July 2, 2026
This heatmap left readers cold
Crossword Heatmap
Puzzle fans love the letter drama, but the color confusion stole the show
TLDR: A crossword fan mapped where letters usually appear in New York Times puzzles, turning hidden board patterns into a clickable visual. Commenters were less obsessed with the letters than with the map’s color choices, jokingly arguing over whether black means hot, cold, or just plain confusing.
A delightful little data project about New York Times crosswords somehow turned into a mini comment-section mystery: people were just as interested in the colors as the letters. The project maps where each letter is most likely to appear on a crossword board, revealing tiny patterns hiding in plain sight. Some letters can apparently pop up almost anywhere, some are strong starters, some love the ending, and poor oddballs like J, Q, X, and Z are basically the rare celebrities of the grid.
But the real action? Readers immediately zoomed past the letter science and into full-on "wait, how do I read this thing?" mode. One commenter openly admitted that Q, U, V, X, Y, Z looked fascinating but asked the question that became the thread’s main character: does black mean more or less? Another piled on with the blunt design roast, saying it should probably be red. In other words, the hottest debate was not about language, logic, or crossword strategy — it was about whether the heatmap was even giving off heat.
Then came the wholesome twist: dataset creator Saul Pwanson showed up like a proud parent, calling the project exactly what he was hoping for and even offering up a bigger dataset for more number-crunching chaos. So yes, the crossword map impressed people — but the comments made it clear that when you call something a heatmap, the crowd expects heat, not a grayscale identity crisis.
Key Points
- •The project analyzes New York Times crossword grids to study positional distributions of letters and empty spaces.
- •The dataset was obtained from Saul Pwanson and limited to two NYT crossword formats: 15x15 daily puzzles and 21x21 Sunday puzzles.
- •The author counted per-cell character occurrences for each grid version to build heatmaps.
- •The visualization is normalized by each character’s maximum per-cell frequency, so it shows where a character tends to appear rather than how common it is overall.
- •The article identifies letter groups with similar positional tendencies, including strong starters, strong enders, middle-heavy letters, and rare letters.