The Ladder — Six Rungs 未成一簣
Each rung has one job, a short work list, and a hard exit criterion. Do not move up until the criterion is met, regardless of what your rating says. That is 弗措也 — the stopping rule from the Five Gates. Rating is a lagging, noisy indicator; the criterion is a direct one.
Why one job per rung
Because working on five things means working on none. The most common failure pattern in adult improvement is a training plan broad enough to feel responsible and thin enough to do nothing. Each rung below deliberately ignores things that matter, because they matter less than the one job, and attention is the scarce resource. The six Train pages are the reference for the job your rung names — not a reading list to work through.
Rung 1 · under 900 · Stop Hanging Pieces
The one job: install a safety check before every move, until it runs without conscious effort.
Three questions. Every move. No exceptions.
Do this on every move for a month. It will feel absurdly slow and you will lose on time occasionally. That is why the plan specifies 15|10 as a minimum. The discomfort is a desirable difficulty in Bjork's precise sense — it is the mechanism.
Also work on
- Counting — before any exchange, count attackers and defenders and run the sequence cheapest first. Most material loss at this level is a counting failure, not a missed tactic. The method, a worked example and the five ways a correct count still loses material are on Calculation.
- Mate patterns only: back rank, smothered, Q+K, R+K, two rooks. Ladder mate and K+Q vs K until automatic — otherwise you will draw won games. Drills on Endgames, Tier 0.
- Puzzles: mate-in-1 and mate-in-2, hanging piece, fork, pin — Tier 1 of the motif curriculum, and nothing beyond it.
- Watch one structured video series once — then stop watching chess videos for six months.
- Build a repertoire once (here), about 90 minutes, then leave it alone.
Why not "just do more tactics"
Because in the only study anyone has run on it, puzzle rating runs ahead of game rating at every band measured — most at the bottom — and the volume of puzzles solved correlates −0.02 with closing that gap. Beginners already recognise more tactics in puzzle form than they apply in games. The bottleneck is not pattern inventory; it is application under game conditions. The full table, and the objection that partly undermines it, are on Tactics.
Exit criterion
Ten consecutive games at 15|10 or slower without losing a piece to a one-move tactic. Not ten out of twelve. Ten in a row. Track it in the log.
Rung 2 · 900–1149 · See Their Threats
The one job: shift from finding your tactics to seeing theirs.
These are different skills, and only one of them is what a puzzle trainer serves by default. Standard puzzles train you to find your own tactics; blunders come from missing the opponent's.
Work on
- Defensive review. Before each of your moves in post-game review, ask what they threatened that you did not see. Log it as its own error class. Tactics lists the few training resources that actually serve this, because standard puzzles do not.
- Basic king-and-pawn endgames: the square of the pawn, key squares, the opposition, rook-pawn draws. That is Tier 1 of Endgames — four rules, about ninety minutes, and free drill trainers cover all of them.
- Interleave your puzzles now. Turn off themed sets. You have learned the themes; the next skill is knowing which one applies — Principle Ⅵ, gate four.
- First over-the-board games. Playing across a real board is a measurably different activity. See Play OTB.
Why endgames now, and not "later"
This is earlier than the usual advice, and it is genuinely contested: the phase-by-phase blunder data puts the endgame at the top at every level, while a separate count finds only 14% of amateur classical games reach a technical endgame at all. Endgames lays out both sides and the real problems with both numbers. The reason it does not matter much here is that Tier 1 is small enough to be worth it even on the sceptic's figures.
Exit criterion
You can win K+P vs K from either side without thinking, and in your last twenty games you can name the opponent's threat you missed in every single loss. If you cannot name it, you did not review properly.
Rung 3 · 1150–1449 · Play With a Plan
The one job: stop making moves that are merely safe, and start making moves that are for something.
Work on
- Pawn structure as the plan generator. Isolated pawn, backward pawn, pawn majority, open and half-open files. The structure tells you where to play — you do not have to invent a plan from nothing. This is the whole of Strategy, and this rung is where it starts to pay.
- Converting winning positions. Most players at this level reach won positions and lose them. Trade pieces when ahead, not pawns. Simplify toward endings. Stop looking for brilliance.
- Rook endgames — Lucena and Philidor. These two positions are worth more rating points than any opening you will ever learn. Both are described move by move in Endgames, Tier 2.
- One annotated-games book, read with the moves covered, guessing each one before you look. Retrieval, not reading — the four-step method is on Strategy.
- First rated tournament. See Play OTB.
- Consider a coach, once a month. By now you have enough games for someone to find the pattern you cannot see in yourself.
Exit criterion
You can state, in one sentence, what your plan is in any middlegame position from your own games — and win Lucena and Philidor cold.
Rung 4 · 1450–1749 · Calculate Properly
The one job: systematic calculation, and thinking about what the opponent wants before what you want.
Work on
- Candidate moves. List two or three before calculating any of them. Most players at this level calculate the first move they thought of, deeply, and never look at the second. Calculation covers how the coaches disagree about this, and what survives the disagreement.
- Prophylaxis. "What does he want to do?" asked before "what do I want to do?" In our judgement this single question is most of the gap between 1500 and 1800 — a coaching claim with nothing measured behind it. The drill is one position a game.
- Visualisation. Calculate four to six ply without moving pieces. Blindfold and "pieceless" tactics trainers exist free; Calculation is honest about how thin the evidence for them is.
- Repertoire depth — reactively only. Now you may extend a line, but only one you have actually lost in, twice.
- Classical time controls. 45|15 minimum, 90|30 ideal. Structured programmes built around classical play start earning their price here.
Exit criterion
You routinely write down two or three candidate moves in critical positions, and your post-game notes contain more sentences about the opponent's plan than about your own.
Rung 5 · 1750–1999 · Technique and Depth
The one job: stop losing games to identifiable errors. At this level the remaining points come from precision, not from new knowledge.
Work on
- Endgame technique in earnest. Not just Lucena and Philidor — rook activity, minor-piece endings, opposite-coloured bishops, the practical drawing and winning methods. That is Tier 3 of Endgames, which is judgements rather than positions.
- Converting small advantages. A pawn, a better structure, the bishop pair. This is where games at this level are decided, and where Strategy's trade rules earn their keep.
- Real repertoire work. The first point on this site where openings deserve meaningful time — you will now meet opponents who have actually prepared.
- Studying complete master games in your own structures, not highlight tactics.
- A coach, if you have not already. Beyond here, self-diagnosis has diminishing returns.
Exit criterion
You convert won endgames reliably, and you lose games to being outplayed rather than to a specific error you can name. When you genuinely cannot identify what you did wrong, you have arrived.
Rung 6 · 2000+ · The Honest Fork
At 2000 you are roughly USCF Class A or above, top ~1% of rated online accounts, and stronger than essentially everyone you will meet socially. The expected gain in this band is around +37 points a year at seven hours a week, and less above it. That number is where most improvement stories end, and it deserves to be faced deliberately rather than discovered by disappointment.
Three legitimate options
- Buy hours. Six to ten a week and a regular coach. This is the only route that reliably works, and it is a lifestyle decision rather than a resolution.
- Switch the goal from rating to competition. Play classical over the board, chase a FIDE rating, enter open tournaments. More satisfying than a number, and it tends to raise the number anyway.
- Stop climbing and enjoy it. Entirely respectable. Nearly all of the pleasure of chess is available at 1800 and does not increase at 2000. Mindset makes the same argument from the other end, including the case for stopping altogether.
Worth knowing before you choose: in a screen of ~8,000 rated players, roughly one to eight per year reached national master in their thirties, zero to three per year in their forties, and there were no cases of adults making large leaps into IM or GM territory — everyone who reached 2500+ was already 2200+ by twenty-one. The 1800–2000 range is genuinely open to adult improvers. What lies above it largely is not, and no amount of method changes that.
A note on rating conversion
The bands above are given in chess.com rapid, because that is what most improvers measure themselves in. Rough equivalences from a sample of ~20,000 profiles: chess.com rapid 1420 ≈ USCF 1415 ≈ FIDE 1670 ≈ Lichess rapid 1740. chess.com rapid 1870 ≈ USCF 1775 ≈ FIDE 1865 ≈ Lichess rapid 2065. Lichess ratings run substantially higher than chess.com's at the same strength — this is a calibration difference, not an achievement.