Using Math To Solve Wordle

What started as a classroom assignment on the application of information theory by Assistant Professor Congyu Wu, of Binghamton University, in New York, resulted in the publication of a study in the Northeast Journal of Complex Systems. Wordle, a puzzle game featured in The New York Times, presents players with the challenge of deciphering a five-letter word within six tries. As players make guesses, the game provides feedback by identifying whether any of the guessed letters are correct and whether those letters occupy the correct positions within the hidden word. Conventional approaches solve the game about 90 percent of the time.

In contrast, the Binghamton team used Shannon entropy, a mathematical concept that measures uncertainty, to prioritize information gathering over direct pursuit of the solution. The objective shifts from achieving a probability of being right to maximizing the expected reduction in uncertainty. The team created a program that recommended the next word most likely to yield the most useful information for solving Wordle, resulting in a reduction in the number of guesses and a success rate of 99 percent.