Shannon Entropy and Game Uncertainty: The Thermodynamics of Information
In 1948 Claude Shannon made 'information' a measurable quantity. His entropy formula is the mathematical reason some game choices feel deep and others feel hollow.
When Claude Shannon published A Mathematical Theory of Communication in 1948, he proved that information is a quantifiable thing. Information equals reduction in uncertainty. The more unexpected a message, the more information it carries. The same formula answers why a game is interesting.
1. The entropy formula
Shannon entropy is written H = -Σ p(x) log₂ p(x), and its unit is the bit. A coin flip carries one bit; a six-sided die, 2.58. Entropy peaks when all outcomes are equally likely — maximum uncertainty.
2. Predictable games have zero entropy
Two perfect tic-tac-toe players always draw. The outcome is known. Entropy is zero. That's why it's a kid's game — no information is being produced for adults. Chess is technically solvable but the space is so huge every move still carries entropy.
Fun lives in the balance between entropy and skill. Pure randomness (max entropy) is meaningless; pure determinism (zero entropy) is boring. The narrow band between them — that's a game.
3. The roguelike formula
Hades, Spelunky, Slay the Spire — all inject high entropy and anchor it to a rule base. Room layouts are random but balanced; enemy placements are random but reactive. The result: every run carries new information, but the information is meaningful.
4. Encounter entropy design
Good opponents oscillate between predictability and surprise. Too predictable and the boss is weak; too random and it's unfair. Dark Souls bosses are deliberately tuned to medium entropy: 4-5 attack patterns, each with a different tell, sequence semi-random.
5. Signal Pitch and entropic pressure
In Signal Pitch the enemy layout is seed-locked — same seed, same entropy. This matters for fairness: player skill becomes separable from the entropy. Two PvP players face the same challenge; the only variable is how they process it.
6. Information budget: how much entropy should a UI show?
- Too little: player is blind, frustrated.
- Too much: cognitive overload, freezing.
- Right amount: 3-7 bits per action. Miller's classic "7±2" rule ties straight into Shannon here.
7. Conclusion
Shannon entropy isn't just a formula — it's the quantitative spine of fun. A game feeling "deep" is directly proportional to the bits per second a player processes. Designers know this instinctively; information theory is the language that explains them.