Bottleneck Theories / Discussion

Charles Matthews To contrast with the discussion at rank and what you know, one can look at the idea that a player stalled at one particular level may lack good comprehension of just one or two key concepts.

There is no particular reason to think that at most levels those ideas are determined in advance. In fact that is relatively implausible, on a global scale. It can easily be that some players are held back because (say) they can't see a snapback. That is unlikely to be true of a community of players; and not likely to be as true, for example, in China and Japan. Where playing styles differ, as is common observation, so different matters affect apparent level.

To take another example, endgame skill is going to be less critical in an environment where players fight all-out in the middle game. Mostly dead groups will matter, up to about dan level.

So one ought to try to filter out factors depending on the individual. Not so much is left. I think for example understanding of light and heavy is vital in the 4 kyu - 1 kyu area.

exswoo -I was thinking about the same thing when I made up the other topic so I'm glad you brought this up.

Kageyama in the Lessons in the Fundamentals of Go says that there are (at least in his experience), 4 main bottlenecks player go through:

1st- 12-13k
2nd- 8-9k
3rd- 4-5k
4th- 1-2k

This idea seems to be pretty dead on when I take a look at the average dispersion of ranks in KGS and IGS...which again goes back the question of what the big difference is between players of players of 12k, 8k, 4k, and 1k ;)

I'm a bit surprised that he didn't note any dan level players in a bottleneck situation. Perhaps he is implying that once you get to dan level, you're bound to make progress sooner or later?

Matt Noonan: I think the conventional wisdom here is the dan ranks are one big bottleneck, just a slow grind on your way up. Still, there seems to be a fundamental difference between a 1 dan and a 4 dan, so maybe there are jumps in the dan ranks too. I can't speak for myself -- I've hit every bottleneck listed above in the kyu range at some point or another, but I haven't explored the dan bottlenecks yet :)

Bill: Worrying about bottlenecks is a way of psyching yourself out.
I do not mean that players do not reach plateaus, stay there for a while, and then move on. Often learning is going on, even on a plateau. Sometimes unlearning bad habits or views is necessary to advance, and that can take a long time. The same goes for areas of neglect or weakness.
Specific insights can make for rapid progress, even at the dan level. When I was shodan one of my teachers told me that I could be 2-dan just by deciding to be. I did not believe him, but later, after being 3-dan for a year and a half I just decided to be 4-dan. It worked! ;-)
It is better to psych yourself up than psych yourself out. :-)

dnerra: I have to try this out! At the next tournament, I'll be 4-dan. I'll report...
By the way, nice to have your contributions again, BillSpight!

Bildstein: I think the rating system in your club or go server can make a difference. For example, the IGS rating system is notorious for becoming very unresponsive when your rating is well established, and it's quite common for people to create new accounts just because they have improved and the rating system will not respond.

I've recently been thinking that this is one of my current problems. I.e. that I'm actually stronger than IGS 4 kyu, but the rating system is not responding to this. And I think it's somewhat self-perpetuating in the sense that I'm continually playing against players slightly weaker than myself, it makes me cocky, and hence I end up playing quite flamboyantly.

Reading this now, I think there's a good chance that I'm just deluding myself. But I've decided to become 2 kyu. At least then I won't be in a bottleneck. And so far so good. The last three games I've played (in the last two days) have been against a 2 dan, and two 2 kyus, and I've won all three. I'll return later and report on how it goes.

... Okay, it's "later" now. I was promoted to 1 kyu on IGS fairly quickly, and I haven't been demoted yet. I haven't been playing a lot, but I certainly feel like a 1 kyu now. Sometimes I have the sense that I'm over-rated and IGS will eventually figure it out, but then I have a serious game against someone and review the game afterwards and realise I did okay.

Charles I'd agree with Bill to the extent of saying that the 'bottleneck' is primarily a coaching concept, rather than a self-study concept.

exswoo Why not just be 9p instead? ;)

(no room left for improvement, of course -mattn)

HandOfPaper 9p players do have room for improvement, even if it's not rank improvement (not counting winning the Judan title, of course). Why not be someone who possesses kami no itte?

adamzero I went straight through from nothing to 1k/1d in the last year, but when I had bottlenecks, they meant that I *had* come to the realization of a new concept, but trying to use it meant shaking up everything that I did have going right, and not yet correctly applying the new concept, thus making me perhaps weaker than I was before. After a little time and a few pointers from the better players, however, things would coalesce and I'd jump a stone quickly as I refined my use of the concept and integrated it into my game. I don't think I've improved much in the last month or so, but, as Dieter said, its more that I need to play more than to study more concepts. Perhaps players who seriously plateau do so because they're not willing to work through that period of getting used to new ideas? They think their losses during experimentation are evidence that their new ideas are bad, rather than good but not worn in yet?

And I agree with those who have already said that there are no particular techniques or concepts peculiar to each level. To draw from my own experience: I started playing go with two other guys and we are all now of about equal strength. However, my fuseki and middle game strategy is strongest, Robert is the strongest fighter and sees more tesuji, and Vinay can win a terribly lost game in the endgame. We're a mini-spectrum of specialties, but we match up about 50-50 wins/losses.

