Tuning Difficulty in Puzzle Games Based on Player Feedback

The nightmare of every puzzle game designer is: "This level is very easy for me, but will players be able to solve it, or will they close the game in the very first minute?" As a designer, because you know the rules and solutions of your own game by heart, you lose your objectivity. Correctly adjusting the difficulty level of a puzzle is a scientific process conducted not just by feelings, but through playtesting and analysis of telemetry data.

Flow State and Difficulty Balance

The Flow Theory, put forward by psychologist Mihaly Csikszentmihalyi, defines the mental state in which a person enjoys the activity they are doing the most. The flow balance in game design is as follows:

• If the game is too difficult for the player's skill, the player experiences anxiety, gets frustrated, and quits (Frustration).
• If the game is too easy for the player's skill, the player gets bored and loses interest (Boredom).

Our goal is to keep the player within the narrow "Flow Channel" between these two cliffs. As the levels progress, the difficulty should increase, but the player's skill should develop at the same rate.

Measuring Difficulty: Telemetry Data

During the development process of Syncron, we collected anonymous gameplay data via Firebase Analytics to measure the difficulty of the levels. The primary metrics we tracked were:

1. Completion Rate: What percentage of players who entered the level were able to finish it?
2. Average Moves: How many moves did it take to solve the level? (Comparison with the designer's optimal move count).
3. Reset Count: How many times did the player restart the level? (A high reset count indicates frustration).
4. Time Spent: How much time was spent on the level?

We encountered interesting results while analyzing this data. For example, we noticed that 40% of players got stuck and reset an average of 15 times on a level we placed as Level 5, thinking it was "very simple." When examined in detail, we realized a combination of mechanics was used that hadn't been taught yet. By slightly modifying the geometry of the level and removing a grid obstacle, we increased the completion rate to 90%, preventing players from abandoning the game.

Interpreting Player Feedback

Data tells us "where" the problem is, but not "why." For this, we used qualitative feedback from beta tests. We saw that when players said "This level looks impossible," they actually failed to perceive the visual cues of the mechanic. By clarifying the color codes of teleportation gates and making the neon lighting more distinct, we managed to lower the difficulty perception without touching the level design or code at all. Consequently, difficulty in puzzle games is not static. Listen to the data coming from your players, put aside your ego, and stretch your game according to their experiences.