Hypothesis - Tips: The Main Difference Between TPK--DDK--SDK--Dan

My overly-simplistic hypothesis is this:

TPK-DDK: The main difference between TPK (20+kyu) and DDK (19kyu…10kyu) is a basic understanding of ‘Eyes’ and ‘Life & Death’.
One of many helpful links on this subject: https://senseis.xmp.net/?LifeAndDeath
So it may truly be said that, ‘Getting up to DDK is a matter of Life and Death.’

DDK-SDK: The main difference between DDK and SDK(9kyu…1kyu) is a good grasp of the initiative. Not just sente&gote but also 'Double-sente, Reverse-sente and Double-gote.
One of many helpful links on the subject: https://senseis.xmp.net/?Initiative

SDK-Dan: The main difference between SDK and Dan is consistency and the ability to prioritise and integrate the issues across the whole board into a single game.
This is the category that I am least qualified to comment on. I can only go by how I feel I am typically losing against Dan players. I particularly look forward to any comments from Dan players on this. After watching one of smuph’s games I was going to say, ‘knowing when to tenuki’ but that is just a better understanding of the initiative.

ps. Dan-Pro: :open_hands: Not even going to guess at this one.


I agree with you that basic life & death is the main thing separating TPK (I always forget what it actually stands for, I can only think of “total party kill” which is when all your D&D characters are slain by owlbears) and DDK.

I don’t agree with you that the main difference between DDK and SDK is initiative. I think actually it’s an improvement in shape knowledge that lets players make stronger groups.

As for SDK - dan, I’ve not crossed that boundary so I can’t say for sure. But my hunch would be that the main thing is in fact timing – not just what to play but also when to play it.

Then finally, the biggest thing separating dan and pro might be positional evaluation using counting of positions that aren’t on the board yet. I think that is one of the big things that help them to make wise choices in the context of the whole board position.

But, that said, we’re all different people and we’re all better (and worse) at different things. What I will gain most from working on is not what another mid-SDK will.


For Reference:
TPK – Twenty Plus Kyu – 20k and weaker
DDK – Double Digit Kyu – 19k–10k
SDK – Single Digit Kyu – 9k–1k

Note: Sometimes I describe my gameplay as that of a TDK – Triple Digit Kyu :wink:

(Depends on the measuring scale, doesn’t it)

Have you looked through a lot of TPK games lately? I watched a few specifically for this thread and I realized that I have no idea what the difference between TPK and DDK is, they all seem to play pretty randomly.


Shape definitely deserves a mention in there somewhere but I regard my own understanding of shape to still be extremely limited. ‘Shape’ was one of my candidates for SDK-Dan but as you say,’…we’re all different people and we’re all better (and worse) at different things.’

I think good shape also affords you more of the initiative so we may simply be reaching the same conclusion from different directions.

Thankyou trohde. This is the least well known one perhaps.

A brave comment. :wink: It makes me wonder what actual ranking random play would achieve.

1 Like

Leela Zero started out on OGS with almost random play. As you’d expect she had a beginner rank like 25k.


I believe there is a huge difference in really random and what one perceives as ‘almost random’. As black a truly random player would open about one in 5 games on the edge of the board for example. Surly something you’ll never see from an 10k (and most likely not even from a 20k). A truly random player wold almost never respond to an atari. A true tenuki expert! :slight_smile:


In her first couple of weeks of playing, LZ made a lot of early moves on the first line. I wouldn’t be surprised if she had opened the game there now and again. And most of the time at that point she didn’t respond to atari. You’re talking about a self-play bot starting from nothing, just like AlphaGo Zero.


Sure, I am aware that LZ really started from randomness. But from where do you get the information that this translated to 25k? I don’t see any translation from the ELO points LZ stats seem to use to ranks. Have you found anything connecting the 2? I’d be very interested in that.

Is “TPK - Twenty Plus Kyu” a relatively a new invention?

Many years ago, I would only see people talking about DDK and SDK, while I’ve only started to see TPK used in a past couple of years or so.

For example, two popular groups that have been around on OGS for a long time are the “Double Digit Kyu Room” and the “Single Digit Kyu Room”, but I’ve never seen a TPK group.


It’s hard to draw a distinct line between 20 kyu vs 19 kyu, or even 10 kyu vs 9 kyu, but I agree that some generalizations might be possible when comparing say 25 kyu vs 15 kyu vs 5 kyu vs 5 dan.

