, there are some problems that are possible to solve just by trying out every possible combination in, say, a depth-first search way like what is described in Davies' Tesuji. Being able to solve that kind of problem with the aid of a board is a very different skill than what you need to do in a game. Unless you're playing on a turn-based server or something, you can't just try out stuff. You have to solve it in your head. Now I'm not saying that trying out stuff on a board is bad. If a problem is too hard, it's probably okay to do that and I think you will learn something. Maybe you will learn to be that systematic in your head, too. Just don't be too proud of finding an answer that way.