Formal Specification
Senegalese Jumping and Stick-Dodging Game — Formal Specification
Senegal · dexterity-small · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 4
# Kotti — AI-optimized specification
> **Reconstructed source.** The entry carries a provenance warning: its rules were written from
> general knowledge, not read from the two works it cites. Gravrand (1990) and Thomas (1958) are
> real books, but the entry does not evidence that either describes this game.
## General idea
A stick-clapping jump game whose formal content is almost entirely **phase**. Two seated holders
beat a fixed 4-count cycle — tap-tap together, open-open apart — and a jumper's foot may occupy
the space between the poles only during the open beats. Everything else follows: the game is a
timing problem against a periodic hazard, the jumper's state is a phase offset, and failure is
being in the wrong place at a beat boundary. Unusually for a dexterity game in this corpus, the
hazard is **not stochastic in its timing at all** — the rhythm is deterministic and public, and
only the jumper's execution is uncertain.
## Formal model
- **Abstraction taken:** Jump execution, foot placement, and balance are collapsed into a
per-beat success probability. The **rhythm itself is not abstracted** — R2 specifies the 4-count
cycle exactly, so it is modelled as a deterministic clock rather than a random process.
- **Player/agent count:** 2 stick holders (non-competing) and 2-6 jumpers (entry §Players).
- **Turn model:** A global synchronous beat clock; jumpers enter in sequence and act on beats.
## Formal rules
R1. Setup (entry §1): two stick holders sit opposite each other holding the ends of **two 2-metre
wooden poles or bamboo**.
R2. Rhythm (entry §2): the holders tap a **4-count cycle** — *Tap-Tap* (poles together), then
*Open-Open* (poles apart). The cycle is fixed and repeats indefinitely
[gap-fill: the entry does not state the tempo, in beats per minute or otherwise, and tempo is
the parameter that determines the game's entire difficulty].
R3. Jumping (entry §3): jumpers **enter in sequence**, hopping in and out on one or both feet
[gap-fill: the entry does not say how many jumpers are in the poles at once, nor whether entry
is on a specific beat. One jumper at a time is assumed, as multiple simultaneous jumpers would
require a spacing rule the entry does not supply].
R4. Elimination (entry §4): a jumper whose foot is **clamped by the closing sticks**, or who trips,
is **eliminated**. Elimination is permanent within the round. Note this can only occur on the
transition into a *Tap* beat, which makes the hazard periodic and predictable rather than
random.
R5. Victory (entry §5): the **longest continuous jumping sequence** wins
[gap-fill: the entry does not say whether this means the last jumper surviving or the jumper
with the most successful beats — two different measures that can pick different winners, since
a jumper eliminated late may have completed fewer beats than one eliminated earlier at a faster
passage. Survival is assumed as the primary reading, being consistent with R4's elimination].
R6. Tempo escalation: the entry records **escalating tempo stages** with handclapping and Serer
songs under Variants rather than Rules [gap-fill: escalation is therefore treated as an
optional layer, not a core rule; but note that without it, R2's fixed tempo makes the game
stationary and a sufficiently skilled jumper would never fail].
## Edge cases
- Tie at the end: two jumpers eliminated on the same beat would tie under R5's survival reading,
and the entry addresses neither that nor the ambiguity in R5 itself. With one jumper in the poles
at a time (R3's assumption), simultaneous elimination cannot occur, which resolves the case by
assumption rather than by rule.
- A player with no legal move: not applicable — a jumper in the poles always has `hop_in`,
`hop_out`, or `hold`, and a jumper waiting to enter always has `wait` or `enter`. Holders have no
decisions at all; they are a clock, not agents.
- Repetition / cycles: the 4-count rhythm repeats indefinitely by design (R2), which is the game's
entire structure rather than a degenerate case. Without R6's escalation this repetition is
perfectly stationary, and that is exactly why the termination argument below depends on
escalation or on imperfect execution.
- Simultaneous claims on one position: the space between the poles admits **one jumper** under R3's
assumed reading, so contention is excluded by construction rather than resolved. If multiple
simultaneous jumpers were permitted, the entry supplies no spacing or collision rule.
