Shannon Guessing Game
Claude Shannon's actual 1951 experiment as a game: a text is hidden, and you guess it letter by letter. Every guess you need is data — the fewer attempts per letter, the more predictable the language. From your guess distribution the tool computes Shannon's upper and lower bounds on the entropy of English, the same calculation that produced his famous ~1.3 bits per letter. Play a built-in passage or paste your own hidden text and test a friend.
Runs 100% in your browser — nothing you paste leaves your device.
Read the full guide to this tool
Notes
- Shannon's insight: a perfect guesser needs exactly as many guesses as the language has entropy. 27 equally likely symbols would be 4.75 bits per letter; his human subjects landed between 0.6 and 1.3 bits.
- The bounds come straight from the 1951 paper "Prediction and Entropy of Printed English": the upper bound is the entropy of your guess-count distribution, the lower bound Σ i·(qᵢ−qᵢ₊₁)·log₂ i.
- Guess at least 40–50 letters before trusting the numbers; text is normalized to A–Z and space, exactly like Shannon's 27-symbol alphabet.
- Runs 100% in your browser — nothing you paste leaves your device.