AshleyF Just for fun, I took the game archive from NNGS and did some analysis. There are 300,000+ games from 1995 up to now. First, it's interesting to see the rank distribution (See also RankWorldwideComparison). This is by game, not by player (could mean that 4 kyus play more often rather than that there are more of them):

Then tracking individual players, we can see the rating progress. I did this based on the average time spent at each rank, not necessarily by tracking individuals for the whole 8 years. I was hoping to see the clear bottlenecks but, if they're there, they're very subtle. It is clear that up to ~14k, progress is noticably quicker:

I was suprised actually to see that it takes 6+ years to reach shodan. Certainly, individual results may vary :-)

Charles Interesting - but I think some bias from the way improving players continue to play, while a player stalled at a level may drop out.

JohnAspinall More statistics, arguing in favor of universal plateaus, can be found here.

Alex Weldon: My own personal experience seems to favor the theory that there are in fact barriers to surmount, or bottlenecks to get through. When I started playing on IGS, I figured I was about 18k* (because people I'd played told me I was about 18k), but I quickly found out I was wrong. When my rank stopped plummeting, I was 23k*. I stayed there for a while, then made fairly quick progress up to 18k*, bounced back, fell to 22k* again, got back to 18k* and stayed there for about at least a month or two. Then, a similar thing happened; I went on a winning streak, gained five ranks in five days (no joke), bounced back again, down to 16k*, then worked my way up to 13k*, where I've been since just before my Christmas vacation. Today, I finally managed to get to 12k*. Perhaps the pattern will repeat itself, and I'll get to 8k* or 9k*, bounce back, then crawl back up and get stuck for a month or two. Or maybe not. But anyway, although I've been gaining, on average, about two stones a month (from 23k* to 12k* in just under 6 months), it hasn't been a uniform gain. I don't make any gain for one month, and then gain four stones the next. I'll update this information as I see my further progress.

abennun? I have a related experience I would like to share: Frequently, my wins and loses come in series. I can somtimes win 10 games in a row, just to lose the same number one after the other! Does it happen to you too?

Chris Ball: I think the easiest explanation is that beginning players aren't aware enough of their mental state to know when they're too tired or irritated to keep playing, and losing while in such a state can often lead to a desire to keep playing in order to finish the session with a win rather than a loss. There's an interesting (if perhaps pretentious) mention of games building self-awareness in an essay by Robert Morrell:

It has only been in recent years that psychologists have moved away from the one-dimensional measure of IQ towards recognition of the multifaceted aspects of intelligence. Chess players have known of this for centuries. Athletes stretch their body's limits, chess players stretch their minds. We are sensitive to its nuances; understanding both its weaknesses and strengths. A chess player learns the smell of self deception the way a runner learns to anticipate leg cramps. We know the blinding effects of greed, and the difference in the textures of confidence and cockiness. We understand careful analysis, but can also trust our intuition and spontaneous insight.

Tamsin: I'm starting to believe that improvement can be blocked by playing the game for the wrong reason, i.e., for improvement. Remember when you started playing and everything seemed so new and enjoyable, because you were constantly learning new things, because every game was an adventure? You improved quickly and as naturally as breathing. Then the desire to win set in, and the desire to become shodan or 5 dan or whatever overcame you, and you placed improvement over enjoyment. Improvement slowed and each game became a kind of torture. Maybe, for many people, the road to improvement can be found by not seeking improvement. I made the mistake I described above, and ruined my enjoyment of the game. Part of that was connected with my medical problems, but now I am a lot better off, and one aspect of that is that I can put playing go in its proper place, as something that is intensely enjoyable, but not something to be done only for the sake of being good at it. Who knows if I shall get any better? Do I really care? I would sooner be a happy 1 kyu than a miserable 1 dan.

Dieter: Hi Tamsin. Desire for improvement is not the problem in my opinion. Desire for rank improvement and pride, on the other hand, are. If you play with the clear intention of learning new things and trying things you think to understand, playing Go is pure joy. The big nuisance comes at rank bottlenecks, when you don't climb the ladder at 1 kyu a month like in the old days. Maybe we have been exposed to ranking too soon and no one was there to tell us that rank doesn't really matter.

Tamsin: Amen to that, Dieter! Seeking improvement to deepen one's appreciation is indeed a great goal to aim at; unfortunately, a lot of people seek improvement merely because they want to beat other people. Improvement should mean better and more entertaining games, with the results being only a secondary consideration.

Imagist: AshelyF's graphs are sort of depressing. After all my efforts I'm just barely above the mode rank (4k is most common, 3k is my rank).

On what Dieter/Tamsin are saying, I think that times when I am able to enjoy the game I improve better. I do have a tendency to focus too much on improvement, and when that happens my improvement generally slows. The opposite seems to be true as well.

Hydro47? I actually find the graphs suprising, I've been exposed to the game for over a year through manga, but haven't even tryed the game till 2 months ago, and now I'm around a 20k on KGS. According to the graph that should take a whole year.

HermanHiddema: I agree. I don't know how progress was measured for this graph, so there might be an error in it. My presonal experience is that almost all people with some talent for the game (and who put in some effort) will reach the first bottleneck (12k) within a year, many will reach the second (8k), some will reach the third (4k) and a few will reach the fourth (1k) within that year.

emeraldemon: One thing a tennis instructor explained to me once: as you attempt to incorporate a new skill into your game, you will often play worse until you master it. For example, let's say I've previously been sheepish about invading the opponent's moyo during the midgame, and generally playing too passive. I decide to correct this by practicing invading earlier and more often. Most likely I will lose a lot of games at first, because I'm forcing myself into unfamiliar and thus more difficult territory: most of my invasions will likely die. Until I learn better how to live small, when to exploit aji, etc. my rank may actually go down. But once I've successfully brought those skills into my general gameplay, my rank will certainly go up past where it was before.