How We Design a Tricky Level 🧠

Start with the trap, not the answer

Most puzzle types start from a solution and hide it. A tricky puzzle works backwards. We start by asking: what assumption does everyone make here? That assumption is the trap. Only once we know the trap do we design an answer that lives just on the other side of it.

Say the level shows three candles and asks how many are left after you blow two out. The trap is arithmetic — you subtract. The real answer is the two you blew out, because the lit one burns down to nothing. We didn't invent the answer first; we found the wrong instinct first, then built the scene to trigger it.

The fairness test

The single rule that separates a clever brain teaser from a cheap one: the solution must be visible or reachable before the reveal. A tricky level is allowed to mislead you with framing, tone, or a distracting character. It is never allowed to require information you had no way to find.

Every level goes through a fairness pass where we ask three questions:

• Could a careful player find this on a calm second look?
• Does the twist use a rule the game has already taught (scroll, tilt, drag, re-read)?
• When you see the answer, do you groan "ohh" — or do you feel cheated?

If the honest reaction is "that's not fair," the level is cut or reworked, no matter how clever it seemed. This is the same principle we hold our characters to: mislead by tone, never hide the only clue.

One twist per level

The temptation is always to stack tricks — a hidden object and a word-play catch and a tilt. Resist it. Two twists don't double the delight; they usually just produce confusion, and the player can't tell which insight actually mattered. Our best levels have exactly one clean idea. The satisfaction comes from the crispness of that single realisation, not from a pile-up.

Tuning the difficulty curve

Early levels teach the vocabulary of trickery — that you can scroll, that words matter, that the frame isn't sacred. We introduce each mechanic gently, with a level where the twist is almost visible, before we start disguising it. By the time you're deep in the game, we can be ruthless, because you've learned to distrust the obvious. A good curve never feels like a wall; it feels like the game and your suspicion levelling up together.

Playtesting the "aha"

We watch real people play, muted, without hints. The metric we care about isn't just solve rate — it's the face. We're hunting for that specific micro-expression: a frown, a pause, then the eyebrows jumping up. That's the "aha." If testers solve it instantly, the trap is too weak. If they give up frustrated and the reveal makes them roll their eyes, it's unfair. The sweet spot is a short struggle followed by a delighted groan. When we hit it consistently, the level ships.