I think the jump from 15 kyu to 5 kyu is a lot more significant than the jump from 25 kyu to 15 kyu, and I think takes a lot more than just a good grasp of initiative.


So I’m SDD? Got everything memorized. :smirk:

From my POV:

DDK usually have

  • basic understanding of some heuristics (corners, sides, center)
  • seen some joseki and tsumego
  • have developed some sort of feeling of where they want to play
  • lack or quickly lose view of the entire board
  • don’t have the shape intuition to reliably predict and therefore read sequences
  • can’t tell if a move is forcing or not, i.e. they respond to every move (ain’t that right, @opuss :smiley: )


  • know a great deal of heuristics
  • have done quite a few tsumego
  • have a lot of game experience, know common non-joseki sequences
  • developed some idea of sente and gote
  • they’re still very easily manipulated into being greedy, envious, defensive
  • can’t read very deeply and therefore
  • don’t know how to handle unusual situations / when things usually considered bad are the ‘best’ option


  • usually loads of game experience
  • strong fighting shapes
  • deep reading, usually recognize life&death in <8 moves right away, can read ~12 forced moves
  • fairly independent of joseki because local outcomes are easy to read
  • decent understanding of sente
  • fairly well-developed global assessment
  • know how to come back from a bad position, maybe due to experience with over-/trickplays
  • as a result of massive game experience, usually developed bad habits > unrefined style
  • bad endgame - no one wants to practice it because it’s tedious
  • lazy => stagnant, because improving takes massive amounts of effort at this point

Quite right! This results in either responding too much or failing to play a move when it is necessary.

I also have difficulty with direction of play.


I see what I said was too ambiguous. Leela Zero, right from the beginning, was playing on OGS as RoyalZero. What I’m saying is that when the RoyalZero account was playing almost-random, it had a “beginner rank like 25k” on OGS. When I am talking about rank here I’m talking about Glicko rating on the site (or perhaps it was still Elo then), not using imagined rank as a proxy for strength. I don’t remember exactly what it was because TPK ranks are unstable, but it was something beneath 20k.

1 Like

Not sure exactly where they are on the DDK-SDK-Dan spectrum, but I’ve noticed that the following make a huge difference. The SDK bits are mainly a reflection of my own weaknesses, the Dan bits are what I think I should do, and the DDK bits are what I think I’m not doing anymore:

  • Direction of play

DDKs focus mostly on small areas of the board, and play very small moves when they should play elsewhere. SDKs play more global moves, but aren’t great at reading out problems that they leave behind, leading to many unnecessary fights. SDKs also have a tendency to go for “big moves” that are fairly passive and just build territory, rather than ones that legitimately pressure their opponents. Dans are better at attacking the important parts of the board, and not leaving weaknesses.

  • Shape

DDKs have minimal ability to understand shape. SDKs understand shape on a basic level, but still leave lots of weaknesses. SDKs are also bad at realizing when a “strong” part of the board becomes weak, as a result of fighting somewhere else. This leads to previously secure groups getting nuked when ladder breakers/influence/etc. shows up elsewhere, and the appropriate steps for fixing the weaknesses they expose aren’t taken. Dans have a more global view of how groups interact, and do a better job at steering fights elsewhere to minimize their own shape defects and expose those of others.

  • Lightness

For DDKs, almost everything is heavy. DDKs will go out of their way to save very small groups, or even single stones, that don’t affect the game in a global sense or have any real meaning for the life and death of groups. SDKs are better at abandoning a few stones, here and there, but don’t have the reading skills necessary to play lightly in the first place. Lightness is more incidental than intentional. Successful use of light stones is at a very basic level. Dans can effectively read, play, and use light stones.

The ability to understand and use these is a spectrum, of course. In general, I’d say that DDKs have little to no ability to understand and utilize these concepts, SDKs know them, and occasionally use them, but in a simple way with many mistakes, and Dans can effectively employ these principles in their games. Right now, I’m firmly in the middle stage of those, understanding my weaknesses is hopefully the first step in moving past them.


… and TPK do not have these things. That’s the main difference there.

(and why a distinction is needed).



What a great, eloquent post! Your point about direction of play being a weakness for DDKs perhaps ties in to what was said earlier about SDKs being better at taking sente, and about strength of shape (weak shapes need lots of defending.) And I also think you’re right about lightness. But to me that is tied up with timing to an extent: knowing when a stone or group is light enough to leave simply because the opponent can’t afford to spend sente capturing it. Like a probe under a shimari.

