Formal Specification
Iroha Proverb Cards — Formal Specification
Japan · card · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: directGap-fills: 3
# Iroha Karuta — AI-optimized specification
## General idea
A real-time auditory recognition and visual search card-snatching game based on the 48-syllable *Iroha* syllabary. A reader chants classical proverbs aloud from reading cards (*yomi-fuda*), while competing players race to spot and grab the matching illustrated card (*e-fuda*) displaying the initial syllable character; grabbing an incorrect card incurs a fault penalty.
## Formal model
- **Abstraction taken:** Continuous physical auditory listening, phrase parsing, and manual slapping speed are discretized into an agent reaction-time model per recited card: `t_response = t_phoneme_recognition + t_visual_search + epsilon`. The player with the lowest valid latency claims the card.
- **Player/agent count:** 2 to 8 players (standard competitive form: 2 competing agents plus 1 automated card-reading module).
- **Turn model:** Sequential card recitation rounds; simultaneous latency-based grab resolution within each round.
## Formal rules
R1. Deck structure: 48 reading cards (`Y_1..Y_48`, each containing a proverb) and 48 grabbing cards (`E_1..E_48`, each displaying the initial kana syllable and illustration).
R2. Setup: All 48 grabbing cards are placed face-up in the central field.
R3. Recitation sequence: On each round, the automated reader draws 1 unread reading card `Y_k` at random without replacement and broadcasts its text string over time.
R4. Grabbing action: Each player agent commits an action `grab(target_card_id, timestamp)`.
R5. Capture resolution:
- If Player A commits the lowest valid timestamp on the matching card `E_k`, Player A captures `E_k` into their personal score pile.
R6. Fault (*Otetsuki*):
- If a player touches an incorrect card `E_j` (where `j != k`), that player commits an *otetsuki*.
- Penalty: The faulted player must return 1 previously captured card from their score pile back to the central field.
R7. Match end: The match terminates when all 48 cards have been read and successfully captured.
R8. Victory: The player with the highest count of captured cards in their score pile wins the game.
R9. [gap-fill] Penalty debt: If a player commits an *otetsuki* with 0 cards in their score pile, they incur a -1 penalty debt applied to their next captured card.
R10. [gap-fill] Exact latency tie resolution: If two players submit identical timestamps (`|t_A - t_B| < 1ms`) on the correct card, the card is awarded via an instantaneous 50/50 pseudo-random draw (*Janken* proxy).
R11. [gap-fill] Forfeited card recycling: Any card returned to the field due to an *otetsuki* is placed into a reserve pool and recited in a supplemental round after the main 48-card deck is exhausted.
## Edge cases
- **Tie at the end condition:** If both players hold exactly 24 cards at game end, the match is declared a draw.
- **A player with no legal move:** Not applicable; an agent may choose to grab any card in the field or choose to pass on a round.
- **Repetition / cycles:** Impossible; deck size is strictly 48 cards, and recycled penalty cards strictly terminate upon clean capture.
- **Simultaneous claims on the same resource or position:** Resolved strictly by response timestamp comparison (R5) or 50/50 draw on exact ties (R10).
- **Player elimination mid-round:** Not applicable (all players remain active for all 48 cards).
- **Resource exhaustion:** Exactly 48 cards are distributed; the game terminates upon field depletion.
- **Termination guarantee:** Guaranteed by finite 48-card deck with non-replacement drawing.
## State representation
- **Persistent state:** `field_cards` (set of active card IDs in the field), `player_scores` (array of ints / card counts), `reading_deck` (shuffled queue of unread card IDs), `recycled_pool` (list of card IDs).
- **Per-round state:** `current_reading_card` (card ID), `syllable_reveal_time` (float), `submitted_grabs` (list of tuples: `(player_id, target_card_id, timestamp)`).
- **Observable information:** Common knowledge: all field card positions and revealed reading text; hidden information: unread cards remaining in the deck.
## Action space
- `grab(target_card_id: int, latency_offset: float)`: Submit grab attempt on target field card.
- `pass()`: Refrain from grabbing during the current recitation.
## Playability notes
- **Single-player-vs-AI:** Highly natural; AI latency distribution can be calibrated to different human skill tiers (novice to master).
- **Group games (n>4):** Scales cleanly to 3–8 players around a common table with multi-agent latency comparison.
- **Physical-skill games:** Physical hand acceleration and visual scanning are abstracted into latency parameters.
- **What is lost in the adaptation:** Physical card sweeping collisions, finger collisions on tatami, and vocal cadence variations of the human reader.
## Implementation hints
Discrete card matching engine with a continuous timestamp comparator. 48 discrete IDs. First milestone: implement 48-card recitation loop with deterministic grab latency comparator and Otetsuki penalty logic.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Iroha Proverb Cards (Japan). (n.d.). Games of Countries. https://gamesofcountries.com/games/jp-japan/iroha-karuta
@misc{gamesofcountries-jp-japan-iroha-karuta,
title = {Iroha Proverb Cards (Japan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/jp-japan/iroha-karuta}
}