Quick comparison
| Concept | Evidence | Applies to | What it establishes |
|---|---|---|---|
| Genbutsu | One opponent’s discard | That opponent | They discarded this tile before |
| Suji | One opponent’s row and number geometry | That opponent | Selected two-sided routes are removed |
| Kabe | Visible copies across the table | Every player | Blocks requiring the wall tile cannot exist |
The evidence differs: behavior, shape inference, and physical count. Only after adding the rules can any evidence become a legal restriction on winning.
What is genbutsu?
Genbutsu commonly means a tile previously discarded by the opponent in question. By itself it states an observable fact: that player did not want the tile at an earlier time.
Without furiten, the opponent may change waits and later wait on an earlier discard. The tile is then a behavior clue, not a guarantee.
In Riichi, furiten prevents a player from ronning a winning tile found in their own discard set. A riichi hand is generally fixed, giving genbutsu against that riichi player formal ron protection. It:
- Applies only against that player
- Does not protect against other opponents
- Does not prevent self-draw
- Still follows the current rules’ exact cases
What is suji?
Suji uses 1–4, 2–5, 3–6, 4–7, 5–8, and 6–9 to remove selected two-sided waits. It must:
- Be calculated against one opponent
- Use the legal meaning of that opponent’s discard
- Preserve inside, double-pair, single, and other exceptions
Suji adds information over an unexamined suited tile but never becomes identical to genbutsu.
What is kabe?
All four copies visible create kabe. Every concealed block needing that tile becomes impossible, removing selected sequence routes around it.
Kabe:
- Does not need the opponent to discard the related tile
- Establishes a structural limit against the whole table
- Uses the same physical count in Taiwanese, Japanese, and Cantonese games
It still removes only structures dependent on the wall tile, not single, double-pair, or unrelated waits.
Why is there no universal defensive ranking?
Against one riichi player, rules-protected genbutsu is normally stronger than a shape-only suji or kabe inference.
In Taiwanese or Cantonese games without the same furiten:
- An early discard may become a later wait.
- Suji lacks the same legal premise.
- Kabe still removes impossible blocks but does not guarantee safety.
A fixed “safe tile ranking for all Mahjong” would conceal these differences. General defense should rank evidence, not labels.
A six-step cross-rules process
- Identify possible tenpai players: keep each opponent separate.
- Identify rules protection: furiten, declarations, fixed hands, and wait changes.
- Build each opponent’s genbutsu set: behavior clue unless rules upgrade it.
- Use suji to remove two-sided routes: name the actual block removed.
- Use kabe and visible counts: four visible removes a block; three is only one-chance.
- Compare your exit cost: shanten, value, remaining defensive tiles, turn, and loss.
What about multiple opponents?
A Riichi genbutsu tile against player A can remain unexamined against player B. Kabe can remove a structural route for everyone, while each player may still use other waits.
A simple matrix is clearer than invented percentages:
| Candidate | Against A | Against B | Against C | Cost to your hand |
|---|---|---|---|---|
| A | Genbutsu or behavior clue | No information | One route removed by kabe | Breaks two-sided block |
| B | One suji route removed | Early discard | No information | Preserves shanten |
Record evidence and uncertainty rather than pretending to know exact risk.
Claims to avoid
- Genbutsu is always 100% safe in every Mahjong ruleset.
- Suji has one fixed safety percentage.
- Every tile around complete kabe is safe.
- Safety against one player protects against all three.
- Honors or terminals always outrank every suji tile.
Each ignores rules, opponents, wait types, or visible conditions.
Scope and limitations
This article layers defensive evidence; it does not calculate a particular deal-in rate. Full push-fold decisions also need estimates of opponent tenpai, likely value, and your own winning value, none of which genbutsu, suji, or kabe can answer alone.
Separating two questions with Tingpai Lab
Tingpai Lab can identify the structurally efficient discard. Defensive evidence asks whether table information justifies overriding that candidate. First understand the efficiency cost, then decide whether the defensive evidence is strong enough to pay it.
Continue reading: