Particle.news
Download on the App Store

Researchers Use Shannon Entropy to Solve Wordle in 99% of Simulations

An April paper shows an entropy-guided guessing rule cuts uncertainty so efficiently that simulations solved 99% of puzzles while its real-time use requires a separate script.

Overview

  • This week multiple outlets reported that a Binghamton University team published a paper in April showing an entropy-based strategy solved 99% of Wordle puzzles in computer simulations.
  • The method treats each guess as a source of information and uses Shannon entropy to pick words that maximize expected reduction in the set of possible answers.
  • In head-to-head simulations the entropy approach reached a 99% success rate compared with about 90% for common-letter or vowel-heavy starting strategies.
  • To use the strategy during live play a person must run a separate program and manually enter Wordle’s color feedback for the script to recommend the next guess.
  • Authors note the work began as a class project turned peer-reviewed paper and commentators warn that fully algorithmic play could diminish the human enjoyment of the game.