Puzzle quality

How We Make and Test Puzzles

A good puzzle should feel fair. This page explains how we build, test and review the puzzles, solvers, printable sheets and strategy games on Logic Puzzles Online.

Solvability Checked before publishing

For puzzle types that need a single answer, candidate grids are checked by solver logic before they are shown to players.

Fairness Logic before guessing

Our aim is that a player can make progress by applying the rules of the puzzle, not by guessing between several equally valid endings.

Review Technical and player checks

Brian focuses on generation, solving and browser behaviour. Karan reviews clarity, feel, wording and whether a page makes sense to a real player.

What we aim for

Logic Puzzles Online is built around playable puzzle pages, solver tools, printable sheets and strategy games that work in the browser without an account. The details vary by puzzle type, but the standard is the same: the page should be clear, stable and useful to someone who has come to play or learn.

For single-solution logic puzzles, fairness usually means one valid answer, clues that are consistent with the rules, and a route to the solution that does not depend on luck. For solver tools, fairness means reporting invalid, incomplete, multiple-solution or uniquely solved positions clearly. For strategy games, fairness means legal moves, reliable state handling and controls that do what a player expects.

  • Rules should be visible and understandable.
  • Generated puzzles should match the rules they claim to use.
  • Single-solution puzzles should be checked for uniqueness.
  • Hints and solution buttons should help without hiding how the puzzle works.
  • Pages should remain usable on phones, tablets and desktop screens.

How puzzles are generated and checked

Most generated puzzle pages start with a candidate grid or board. The generator builds a puzzle for the selected size and difficulty, then a checking step tests whether the puzzle is valid. For puzzle types where uniqueness matters, the checker counts solutions and rejects puzzles that have no solution or more than one solution.

The exact generator is different for Sudoku, Kakuro, Hitori, Hashi, Akari, Nonogram and the other puzzle types, because each puzzle has its own rules. The common idea is to separate creation from checking: the page should not simply trust a generated board because it looks plausible.

Some pages use a prepared bank of base puzzles and then add variety through safe transformations such as rotation or reflection. Others generate a fresh puzzle in the browser. In both cases, the puzzle should still respect the published rules and the difficulty selected by the player.

1

Build a candidate

The generator creates a board, clue set or puzzle layout for a chosen size and difficulty.

2

Check the rules

A solver or validator tests whether the board obeys the puzzle rules and can be completed.

3

Reject bad cases

Invalid, unsolved or multiple-solution puzzles are rejected for puzzle types that require one answer.

4

Show the puzzle

Only accepted puzzles are shown to players, with controls for checking, hints, solving or starting again where appropriate.

How difficulty labels are assigned

Difficulty labels are practical guides rather than mathematical promises. A puzzle can feel easier or harder depending on the solver, the device and the puzzle type. Still, the labels should be meaningful.

We use a mix of technical checks and play review. Depending on the puzzle, difficulty may be influenced by grid size, clue density, how constrained the early moves are, how much branching the solver sees, and which rule interactions are likely to appear.

  • Easy puzzles should have clear starting points.
  • Medium puzzles should require several linked deductions.
  • Hard puzzles can demand more scanning, delayed consequences or tighter constraints.
  • A larger grid is not automatically harder, but it often increases the amount a player must track.
  • If a difficulty label feels wrong after testing or player feedback, we can adjust the generator or the copy.

What Brian and Karan review

Brian Hamilton works mainly on the technical side of the site: puzzle generators, solvers, board behaviour, page performance, printable tools and the browser features that make the puzzles work across devices.

Karan Hamilton brings the player and family-game perspective. She reviews puzzle ideas, wording, instructions, page flow, printable usefulness and whether a game feels clear and fair enough to return to.

That split matters because a puzzle can be technically valid and still feel confusing. We want both: sound rules underneath and a page that explains itself well enough for a player to enjoy.

How to report a puzzle issue

If you find a puzzle that looks wrong, a solver result that seems off, a typo in a rule explanation or a control that behaves badly, please tell us through the contact page. The most useful reports include the page URL, puzzle type, selected size and difficulty, what happened, and what you expected to happen.

After a report, we try to reproduce the issue. If the problem is technical, Brian checks the generator, solver or page code. If the problem is wording, clarity or a confusing experience, Karan helps review the page from the player's side. When a correction is needed, we update the page, generator, solver, puzzle bank or explanation and republish the site.

Report a puzzle issue