BWB

What makes BWB different

BWB is based on a simple premise: humans do not think like robots, and robots do not think like humans.

Yet the official NYT Wordle Bot benchmarks our human efforts to solve the puzzle as if we were made of silicon, not cells. The Bot assumes that each of us has memorized the 13,000-word Wordle dictionary and can instantly calculate the portion of those words remaining after each guess. Making matters worse, the Bot actually rewards you for taking more guesses instead of fewer.

But humans don't solve that way. We make our early guesses based on likely letters, letter pairs and letter placement. Only after narrowing down the potential solutions do we switch to a Bot-style analysis in which we can think about all remaining words and how to winnow down the list.

So BWB takes a different approach. We built an army of human simulators to solve Wordle the way humans do, focused on letters, letter groups and placement. We then compare your guesses to those of that sim swarm, recalculating what each of those sims would have guessed after each of your attempts. Over the course of your game, BWB tallies up thousands of faux-human games and uses those as a benchmark against which your game is gauged. In that way, BWB scores your game based on how a human solves Wordle, not how a computer does.

And we check that against reality. The words our simulators open with are measured from millions of real games, not invented — and when we compare our sim swarm's solve rate against what actually happened to a million real players, they match to within a hundredth of a guess.