Most of all though, this rings true to me: “SDKs are bad at realizing when a “strong” part of the board becomes weak, as a result of fighting somewhere else. This leads to previously secure groups getting nuked when ladder breakers/influence/etc. shows up elsewhere”

Out of anything this might be the hallmark of the SDK game.


tl;dr: Go is a game that is fundamentally about flexibility, reading and positional judgement. From these arise concepts such as exchange, urgency, big moves, thickness etc… Players across the spectrum have various degrees of mastery in these fundamental things.

Having had a two year journey starting at the bottom of the TPK ladder to somewhat of a solid shodan player, here’s what I have come to learn about the rank ranges.

Of course, these are my personal observations only. It is by no means an objective statement.

TPK (25k-20k)
New players who are still trying to understand the bare basics of stones, edges and corner intersections.

Typical characteristics:

  • Will try to save every stone

  • Will try to run out an unfavorable ladder

  • Sometimes do not see ataris or self ataris

  • Next to no concept of territory or moyo or anything beyond what is physically present on the board

  • Highly devoted to capturing stones and avoid being captured

  • No concept of weaknesses, shape or haengma

DDK (19k - 10k)
Players in this range have either played for merely few weeks or months to years or even decades.

Key improvements from TPK:

  • Recognized Go is about making points than capturing stones.

  • Knows the bare minimum of life and death

  • Knows few commonly played shapes to maintain connection or break connection (usually by experience from past games)

  • Rough idea of what is territory, what is moyo (especially 13k+)

  • Understands basic go theory and few common josekis

  • Will sometimes defend a cutting point in their own position

  • Knows how to use some basic shortage of liberty aji during endgame.

Bad habits and weaknesses that leads to rank stagnation:

  • Weak understanding of shape (no concept of vital points of stones)

  • Laziness (playing moves that seem to work from past experience instead of reading at every move to make an informed choice)

  • Inflexible thinking (I have stones here so this area must be my territory, if you invade in my area I will kill you. If you reduce me I will hold on to what I have instead of making a trade.)

  • No concept of reasonableness (Will invade in gaps larger than two spaces apart), this is a common habit further encouraged by the opponent not knowing how to punish severely.

  • Tends to follow where the opponent played last

  • Pays no attention to the whole board position, thinks very locally.

  • Plays aji keshi and erases almost all good aji in every position.

  • No concept of stone efficiency

  • Creates extremely heavy groups everywhere, no concept of light play or sacrifice.

I want to say that the longer a player remained DDK the more profound these mistakes are.

Weak SDK (OGS 9k - 4k)
Players who have played for a few months or a year can usually reach this stage if they are critical of their games and keeps an open mind for learning new things.

This is also by far the most volatile level gap. There are players of all styles and habits here because different people would have spent different amount of time at this stage and accumulated certain mindsets.

Improvements from DDK:

  • Better reading skills overall, most actually reads a few moves ahead somewhat accurately.

  • Better understanding of common tesuji and can utilize them to get out of certain situations.

  • Understanding basic direction of play and knows how to build a position for themselves.

  • Understands which moves are big and urgent to a degree and can apply these occasionally.

  • Understands some common shapes for corner enclosure, extension, invasion, reduction and just shapes overall in middle game. Also has an awareness for large points and urgent points.

  • Understands, to a certain degree, strength and weakness of groups and will have an awareness to defend a weak group. Knows common techniques to attack in the middle game.

  • Understands the concept of exchange and is sometimes willing to exchange one dead group for another. Will not always passively defend own territory.

  • Has a general idea of how to use influence.

  • Is willing to sacrifice a stone or two, or maybe three for the greater good…


  • Because of their over confidence in escape tesuji and their ability to escape or make life, SDK s at this range is the most likely to invade aggressively everywhere. Opponents of equal strength usually finds these invasions hard to catch and difficult to punish. Playing like this is one major reason a SDK find their rank stagnate.

  • Some SDKs are heavily focused on destroying the opponents position instead of building their own throughout the entire game. Some may find this style annoying, but dealing with it is a test of your go fundamentals. This is the second major factor of rank stagnation at this stage.

  • Sometimes will choose an inadequate shape to defend or attack, leading to future problems as the game develops.

  • Sometimes overlooks a weakness in their own position when attacking or defending.

