How the puzzles are built
A look at puzzle generation, original content and the practical limits of each implementation.
A starting position you can reach
Tile Shift starts in the completed arrangement and is mixed through legal moves of the empty space. That keeps the generated position reachable; it does not measure its difficulty or prove an optimal move count. Signal Grid starts with all lights off and applies legal switches. Repeating those switches clears it again, although finding a short solution is a separate challenge.
Pair Search creates two cards for each of eight symbols, then shuffles their positions. New puzzle creates a new layout; Restart returns to the layout that was dealt. No hidden global score or comparative ranking is attached to your result.
Guides explain this implementation
Our field notes are written for the rules shown beside these puzzles. They include specific positions, overlaps and decisions you can reason through. A small teaching board may be used to show an idea; it is labelled separately from the size of the playable board.
A useful explanation should describe its limits. A planning habit is not a proof of an optimal solution, and a memory technique does not guarantee a score. When an example contains an error, we aim to correct the reasoning rather than cover it with more words.
Simple delivery and honest limitations
The game logic runs in your browser. The pages and rules can be read without JavaScript; the interactive boards need it. Your current puzzle state is held in the page and is lost when you reload or leave. We do not promise that every assistive technology and browser combination behaves identically. The help page describes the available controls and how to report a problem.