How We Verify Clues and Lexicon Sources
How the application records clue evidence, review fields, answer lengths, and current sourcing limitations.
Verification is a layered process. Mechanical checks catch letter-count and grid errors, while editorial review evaluates meaning, grammar, ambiguity, and whether the evidence label accurately describes the source. A candidate is not presented as externally verified unless its record identifies external evidence.
What each clue record contains
A solver candidate stores the proposed answer, declared letter count, definition, part of speech, clue type, evidence type, source label, and internal review status. Some entries also explain which sense of an ambiguous clue they match and when they may not fit.
These fields make limitations visible. Two answers can be valid synonyms while only one matches a particular grid length, crossing pattern, or grammatical form.
Automated integrity checks
Before a curated clue page is eligible for indexing, the application checks that at least two candidates are present, every candidate is marked as internally reviewed, every answer length matches the stored answer, the page has a meaningful explanation, the editorial status is published, and an explicit index decision and expected canonical route are recorded.
Puzzle tests separately check grid dimensions, entry coordinates, declared lengths, black-square collisions, and matching letters where Across and Down entries intersect. These checks catch structural defects; they do not replace a human judgment about clue fairness or nuance.
Meaning and grammar review
Review starts with the precise sense of the clue. The candidate must match the requested part of speech, number, tense, and level of formality. When a word has multiple meanings, the explanation should identify the relevant one rather than presenting every dictionary sense as interchangeable.
Rare, archaic, foreign-language, or specialized answers need clear clue signals. If the clue is too broad for a single confident answer, the solver should offer candidates and explain the uncertainty instead of claiming one guaranteed solution.
Evidence labels and current limitations
The current solver uses an internally maintained seed list. A record labeled “editorial” means the entry has passed the site’s internal checks; it does not imply a licensed corpus, independent dictionary partnership, or external certification. Other evidence types should be used only when the corresponding source is actually recorded.
Definitions and candidate lists can still contain mistakes or omit a valid regional, historical, or themed answer. Crossing letters and the original puzzle’s editorial context remain decisive.
Corrections and change handling
A useful correction identifies the page, clue wording, proposed answer, letter count, and the reason the existing entry is wrong or incomplete. Until a dedicated reporting service is connected, corrections can be sent through the site’s contact page.
A correction should update the candidate record, explanation, review date, and any affected tests together. That prevents the visible page and the solver data from drifting apart.
What verification does not promise
Verification does not guarantee that a suggested answer belongs in every puzzle using similar wording. Constructors use themes, abbreviations, puns, proper names, and publication-specific conventions. The tool narrows possibilities; it cannot replace the crossing pattern and context supplied by the original grid.
Key Takeaway
The site distinguishes mechanical validation, internal editorial review, and external evidence. That boundary is more useful than an unsupported claim that every candidate is universally correct.