  • Highly unreasonable (some SDK players), wants everything in the board and wants you to have nothing. This almost never works out well.

  • Overly aggressive to the point of overplay, but being able to punish takes a greater deal of effort, which is why they can sometimes get away with it especially against weaker players.

  • Often will misjudge the result of a local exchange and make unfavorable exchanges.

  • Still have little understanding of preserving aji and will play aji keshi and crude moves.

  • Can still be inflexible with decision making. Will often stubbornly hold on to what they have instead of venturing out to attack. Reluctance to sacrifice stones is one major trend.

  • Some players overplay to solely exploit mistakes of the opponent (whether due to time pressure or aji) instead of playing something that actually works.

Strong SDK (3k - 1k)
By now these players know most of the fundamentals.


  • All of the aforementioned improvements done better.

  • More flexibility in their thinking, willingness to exchange and trade.

  • Occasionally mindful of sente and gote. Will often try to defend in sente.

  • Better judgment of board position and a growing awareness of thickness and thinness of groups.

  • Usually have a fine instinct for common tesuji points and haengma vital points.

  • Basic awareness of aji, but will still play many crude or aji keshi moves.

  • Familiar with lighter play when invading deep in enemy territory. Knows to create aji with attachments etc. Has a degree of flexibility in their moves when the situation demands it.


  • Still unreasonably aggressive sometimes. Often will overcommit to an attack that may or may not work… Will often overplay to kill. Usually my games against opponents of this level ends up with a dead group because of an overaggressive attack.

  • Will still invade occasionally in questionable places. Though this will occur less than those in the last group.

  • Questionable joseki choice or direction of play. Often still thinks locally.

  • Sometimes will make questionable shapes in attack and defense.

  • Still poor evaluation of local exchanges. Will play exchanges that leads to a degree of local loss.

  • Sometimes makes questionable haengma that can be exploited in some way, leading to a totally collapsing position.

  • Sometimes plays slow moves

Low dans (1d - 3d)
The one biggest misconception I had for a long time as a DDK player is that dan level players can make magic happen out of seemingly nowhere. At the end of the day, they are still players who cannot bend the rules of the game. Instead of getting caught in a tricky spot and think a dan level player can find a magical tesuji to reverse it all, what makes dan level players generally stronger is that they can anticipate the danger and avoid it before it happens. This is a result of game experience and honed intuition.

If anything I just want to emphasise 1 thing low dans do more consistently compared to SDKs:

  • In some situations, they will invest one move to defend themselves in gote and recognise that is the largest move on the board

Everyone loves to attack and enjoy the thrill of watching the opponent squirm. But sometimes the secret to strength is the most mundane :smiley:

That being said, low dans make low dan mistakes. Much of the comparison used herein is in comparison against the high dan standard.

  • Unwillingness to sacrifice former attacking stones. Does not recognise consistently that vital attacking stones that were useful a few moves ago can be worthless and not need saving a few moves later. This results in the creation of a heavy group – a burden – for the rest of the game.

  • Though decent with local josekis, will still make questionable joseki choices.

  • Often lacks consideration of full-board position and play sequences that are equal locally but doesn’t work with rest of the stones on the board.

  • Unstable grasp of using aji.

  • Unsensitive to shape of stones and the aji in said shapes.

  • Will still make unfavourable exchanges that give opponent good shape, or weaken (or even erase) the aji in opponent’s position.

High dans (4d - 6d+)
I happen to be in a club full of 4d+ players (and I’m the weakest player there). From all the games I played against them, I can safely say the following:

  • They all have a solid grasp of Go fundamentals

  • They are good at handling weak groups to an extent where you cannot capture them (save for a blunder which still happens rarely).

  • They know almost all the common and special tesujis during local fights

  • They are familiar with vital points of attack and defense in the middle game

  • They have better endgame compared to low dans, but still questionable sequences by professional standards. Endgame is a common weakness amongst amateur players.

  • Because of subjectivity in positional judgement, one player might end up losing out in a position when they think they’re still ahead.

  • They will make an effort to count territory and analyse their position regularly and develop their strategy.

  • Much more flexible than previous group, but still not flexible enough by professional standards :slight_smile:

  • Understands thinness and thickness of a position

  • Will have the right ‘direction’ and ‘instinct’ to attack, but the exact stone location can often be better by professional standards.

