The Worst Way to Teach a Game
A pop-up appears. A hand icon points at a button. Text explains what the button does. You tap it, the pop-up dismisses, and four seconds later you have forgotten everything it said.
Tutorial overlays are the least effective teaching tool in the medium, and they are ubiquitous on phones because they are cheap. The alternative — teaching through the arrangement of the levels themselves — costs real design effort and leaves no visible trace, which is precisely why it goes unremarked. At The Night Notebook we spend a lot of a game's first hour working out which of the two it is doing, because the answer predicts almost everything about the following ten hours.
Escalation as a Sequence of Questions
The cleanest way we have found to think about a difficulty curve is as a series of questions the game asks. Each stage asks one, and the useful property of a good sequence is that every question is answerable using something the previous questions established.
Consider how a well-built sequence introduces any new element:
- Present it in isolation. The new element appears where nothing else is happening and where failure is impossible or trivial. The player experiments without pressure.
- Make it necessary. The next stage cannot be completed without using it. The player moves from awareness to competence.
- Add a cost. Using it now has a downside — time, position, a resource — so the player must decide when rather than whether.
- Combine it. The element is paired with something learned earlier, and the interaction between the two produces a behaviour neither had alone.
- Apply pressure. The same combination, now under a time limit or against active opposition.
- Withhold it. Later, a stage is built around not having the element, which tests whether the player understood the principle or only the button.
Six steps, and not a word of text in any of them. When we play a game and can reconstruct this sequence afterwards, we are looking at deliberate design. When the first stage after the tutorial already sits at step five, the game is not teaching — it is filtering.
The Difference Between Losing and Being Stopped
Failure is not a design flaw. It is the mechanism by which a player learns anything in an interactive medium. But there is a large difference between failing and being stopped, and mobile games are unusually prone to the second.
Failing returns you to the attempt with new information and no cost beyond the time spent. The loop is fast, the loss is legible, and the natural response is to try again immediately.
Being stopped interrupts the attempt with something that is not the game: a countdown before another try, a consumed attempt allowance, an advertisement, a screen offering assistance in exchange for a purchase. The loop is slow, the loss is framed as a shortage rather than a mistake, and the natural response is to put the phone down or to pay.
The design question we always come back to is whether the interruption serves the player's understanding or the developer's revenue. Both can be true at once, and often are. What we look for is which one determined the placement.
Reading the Shape of a Run
Difficulty is best observed at a distance, so we keep notes across a long run and then look at the shape.
| Curve shape | What it usually indicates |
|---|---|
| Staircase — rises, then a comfortable plateau | A designer letting the player enjoy new competence |
| Smooth continuous ramp | Careful tuning, though it can feel monotonous |
| Sawtooth — sharp rise, immediate relief, repeat | Deliberate pacing, common in level-based puzzles |
| Flat, then a single cliff | Frequently a wall placed rather than designed |
| Ratchet — rises and never relaxes | Pacing pressure rather than teaching |
| Erratic — no relationship between stages | Levels authored independently without a curve |
None of these is disqualifying on its own. A cliff can be a genuine skill gate in a game that is honest about being demanding. A ratchet can be intentional in a short experience meant to end. The shape is a question to investigate, not a verdict.
The Thing We Watch For Most
If we had to reduce all of this to one test, it would be this: after a failure, can the player state what they will do differently?
Ask that after every loss and the answer sorts games quickly. In a game with a designed curve, the player almost always has an answer, even a wrong one, and the wrong answers are themselves productive. In a padded game the answer is some version of "try again and hope for a better start", which is not a plan. It is waiting.
That test also happens to be resistant to production values. An expensive game can fail it and a simple one can pass it, which is part of why our rankings sometimes look unusual next to a store listing sorted by popularity.
Mobile Makes This Harder
Two things about phones sharpen the problem.
Sessions are short and often interrupted, so a game cannot rely on the player holding a complicated lesson in their head between attempts. Teaching has to survive being put down mid-thought, which favours ideas introduced one at a time and revisited often.
And free-to-play economics put a genuine tension into the curve. A game funded by purchases has a commercial reason to make certain moments uncomfortable. Plenty of studios handle that tension honourably — selling convenience, cosmetics or content rather than relief from an artificial obstruction. We say so when they do, and we say so when they do not.
Our published figure is one editorial score, arrived at by our own judgement, and deliberately not the store's star average. Star averages are particularly unhelpful here, because a well-designed hard game and a padded one produce a similar spread of frustrated and satisfied ratings. The reasoning behind our scoring is on our methodology page, the current standings are on the ranking, and the desk that plays these games through to the wall is described on our about page.