Why Some Puzzles Feel Fair

A difficult puzzle can keep you stuck for an hour and still feel impeccably fair.

Another can annoy you in thirty seconds.

The difference is rarely difficulty itself. Solvers will tolerate formidable crosswords, murderous Sudoku and puzzle boxes that seem to have been designed by a watchmaker with a grudge. What they resent is the suspicion that thinking harder will not help: that a rule has shifted, a necessary clue was withheld, an object behaved differently this time, or the intended solution depends on guessing what happened to be in the setter’s head.

Fairness is the quiet contract underneath a puzzle. You may be deceived. You may be sent down an attractive dead end. You may discover that the thing you dismissed ten minutes ago was the whole point. But when the answer arrives, the puzzle should be able to show its receipts.

Fair is not the same as easy

This distinction matters because making a puzzle easier is often the wrong cure for making it fairer.

Consider a Sudoku. A hard grid may require a long chain of deductions that few casual solvers will spot. That is difficulty. If the printed givens admit two equally valid completions while the instructions promise one solution, that is a different problem. Nothing about greater patience or skill can tell the solver which answer the publisher privately prefers.

Programmer and puzzle designer Simon Tatham makes the distinction unusually strict in the editorial policy for his long-running Portable Puzzle Collection. A game presented as fair must be solvable, and where hidden information exists, a correct move must be deducible from the information currently available. An arbitrary choice that only becomes right in retrospect fails the test.

That is a useful hard edge to the idea, even if not every puzzle tradition needs to adopt it wholesale. Some games deliberately contain risk, incomplete information or multiple solutions. The important thing is that the player knows what sort of agreement they have entered.

Chess does not promise that every move can be proved correct. A cryptic crossword does promise that, once solved, the clue can be parsed. A logic puzzle usually promises that deduction rather than clairvoyance will get you home.

Fairness begins with knowing which promise is being made.

The rules have to stay put

Puzzles teach us how to read them.

If pushing a wooden crate onto a switch opens a gate in the first five rooms of a game, the sixth room has already inherited a rule. If one particular crate suddenly cannot be pushed because the designer needs it to remain decorative, the problem is not merely that the player is confused. The game has spent five rooms teaching the confusion.

This is why consistency is more valuable than exhaustive instruction. Good puzzle systems rarely explain every consequence in advance. They establish a small grammar and let the player become fluent in it.

Once that happens, difficulty can rise dramatically. A familiar rule can be combined with another familiar rule in a way the player has never seen. The surprise feels earned because its ingredients were already on the table.

The opposite is the infamous adventure-game solution that asks you to combine two objects for a reason no ordinary person would infer. The game may technically contain everything required. That is not the same as having communicated a usable logic.

A fair puzzle can conceal an answer. It should be careful about concealing its language.

Necessary information should be available, not necessarily conspicuous

There is a lovely tension here. If every important clue flashes, pulses and announces itself as IMPORTANT CLUE, there is not much puzzle left.

Information can be subtle. It can sit in plain sight. It can even be designed to look unimportant until a later observation changes its meaning. What matters is that the solver had a reasonable opportunity to encounter it and understand its relevance from the puzzle’s established vocabulary.

This is one reason a good misdirection feels so different from a cheap trick. Misdirection encourages a plausible interpretation, then lets a better interpretation displace it. A cheap trick asks you to notice something the puzzle gave you no reason to inspect.

Crossword setters live with this distinction every day. As we argued in A Crossword Clue Is a Tiny Promise, a clue can lie beautifully on the surface while remaining exact underneath. The joke works because the deception and the accounting coexist.

Visual puzzles have the same obligation. If colour is normally decoration, one level cannot quietly make a tiny colour difference essential unless the puzzle has done something to establish colour as information. If background objects are normally inert, requiring the player to click a background brick is not ingenious merely because the brick was technically visible.

The relevant clue need not be loud. It needs to belong to the conversation.

Failure should teach you something

A fair puzzle can let you fail.

