Isidor SelectionCriteria before scores

Puzzle & Logic

Testing Solvability Without Publishing the Solution

A solvability criterion is only meaningful if it can be assessed from what the game itself tells the player. Here is how to check that without becoming a walkthrough site.

Board: Isidor Selection Editorial Team·4 min read·Published 2026-11-02

“Solvable” sounds like a binary property. In practice it is a claim about a relationship: between the information a game gives and the deductions it requires. That makes it assessable without ever writing down an answer — which matters, because the moment a selection desk publishes solutions it stops being a judge and becomes a walkthrough site.

Solvability is not fairness

Wikipedia’s article on puzzle video games frames the category around problem-solving as the core activity rather than reflexes or reaction time. That framing is important because it defines what solvability is a property of: the puzzle as a reasoning problem, not the experience of playing it.

Fairness is the neighbouring claim, and it is weaker. A puzzle can be fair — no trick, no hidden rule — and still be unsolvable from the game alone, if the game never supplies the one fact the deduction needs.

Three solvability questions

A record can answer these without leaking anything:

  1. Is the goal stated in the game? Not in a store page, not in a community guide. In the game.
  2. Is every rule the solution depends on discoverable in the game? If a mechanic is taught in a tutorial, it counts. If it is learned by reading patch notes, it does not.
  3. Is the deduction chain unbroken? Each step should follow from something already established. If a step requires outside knowledge — a real-world date, a language the game does not offer, a cultural reference — the solvability claim has a qualification attached to it.

Note what none of those questions ask: what the solution is. A desk can state “every rule required is taught in the first two tutorial screens and restated in the in-game help” and give a reader a genuinely useful answer without spoiling a single puzzle.

The hint system is evidence

The Game Accessibility Guidelines’ cognitive section lists several items that bear directly on solvability and are all checkable from a menu or the first minutes of play:

  • Provide contextual in-game help, guidance and tips.
  • Include a means of practising without failure, such as a practice level or sandbox mode.
  • Include interactive tutorials.
  • Allow players to progress through text prompts at their own pace.

Each of those is a solvability asset. A game with an in-game help system that restates its core rules has a defensible solvability claim, because the information exists inside the product. A game whose only tutorial is a splash screen that cannot be revisited does not, even if the information technically appeared once.

When information is carried by images

A category of solvability failures hides in presentation. If the solution depends on telling two shapes apart, and the only cue is colour, then the game is formally solvable and practically unsolvable for a colourblind player.

The W3C’s Success Criterion 1.1.1 is relevant here: all non-text content that is presented to the user must have a text alternative serving the equivalent purpose, except for listed situations, and controls must have names describing their purpose. A puzzle that requires reading an image as if it were text is exactly the case the criterion is aimed at.

This is why solvability and accessibility cannot be assessed fully independently. A record can note the overlap and say so rather than filing the observation under one criterion only.

Playtesting is the honest method, but it has limits

Playtesting is the standard way to find out whether a puzzle communicates what its designer thinks it communicates. Wikipedia’s article on the practice describes it as a way for designers to test a game’s completeness, and notes that a game’s testers are generally not told everything about the product so that balance and flow can be checked honestly.

For a selection desk, the useful version of this is a short, deliberate test: play the tutorial, then attempt the first puzzle without consulting anything else, and record where the information ran out. That is a bounded observation, not a session report, and it can be described in a record honestly without claiming more than was actually done.

What a single evaluator cannot establish is a population-level claim. “Solvable from the game alone” means “solvable by someone using only the game”, and one person solving it is evidence, not proof.

Writing the finding

The safe form of a solvability finding has three parts:

  • What the game states, quoted or closely paraphrased from in-game text.
  • What the deduction requires, described by rule, not by answer.
  • What is absent, if anything — with a note of whether its absence is acknowledged anywhere in the product.

That structure is publishable, checkable and spoiler-free. It also forces the assessor to distinguish a game that is hard from a game that is incomplete, which is the distinction a solvability criterion exists to make.

Sources

Written for: A designer or judge who has to state whether a puzzle game is solvable from its own information, without publishing hints that would spoil it for readers.

AI assistance recorded as: AI draft, not edited. Review status: approved.