Stuck at 12 Kyu

[palapiku]

One guy I know got to 2d in a year playing only blitz and not doing any tsumego... Just imagine, if he played serious games and did tsumego properly, he could have reached 3d!

[gaius]

A lot of the tips above are good.

One tip to apply at every move: before you play, ask yourself why you need to play there, what happens if you omit the move, and which other moves are urgent (look at the whole board). If, after that, you still play a wrong move, it's okay, mistakes happen. But at least it should never happen that somebody asks you "why did you play that move?" and you didn't have a good reason...

[entropi]

I heard of a lot of people being stuck around 12-11k, 5-4k and 1k-1d.

It is natural that there will be plateaus for everybody. But I find it strange that many people get stuck at similar levels. Why are less people stuck around 15k or 8k or 2k for example? Or is it just a false impression?

It could be that each of these blocks are caused by a similar lack of understanding, as they are concentrated at similar levels. For example the block at 12k may be due to lack of basic life&death, the block at 5k may be the direction of play or efficiency, etc.

It is of course possible that this does not make sense at all, but if that's true this may give people (including me) a guideline to study efficiently.

What did people learn/understand to break each of these blocks?

Any comments from experienced teachers?

[dfan]

I heard of a lot of people being stuck around 12-11k, 5-4k and 1k-1d.

I think that if these blocks truly existed, the graph of number of players by rank (like this one) would have bumps in it, as players would bunch up at those points before moving on.

So my assumption is that a lot of the sense of common roadblocks at particular levels is due to confirmation bias.

[Magicwand]

in my opinion rank reflect the level of understanding of the game.
it will take time for them to gain deeper reading ability and that might be why they plateau.
for example: you can achieve 12 kyu by avoiding getting killed.
but for you to pass that level you have to understand what thickness and influence does.
once you understand thickness and influence you might be stucked at 1k.
once you understand the flow of the game you can pass 1k~1d level.
after that every dan level is a plateau for me.

[entropi]

I heard of a lot of people being stuck around 12-11k, 5-4k and 1k-1d.

I think that if these blocks truly existed, the graph of number of players by rank (like this one) would have bumps in it, as players would bunch up at those points before moving on.

So my assumption is that a lot of the sense of common roadblocks at particular levels is due to confirmation bias.

Makes sense. Even though the graph is not a watertight proof that the plateaus are distributed evenly, it is a good indication therefor.

[Cassandra]

I heard of a lot of people being stuck around 12-11k, 5-4k and 1k-1d.

I think that if these blocks truly existed, the graph of number of players by rank (like this one) would have bumps in it, as players would bunch up at those points before moving on.

You have to look at changes in the declination of the graph !
For example, the number of 6k is about the same as the number of 5k. There seems to be some kind of barrier.

By the way: The static view on the number of players by rank does not say anything about how long someone is x kyu or y dan.

[dfan]

The static view on the number of players by rank does not say anything about how long someone is x kyu or y dan.

It is not the same thing, but it certainly provides very relevant information. When you take an aerial snapshot of a highway, the places with the highest densities of cars are where the traffic jams are.

[Cassandra]

The static view on the number of players by rank does not say anything about how long someone is x kyu or y dan.

It is not the same thing, but it certainly provides very relevant information. When you take an aerial snapshot of a highway, the places with the highest densities of cars are where the traffic jams are.

This may be.
If there are traffic jams, they will move over time.
And every car using the entire road will need the same time.

And - compared to the rank graph - you must not forget that there are a lot of cars not moving any more.

[entropi]

The static view on the number of players by rank does not say anything about how long someone is x kyu or y dan.

It is not the same thing, but it certainly provides very relevant information. When you take an aerial snapshot of a highway, the places with the highest densities of cars are where the traffic jams are.

This may be.
If there are traffic jams, they will move over time.
And every car using the entire road will need the same time.

And - compared to the rank graph - you must not forget that there are a lot of cars not moving any more.

If there are really too many non-moving cars (players with definitely stabilized ranks) this would also explain the lack of bumps around 12k, 5k and 1d.

Therefore, whether the static view of graph says something about the barriers depends on the ratio of stabilized ranks at each rank and the ratio of newcomers. Very difficult to tell. I agree it does not prove anything but it gives at least a rough indication.

[dfan]

If there are really too many non-moving cars (players with definitely stabilized ranks) this would also explain the lack of bumps around 12k, 5k and 1d.

What is the difference between a "player with a definitely stabilized rank" and someone who has hit a roadblock?

If you have some programming facility, it would be an educational exercise to hack up a little simulation of a Go-playing population, including these roadblocks (e.g., automatically reduce the rate of improvement while a player is 5k). Run it for a few generations and take a snapshot of the rank distribution. I would be surprised if you could design a system that both had those roadblocks and didn't display bumps at those points on the rank distribution graph. I've been wrong before, though!

[palapiku]

Perhaps the original poster hasn't hit a roadblock, 12k is just his definitely stabilized rank...

[entropi]

Perhaps the original poster hasn't hit a roadblock, 12k is just his definitely stabilized rank...

No, I am sure he will improve sooner or later if he keeps trying, this is just a roadblock.

But if he stops trying or even stops playing, then his kgs account will of course become a "stopped car".

If we assume that a static graph can say something about the roadblocks, then the distribution of the passive (rarely used and not improving) accounts will distort the graph in that sense because this distribution will be different than the one of actively improving players. If the distribution is the same, then the initial assumption that the static graph can say something about the roadblocks would be questionable.
Roadblocks at 12k, 5k, 1d, etc are defined only for those actively improving players (even if the improvement is slow).

What is the difference between a "player with a definitely stabilized rank" and someone who has hit a roadblock?

By stabilizied rank, I mean the rank of players who rarely use their account and neither expect nor are concerned with improving. As said above, roadblocks are defined only for active players.
I have no idea about the ratio of passive (better say non-active) players on kgs, but I would guess their rank distribution differs from the general rank distribution or from the one of active players.

[Cassandra]

Here are some results of an analysis of the EGF database. This database includes all players in all tournaments in Europe since 1996.

Shown is the mean value of the time in month between the first appearance in the list as "rank" and the first appearance as "rank + 1" (for example, 8k >>> 7k, 2d >>> 3d) for various periods of time.

Be aware that the longer the period is, the more "long-term-developers" are included. So it is natural that the general level of the values shown will increase.

But what is stable anyway are some peaks at very special ranks, as at 11k, 8k, and 5k.

2007.gif (9.73 KiB) Viewed 14973 times

2003.gif (9.87 KiB) Viewed 14973 times

1999.gif (9.9 KiB) Viewed 14973 times

[Cassandra]

Here are some other results, based on accumulated medium values over different periods of time, 2, 3, 4, 6, and 8 years. Highlighted are "4 years".

As before, level of the graphs rises with the amount of evaluated years, due to the "long-term-developers".

What may be more interesting than the still visible peaks are the three different ranges of development "speed".

Speed seems to be constant until approximately 12k, developments becomes a bit more difficult until approximately 3k, thereafter you have to invest much more time than before to get stronger.

This corresponds with the winning probability of "rank" in an even game against "rank + 1".

Monate3.gif (12.93 KiB) Viewed 14928 times

probabilities.gif (7.73 KiB) Viewed 14928 times