Challenge only makes sense next to skill
Difficulty is not a property of a game on its own. A fast platforming section is brutal for a beginner and dull for an expert. What a player feels is the gap between the challenge in front of them and the skill they bring, and that gap changes as they learn.
This is why a difficulty curve is best thought of as a relationship over time: how the challenge rises as the player’s skill rises. A curve that rises too fast loses people to frustration. One that stays flat loses them to boredom.
Flow, frustration and boredom
Psychologists use the idea of flow for a state of absorbed attention, where a task is demanding but achievable, goals are clear and feedback is quick. Games are often cited as good at producing it. A simple picture is a channel: if challenge climbs well above skill, players tend toward anxiety or frustration; if skill sits well above challenge, they tend toward boredom; in between lies the zone where play feels engaging.
Treat this as a useful model rather than a law. It is hard to measure flow directly, players differ in how much difficulty they enjoy, and some deliberately seek punishing games where frustration is part of the appeal. The same hurdle can feel like a satisfying test to one player and a wall to another.
Frustration is not automatically bad either. What matters is whether it feels earned. Players tend to tolerate repeated failure better when:
- the cause of failure is understandable (“I mistimed the jump”)
- they can try again quickly
- they can see themselves improving
- the penalty for failing is small
Failure that seems random, unexplained or expensive tends to feel unfair, and that is where players leave.
Signals that suggest how a curve feels
Designers cannot read minds, so they look for behavior that hints at the experience. Two signals are especially common.
Retry counts. How many attempts does it take to clear a given section? A few retries is normal and often a sign of a healthy challenge. A median that is far above its neighbors suggests a spike. The median is usually more informative than the average, because a few players stuck for dozens of attempts can drag the average up.
Quit points. Where do players stop playing and not come back? A sudden concentration of quits at one location is a stronger sign of trouble than a slow, even trickle.
Neither number settles the question alone. A high retry count can reflect a boss fight that players love; a quit point can mark the natural end of a short session. They are prompts to investigate, ideally alongside direct observation of players, which is what playtest sessions provide.
A hypothetical example
These numbers are invented to show how to read the evidence. Suppose 1,000 new players start a four-section level. The table shows how many entered each section, how many quit there and never returned, and the typical number of attempts.
| Section | Entered | Quit | Quit rate | Median attempts |
|---|---|---|---|---|
| 1 | 1,000 | 30 | 3.0% | 2 |
| 2 | 970 | 25 | 2.6% | 2 |
| 3 | 945 | 189 | 20.0% | 11 |
| 4 | 756 | 23 | 3.0% | 3 |
The arithmetic holds together: 1,000 minus 30 leaves 970; 970 minus 25 leaves 945; and 945 minus 189 leaves 756. The quit rate in section 3 is 189 divided by 945, which is 20 percent, compared with roughly 3 percent elsewhere. The median attempts jump from 2 to 11 at the same spot.
That pattern points to section 3 as a likely spike, but it does not say why. The cause could be a surprise enemy with no warning, a missing checkpoint before a long stretch, an unclear objective, or a skill check that was never taught. Looking at the section, watching players try it and asking them questions are the next steps. Note also that only 756 of 1,000 players saw section 4, so any feedback about it comes from a group who already survived the spike.
Common ways a curve goes wrong
- A sudden jump. A new mechanic arrives with a test attached before the player has practiced it. Teaching in a safe space first helps, which is the job of good player onboarding.
- Long penalties. If failure sends you back several minutes, each retry costs more, so the same challenge feels harder. Checkpoint placement strongly shapes how a section feels.
- Hidden rules. Challenge that comes from not knowing what the game wants is a different thing from challenge that comes from execution. Players generally dislike the first more than the second.
- One size for all. A single difficulty setting rarely fits everyone. Selectable levels and assist modes help more players reach the part of the game they came for.
Balancing different players
Average numbers hide groups. Newcomers, veterans, players using different controllers and players with different needs can all experience the same section very differently. Splitting data by experience level, where you have it, often shows that a section is fine for one group and a wall for another. Designers can respond with optional help, adjustable settings or alternative routes rather than flattening the challenge for everyone.
What to take from it
Difficulty is judged by feel, and feel can be investigated but not read off a single statistic. Useful evidence combines numbers such as retries and quit points with direct observation and player comments. A good curve is rarely one that is easy; it is one where challenge, learning and recovery from failure fit together so that trying again feels like a reasonable choice.