Games between high dans comes down to which players are making more efficient exchanges compared to another. If one player can take advantage of the others’ shape and gain 3 points in sente, it would be a decent lead. Since high dans are usually players who played for a few years, each of them understands the game in a very different way. High dans usually argue over whether a given position was good for white or black based on their biased views on the game. There’s plenty of that at the club :^)

As for even stronger players…

A Chinese 4p once said: “A player many stones stronger than you will make you feel disgusted at your position. They will demand you to settle your stones in a way that is highly inefficient, but not settling it will be even worse. It’s a very uncomfortable feeling, but that’s what professionals do. They don’t aim to capture you, but they want their stones to work better than yours.”

I am not as qualified to comment on players around this range since I haven’t been there yet. However, from many lessons with my 6d+ go teacher, whose strength is on par with a weak professional player, I can infer the following:

  • They are able to read deep and read accurately in any given position

  • They are highly sensitive to the shape of stones and the aji in said shapes.

  • They are able to play moves that threaten to utilise aji in their zombie stones.

  • They are flexible with their position and is usually willing to sacrifice stones for a larger gain. In particular, they value thickness more than lower dans because they can make better use of them.

  • They can judge a position quite accurately and recognise which player has gained or lost from an exchange locally. Though this is still highly subjective, which is the cause for strength disparity among stronger players.

  • Wins or losses at this stage is usually determined by who has better reading and strategy.

  • Their life and death is very robust; given 2 minutes they can read most common local positions if they don’t know them by heart already.

  • Very cunning abuse of byo-yomi time controls to maximize thinking time :wink:

Some final remarks
Below are some of my opinions on the nature of Go. Personally I see some truth to these opinions, though I would encourage people to discuss it and share their thoughts, perhaps in another separate thread.

What makes the difference between players?

  1. Judgement

In my honest opinion, go is fundamentally a game about reading and judgement. Kobayashi Koichi 9p loves territory, Takemiya Masaki 9p favors influence. Rumour has it that an amateur asked both players: “What’s the weakness of the star point?”. To which Kobayashi responded: “The 3,3 invasion”, and Takemiya said: “The cover at 5,5”. Before AIs arrived in the go scene, it’s hard to debate who is right and who is wrong at the top level of play. But given a position, how you understand it plays a vital role in how you plan to play your subsequent moves. Whether you play risky because you think you’re behind, or you play solid because you think you’re ahead. At the top level, every point matters, which is why judgement is extremely important.

Of course, as amateurs, the first step is to be able to ‘judge’ a position. Then it’s a matter of judging accurately, and judging fast. Judge a position globally to see who’s ahead territorially, who has thickness, where are the thin areas in opponent’s position. Next it’s the ability to evaluate the relative gains and losses in between exchanges. Did white gain more than black after a local non-joseki exchange? Without these concepts, one would be playing moves they think were necessary when in fact they were losing lots of points. In the end they might not even know why they lost!

  1. Reading

Judgement has its foundations on reading. Without accurate and deep reading, you cannot judge accurately whether a variation is good or bad for you. And Go players, amateur and professionals alike, dedicate their life to hone these two fundamental skills. Life and death is a must for all who seeks to improve.

  1. Intuition and creativity

Everyone, given enough time and study, will understand common shapes for attack, defense, sabaki and the like. But those with creative intuition can see unconventional moves that most people won’t consider. Often times these creative moves can throw the opponent off-guard, or be a straight-up beautiful tesuji that the opponent did not consider when playing his move. Keep an open eye for moves on the board beyond the conventional shapes. Not all empty triangles are bad moves!

  1. Flexibility

All the above points lead to the final one. You don’t have to connect to every peep. Nor do you have to defend every attempt your opponent makes to come into your territory. If they break 50 points here, you break 50 points there. The game will still be even. Be willing to give up stones for thickness or for sente.


Mr. Gorbachev, tear down this text wall!

No, really, I am pleasantly amazed that you wrote that much.

… And I’ve now read it all :slight_smile:


On my way to the high SDK ranks (and hopefully dan at one point :stuck_out_tongue:) ive experienced improvement as a struggle for confidence mostly. confidence in ones own reading ability and judgement of big points is necessary in order to be able to play away at crucial moments.
In my opinion knowledge of who has sente, how to take and how to keep it is the most consistent measure of strength throughout the ranks. it distinguishes DDKs from SDKs and SDKs from Dans likewise. that is because it involves knowledge of L&D and shape points, reading skills, positional judgement and (not infrequently) requires quite precise counting of plays.

great point!