- Elimination mid-round and turn-order repair: central to the game (R4) and permanent within the
round. Jumpers enter in a fixed sequence, so an eliminated jumper is simply removed from the
queue; the beat clock is global and unaffected, so there is no turn order to repair.
- Resource exhaustion: not applicable — nothing is consumed. The poles and the rhythm are
inexhaustible. Jumper stamina is the real limit and, as in several entries here, is unmodelled.
- Termination guarantee: **guaranteed with probability 1, conditional on imperfect execution.**
The jumper pool is finite and strictly non-increasing under R4, so the round ends when it empties.
Each beat carries a fixed positive probability of a clamp for the jumper in the poles, so the
pool drains almost surely in finite expected beats. The conditionality is worth stating plainly:
if a jumper's per-beat failure probability were zero, R2's stationary rhythm would let them
continue forever — R6's tempo escalation is what makes failure eventually certain in the real
game, and it sits in the entry's Variants rather than its Rules.
## State representation
- Persistent (round) state: `alive[j]` (monotone non-increasing under R4), `jumper_queue`,
`beat_index` (mod 4), `tempo` (per R2's gap and R6's escalation), `beats_survived[j]`.
- Per-beat ephemeral state: `pole_state` ∈ {together, apart} — a deterministic function of
`beat_index`, not an independent variable — and `foot_position[active_jumper]`.
- Observable information: fully observable, and importantly so: the rhythm is public and periodic,
so a jumper always knows exactly when the poles will close. There is no hidden information and no
uncertainty about the hazard's timing — only about one's own execution.
## Action space
- `enter()`: the next queued jumper; precondition: the poles are `apart` and no jumper is inside
(R3).
- `hop_in()` / `hop_out()`: the active jumper; the two transitions across the pole boundary, each
resolving a success check against the current tempo.
- `hold()`: the active jumper; remain in the current position through the next beat.
- Holders have **no** action space — R2 makes them a deterministic clock, and modelling them as
agents would be adding decisions the entry does not give them.
## Playability notes
- Single-player-vs-AI: reasonable as a rhythm-timing challenge with a human jumper, and near-
pointless as an adversarial problem — there is no opponent. The holders do not oppose the jumper;
they execute a fixed pattern, so the only contest is against the clock and against other jumpers'
survival lengths.
- Reducing n > 4 to 2 agents: mechanically lossless, since jumpers never interact and are compared
only at the end. What disappears is the queue and the social escalation of watching others fail.
- Physical-skill games: footwork, timing, and balance are the entire contest and collapse into a
per-beat success probability. This is one of the few entries where the abstraction is *nearly*
adequate, because the hazard's timing is genuinely deterministic and only execution is uncertain.
- What is lost: the music. The entry ties the game to Serer songs, handclapping, and initiation
dance traditions, and the rhythm is simultaneously the game's clock and its cultural content —
a `beat_index` modulo 4 keeps the former and discards the latter entirely.
## Implementation hints
Model `pole_state` as a **derived** function of `beat_index`, never as independent state: the poles
close on a schedule, and any implementation that lets them close probabilistically has converted a
timing game into a luck game. Make tempo an explicit parameter and implement R6's escalation as an
opt-in layer — with a fixed tempo and a skilled agent, rounds do not end, which is the practical
consequence of the entry putting escalation under Variants. R5's ambiguity must be resolved
visibly, since survival and total-beats can name different winners and the entry supports both
readings. The representational trap is R4's timing: a clamp is only possible on the transition
*into* a Tap beat, so an implementation that checks for clamping every tick will eliminate jumpers
at roughly four times the correct rate. Suggested first milestone: the 4-beat clock with one jumper
and a tempo-dependent hop success rate, verifying that eliminations occur only on close
transitions and that raising tempo shortens survival.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Senegalese Jumping and Stick-Dodging Game (Senegal). (n.d.). Games of Countries. https://gamesofcountries.com/games/sn-senegal/kotti-senegal
@misc{gamesofcountries-sn-senegal-kotti-senegal,
title = {Senegalese Jumping and Stick-Dodging Game (Senegal)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/sn-senegal/kotti-senegal}
}