Competition works when players can understand the rules that produced the result. These pages explain the scoring and opponent decisions instead of hiding them behind a difficulty label.
Explore Competitive Word Games: Fair Play and Rankings
A direct score comparison needs the same rack, clue set, timer, and validation policy. Public ratings should update only after a defined completed match. Solo practice, abandoned rooms, and local pass-and-play belong to different records. Record the rule version and match state with the outcome. Two totals are comparable only when players faced the same information and validation. A rematch after a rule change belongs to a new comparison, even with the same players.
Difficulty without theater
An AI level should state what changes, such as solve probability or crossing strategy. A delay is interface timing, not evidence of thought. Wordmap does not describe automated choices as human feelings or experience. Easy, Medium, Hard, and Expert should produce testable differences in move selection or solve probability. Re-run the same puzzle set when comparing levels; a single lucky answer cannot establish that a label behaves as intended.
A rating is contextual
The Duel rating summarizes results in one format against the available player pool. It does not measure general intelligence or complete language ability. The methodology keeps that boundary visible. A new rating can move quickly because little match evidence exists. Later changes may be smaller as more completed results accumulate. Read the rating beside match count and format rather than treating the number as a fixed personal attribute.
Local and online play solve different needs
Pass-and-play works for two people sharing one device and requires private turns. Online Duel uses room tokens and server-scored results so two browsers can compete. The site labels these modes instead of merging them into one vague multiplayer claim. For shared-device play, hide the rack or answer state while control changes hands. For online play, room state must be shared and scored consistently by the server. The privacy and integrity problems are different even when both modes have two players.
Use outcomes for review
After a match, choose one move that changed the result and compare it with another legal option. Do not infer a broad ability from one win or loss. A specific board decision provides a clearer practice target. Save the board position, legal alternatives, time remaining, and score effect for one pivotal move. That record supports a useful rematch goal, such as checking crossings before placement, without turning the losing player into the subject of the review.
Report reproducible problems
A useful fairness report names the game, mode, visible state, submitted answer, and unexpected result. That evidence can be tested without collecting private account details or accepting a general claim that a level felt unfair. A screenshot can help when it excludes private room tokens and account details. Include the exact input and expected rule, then compare it with the published method. Reproducible evidence lets a correction address the system rather than one person's impression.
