Methodology
LexSweep publishes one fresh symmetric word-square puzzle every UTC day. Every puzzle is generated by program, validated for unique-solvability, and proven solvable by deduction before it goes live.
The format
A symmetric word square is a grid of letters that reads identically when transposed across its main diagonal. A 5×5 LexSweep puzzle contains five distinct 5-letter words; each word appears once as a row and once as a column. Solving a single row is equivalent to solving a single column (because they share letters); identifying the diagonal is structurally the most valuable early-guess outcome.
The format predates crosswords by centuries — the Sator Square (first-century Rome) is the earliest known example — but it has never had a daily-rotation digital home. LexSweep built one to fill the gap.
Generation
The daily generator runs at 00:00 UTC, seeded deterministically from the calendar date. It selects a top-left anchor word from the valid 5-letter TWL set, derives the row-and-column constraints implied by that anchor, and exhaustively searches the candidate space for grids that satisfy all constraints AND are symmetric AND have exactly one valid solution given the published clue pattern.
The candidate pool is filtered by difficulty: candidate grids whose pilot solve metrics (median solve time, median guess count, solve rate) fall inside the target difficulty band for the day of the week are eligible for publication. Mondays start easier; weekends run harder.
Validation
Before publication, the in-house solver attempts to derive the puzzle's solution using only logical deduction from the published clue pattern. The solver does not guess. A puzzle the solver cannot resolve through deduction alone gets rejected; a puzzle that admits multiple deductive paths to different valid solutions also gets rejected. The published puzzle is one we have proven to be uniquely deductively solvable.
We also re-validate every puzzle against the published TWL word list at publication time. This catches a class of bugs where the generator and the published dictionary drift apart between releases (an accepted word in the generator that is not in the player-facing dictionary, or vice versa).
Dictionary scope
LexSweep uses the TWL (Tournament Word List) — the standard North American Scrabble tournament dictionary — as the accepted word set. We filter it against a community-maintained block-list to exclude offensive terms and we exclude proper nouns and abbreviations. The TWL contains about 178,000 words, including some technical, archaic, or regional terms rarely used in everyday writing.
We do not modify the dictionary in response to individual player feedback that a word feels obscure. The TWL is a published standard; using it as-published makes the puzzle reproducible and the rules stable across the life of the site. Where the puzzle contains a word likely to be unfamiliar, we publish a definition link beside the share string so the encounter reads as educational rather than gatekeeping.
Limits of what we publish about the methodology
We deliberately do not publish the generator's seed-to-puzzle mapping ahead of time, the exact difficulty thresholds, or the pilot-solver pool composition. Doing so would let players (or aggregators) precompute upcoming puzzles in ways that would degrade the daily-puzzle experience for everyone.
We do publish: the dictionary in use, the validator's correctness checks, the symmetric-square specification, the difficulty calibration approach (without the numeric thresholds), and the open-source generator code. Anyone who wants to verify a published puzzle's correctness or build a similar daily puzzle can do so from the published methodology and code.
Frequently asked questions
What is a symmetric word square?
A symmetric word square is a grid of letters where each row reads as a valid word AND each column reads as the same word — the grid is identical when transposed across its diagonal. A 5×5 symmetric square has only five distinct words but produces 10 word-reading lines (5 rows + 5 columns) all of which must be valid.
Which dictionary does LexSweep use?
LexSweep uses the TWL (Tournament Word List) North American Scrabble dictionary as the accepted word set, filtered to exclude offensive terms and proper nouns. Both the generator and the validator share the same word list, so a word the puzzle accepts is exactly a word the validator confirms.
How is the daily puzzle generated?
The generator selects a top-left word from the valid 5-letter set, derives the constraints that crossing words must satisfy, and exhaustively searches the constraint space for symmetric grids that satisfy them. Candidate grids are filtered by uniqueness (only one possible solution given the published clue pattern) and difficulty (target solve metrics from pilot data).
Why is LexSweep harder than Wordle?
A Wordle guess produces 5 letter-position feedback signals against one target word. A LexSweep guess produces feedback signals against five words simultaneously (each row), interleaved with the column constraints. Each guess carries more information but the constraint surface is larger; the puzzle is harder per-guess but more deterministic per-clue.
Are non-English players at a disadvantage?
Yes — the dictionary is American English. We have considered adding non-English variants but have not yet decided whether to fragment the daily-puzzle pool across languages. The site assumes English-language familiarity at roughly a high-school reading level.
Can I dispute a word the puzzle accepts?
Yes, write to hello@lexsweep.com. The Scrabble TWL contains some words rarely used in everyday English (technical, archaic, regional). We do not remove TWL words from the puzzle pool on the grounds of obscurity (the whole dictionary is the standard), but we publish a definition link beside any unusual TWL word so the puzzle reads as educational rather than gatekeeping.