which wordlist download should i use?

we provide a choice of either lowercase, uppercase, or surface forms (cased/spaced/punctuated for answers scored 30+). they've been tested on the latest beta version of ingrid to ensure they're equivalent for purposes of filling. each is available in both .txt and .dict.

what do the scoring levels mean?

scores of 50-55 mean “clean” (well-attested and/or valid), while scores of 40-45 mean “decent” (well-attested but not ideal, less-attested but likely ok). everything lower is dubious. answers that aren't proper/abbrev/partial get a 5 point boost.

how do you determine each answer's score?

the process is mostly automated based on historical appearances, clue usage, filters, and external resources. short answers are held to a stricter standard than long ones. let us know if anything seems off.

why are some answers scored zero?

we screen words against a hand-built “blocklist” and set them to zero for a variety of reasons (harmful, X-rated, etc.). you’re welcome to rescore them.

where do you get your answer data?

we get historical answer and clue data from saul pwanson’s database and more recent puzzle data via parker higgins’s downloader. we also use some external resources with terms that would make for good answers but don't appear historically.

can i make changes to your wordlist?

we highly recommend making your changes in a separate, higher-priority list so you can continually replace spread the word(list) with its latest update without losing your edits.

what is the puzzle visualizer?

it's a review tool that takes a .puz file and shows you every answer in the grid scored against our list, alongside appearance counts across 26 venues since 1994. you can highlight answers in the grid according to score, appearance counts, and labels. none of your data is uploaded.

what is the answers browser?

it's a search tool where you type in an answer and see its wordlist score and labels, alongside appearance counts across 26 venues since 1994, as well as its most recent clue in each venue that has run it. none of your data is uploaded.

what are the labels and surface forms?

our tools show proper/abbrev/partial labels and surface forms (cased/spaced/punctuated) for answers scored 30+. these are data-driven from clue usage; when an answer has multiple senses, we follow the recently predominant one. the chosen sense might not always match the "latest clue" or the clue in your .puz file.

was AI used to build these tools? why?

enrique had built versions of the visualizer and browser by hand over the years; to bring them to the community, he used Claude for web development and to help derive clean clue-driven labels and surface forms at this scale. as ever, our goal is to make free data-informed tools that help creators bring their crossword ideas to life, not to replace craft or creativity.

how do i make a crossword?

here’s a helpful list of construction resources.

where can i submit crosswords?

here’s a helpful list of publication specs.

where can i talk to people about crosswords?

try the crosscord discord server or, if you’re underrepresented, the crossword puzzle collaboration directory facebook group.