Library / Learning / Multi-Armed Bandit
Multi-Armed Bandit
Balance pulling the best-known lever against testing new ones. Regret is the score.
INTERACTIVE LAB
Tolerance sandbox (original recreation)
happy 76% • avg. similar neighbors 49% • red ring = wants to move
LEARNING
When to use / misuse
Use → Trials, dosing, ad creative. Run small experiments, then exploit.
Difficulty → Core
REDCAPE → Act · Explore
agent: bandit-checker family: Learning combine_with: [two other families] rule: state failure mode before acting
Free sources: Book final chapter (opioid case) — original summary
Attach to a lattice — combines with
Different families, shared jobs. Click to open alongside.
DIFFUSION
SIR Contagion
Susceptible–Infected–Recovered with an R0 tipping point. Same math for flu and pop-star fever.
SHARED JOBS: ACT · EXPLORE
GROWTHCapital Share (Piketty-style)
When returns on capital outrun growth, wealth concentrates without any villain.
SHARED JOB: ACT
DIFFUSIONCellular Automata & Game of Life
Tiny local rules aggregate into gliders, blinkers and chaos. Emergence needs no boss.
SHARED JOB: EXPLORE
More in Learning
Replicator Dynamics (Fisher)
Winning strategies reproduce faster; adaptation speed rises with variation.
Diversity Prediction Theorem
Crowd error equals average error minus diversity. Diverse models beat the lone genius.
System Dynamics / Feedback Loops
Stocks, flows and reinforcing loops turn small pushes into addiction spirals or virtuous cycles.