🧟 Tilt Contagion — Zombie SIR at the Poker Table

Calm players catch tilt from tilted neighbors at rate λ per tilted neighbor. But a tilted player also gets pressured to bust out of their seat at a rate equal to how many calm neighbors are glaring at them — the more wary, level-headed players around you, the faster you're pushed out. Deal a table, watch it unfold live, and see how the probability the whole table goes up in flames depends on λ.
Grounded in "The Zombie Infection Model" — arXiv:2607.29409. The paper studies a stochastic SIR process on graphs, originally framed as a zombie outbreak: susceptible nodes get infected at rate λ times their infected-neighbor count (standard SIR), but infected nodes get removed at a rate equal to their susceptible-neighbor count, not a flat recovery rate. That twist means dense, wary neighborhoods actively fight the outbreak back harder — and the paper proves the probability the infection survives forever is monotone in λ for the graph families it studies. This demo reskins S / I / R as Calm / Tilted / Busted at a poker table: run a single live game with a Gillespie exact-event simulator, or sweep λ across hundreds of trials to trace that survival-probability curve yourself.
Higher λ = calm players catch tilt faster per tilted neighbor.
Immediate seat neighbors only — degree 2.
avg degree —
Scales how fast simulated events play out on screen (doesn't change the math).

Live table Ring Table

Calm Tilted Busted / left table
0
Calm
0
Tilted
0
Busted
0.00
Sim time
Deal a table to begin.

Probability of runaway tilt P(every calm player gets pulled in)

Computing sweep…
—
What "runaway" means here. On a finite table the outbreak always halts eventually — but it can halt in two very different ways. Either it dies out with calm players still left, or it consumes every calm player, at which point busting also stops (a tilted player only gets pressured out by calm neighbors — once there are none left, nobody can bust anyone, and the survivors are frozen tilted forever). That second outcome — calm count hits zero — is the finite-table analogue of the paper's "infection survives forever," and its probability rises monotonically in λ, exactly as proven for these graph families.