RoyalDoku

The game

A hidden court of chess pieces sits on the board: exactly one piece per row, column and colour region, and — because they are chess pieces — no piece may attack another. You rule squares out, pencil in suspects, and reveal the court by pure logic.

Guess-free is not a slogan, it is a construction constraint. Every deal is generated offline and machine-verified to have exactly one solution — then checked again by a logical solver that must reach that solution using deduction steps a human can make. If a board would ever reduce you to guessing, it does not ship.

How it plays

1. Tap a square to rule it out; double-tap to reveal when you are certain. A wrong reveal costs a heart and a time penalty — certainty is the currency.
2. Pencil marks note which piece types could stand on a square — five types, five fixed positions, so a glance reads the possibilities. Flags watch squares you are reasoning about.
3. The census under the board counts what remains of each piece type — a real given that real deductions hang on.
4. Undo steps back marks only — it never un-reveals, never refunds. Knowledge is permanent.

Two rulesets — two ways of thinking

Classic is Sudoku law plus chess law. One piece per row, one per column, one per region — the lines do the heavy lifting, exactly as they do in Sudoku: a placed piece instantly sweeps its whole row and column, and you scan lines the way a Sudoku player scans boxes. The chess layer refines it — no piece may attack another — and the last few deductions are usually attack constraints closing what line logic alone cannot.

Advanced deletes the lines as law. Only two rules survive: one piece per region, and no piece attacks another. Nothing is universally forbidden any more — two knights may happily share a row — but nothing shares a line with a rook, because the rook attacks along it. The grid law you leaned on returns only where a piece’s own reach re-imposes it: a rook privately restores Sudoku on its lines, while a knight ignores lines entirely and forbids an L-shaped constellation instead. Constraint stops being a property of the grid and becomes a property of each piece.

Same board, same piece, opposite reasoning. Classic is bookkeeping: track the lines, refine with attacks. Advanced is coexistence geometry: every deduction flows from where power projects, so you think in shapes rather than rows — and the piece census becomes the spine of the solve. Eight levels ladder from kings-only up to a full mixed court, and the jump between rulesets is the game teaching you its second language.

The daily

One global board a day — every player in the world gets the same deal, selected deterministically from the date, on a weekly rhythm: gentle Monday, advanced weekend. No account, no server round-trip, no way for two players to see different boards.

The ruleset is data

Under RoyalDoku sits a general deduction engine: a piece is a name, a glyph and an attack footprint, and the solver cascade — eliminations, census counting, uniqueness verification — never asks which game the footprint came from. Classic and Advanced are already two rulesets running on one engine. A set with entirely different movement — fantasy creatures, card suits, anything with a definable footprint — is a data file, not a rewrite, and the no-guess pipeline (generate, count solutions, replay the human-legal deductions) verifies it the same way.

What ships today is chess. What was built is a deduction platform.

Worth judging closely

• The no-guess guarantee is auditable: deals come from an offline generator that counts solutions and replays the actual rules — not from a difficulty heuristic.
• Castle walls view: an optional 3D-walled board where the masonry itself is the region map — the information load drops, and the board becomes a place.
• A deduction vocabulary, not just an eraser: eliminations, flags, per-piece marks and soft what-if marks each claim something different.