SNAKE
DECISION LAB

Watch the AI score every legal move. The dotted line is its current plan; the full-board map reveals food distance, escape room, tail access, and when occupied tiles will reopen.

CURRENT STRATEGY

Balanced safety

DEFAULT

The AI combines food distance, flood-filled open space, and whether it can still reach its moving tail. This is a compact example of a weighted evaluation function.

score = -3F × distance + S × open space + tail safety

ONE-SPAWN LOOKAHEAD

Next-food trap rate

0.0%

0 of 318 possible next food locations have no route with a tail-safe continuation after the current plan reaches its food.

Computed once when this food appears and held steady until the next food. It is a one-spawn heuristic, not a proof that the entire future game is solvable. Open the “Next-food trap risk” map to see the locations.

LIVE TRACE

Next move evaluation

8 TILE PLAN · 0.0% KILL

INSPECTED TILE8, 9

OPEN

FOOD ROUTE87-MOVE SPACE107OPEN EXITS3TAIL ACCESSYESBODY CLEARS5 movesAI UTILITY77.0NEXT SPAWNSAFESPAWN CHECKTAIL-SAFE ROUTE

This tile is open. The utility value applies the current balanced safety weights to its food distance, local escape space, and tail connection. As the next food spawn after the current plan, it is classified safe: tail-safe route.

MOVEFOODSPACETAILSCORE

UP7320YES296.0

LEFTBODY———

DOWN9320YES284.0

RIGHT7320YES296.3

The AI simulates one move, flood-fills the remaining board, checks whether its tail is still reachable, and then applies the selected scoring rule.

Tune the experiment

Strategy

RANDOM SEQUENCE0x00051a7e

Reset behavior

Run seed

Share a seed to replay the same random-number sequence and compare AI settings fairly.

Simulation speed5×Food-distance weight2Open-space weight1Reward paths that can still reach the tailShow full-board analysis heatmap