In fact, failure is often useful. You try a hypothesis, watch it break and return with a better model of the system. The failed attempt has bought information.

But there is a nasty version of failure in which the player makes a mistake, continues for twenty minutes and only then discovers that the puzzle became impossible long ago. Sometimes that delayed consequence is the puzzle. Often it is just expensive feedback.

Undo buttons, reset controls and visible state are not concessions to weak solvers. They change the economics of experimentation. When recovery is cheap, a player can test an idea without wondering whether curiosity will cost an evening.

Research work at the MIT Game Lab on procedural puzzle tools treated solvability and the prevention of unknowingly entering unsolvable states as explicit design goals. That is a technical version of a very human concern: if a system lets the player ruin a puzzle, the system should have a good reason for doing so.

This does not mean every mistake needs to be instantly reversible. A physical puzzle box may be interesting precisely because actions have consequences. A Sokoban position can be satisfying because you learn to recognise deadlocks. Fairness lies in whether those consequences can be understood and whether the puzzle has equipped you to reason about them.

The player should lose an argument with the puzzle, not discover that the puzzle left the room twenty minutes ago.

Surprise needs an alibi

The finest puzzle solutions often feel impossible a moment before they feel inevitable.

That transformation is the source of the Aha moment, but it also explains why fairness matters so much. Surprise alone is cheap. A designer can surprise a player by changing any rule at any time. The satisfying surprise is one that survives inspection.

After the reveal, you can go back.

There was the clue. There was the constraint. There was the odd wording you dismissed. There was the empty square whose shape suddenly matters. The answer has not merely appeared; it has reorganised evidence that was already present.

This retrospective quality is one of the best tests of puzzle fairness. The solver’s first interpretation can be wrong. The puzzle’s evidence should not be.

It is also why an answer can feel fair even when almost nobody finds it. Rarity is not proof of obscurity. A brilliant deduction may be hard because several ordinary facts must be held together at once. Once shown, though, each step has to bear weight.

Trust is accumulated in small deposits

A single puzzle is judged partly on its own terms. A puzzle game or newspaper section has something more powerful: a reputation.

Solve ten carefully constructed puzzles and you begin to grant the eleventh a useful benefit of the doubt. When you get stuck, you assume there is something you have not seen rather than something the designer forgot. That assumption keeps you thinking.

Break the contract often enough and the mental posture changes. Now every impasse produces administrative questions. Is this solvable? Did I misunderstand the controls? Is the clue wrong? Is there a bug? Am I expected to look something up? The puzzle has stopped being the only puzzle.

This is why fairness is not merely an ethical nicety or a checklist item in playtesting. It changes what difficulty feels like. Trust lets a designer ask more of the player.

The same principle sits behind A Good Puzzle Should Let You Leave. Respect for a player’s time does not require removing friction. It requires making the friction belong to the thing they came to do.

A hard puzzle says: you have not seen it yet.

An unfair one leaves open a more corrosive possibility: there may be nothing worth seeing.

The answer should make the puzzle look better

There is a simple test worth applying after a solution is revealed.

Does the puzzle become more impressive?

A good answer illuminates earlier details. The peculiar clue wording becomes precise. The awkward shape turns out to have a job. The rule you learned in an easy level acquires an unexpected second use. Even a deliberate red herring looks fair because you can see exactly why you followed it and exactly what should have made you turn back.

A bad answer needs excuses. You were supposed to assume… You just had to try… Technically, the clue could mean…

No formal definition will settle every case. Puzzle traditions make different bargains, and expert communities happily tolerate conventions that would bewilder a newcomer. But the durable principle is less slippery than it first appears.

A fair puzzle may withhold the answer for as long as it likes. It does not withhold the possibility of understanding why that answer is right.

Sources

• Simon Tatham, “How to write a new puzzle”, especially the collection’s fairness criterion.
• MIT Game Lab, “Procedural Puzzles as a Design Tool for Games”, on solvability and preventing unsolvable states.
• Josh Bycer, “A Study Into Puzzle Design”, Game Developer, on the informational contract between puzzle and player.