Game Design
Designing fair-but-hard puzzles across 1000+ levels is harder than it sounds. Here is how the Stika Studio team approaches difficulty, flow state, and the
Game design advice usually sounds like this: "make it fun." But fun in a puzzle game is a fragile thing. Too easy and players feel bored and patronized. Too hard and they feel stuck and quit. The window between those two failures is narrower than you'd expect - and hitting it consistently across 1000+ levels is one of the hardest problems we've worked on at Stika Studio.
This is a behind-the-scenes look at how we approach puzzle difficulty in MyRin: what we got wrong first, what we eventually got right, and what we'd do differently starting over.
When we started designing MyRin's level set, we had a clear goal: a brain-training puzzle game that genuinely improved the player's cognitive ability - not one that just felt productive while accomplishing nothing.
That goal created an immediate tension. Genuine cognitive training requires working at the edge of your ability - tasks that are hard enough to demand real effort. But mobile games live and die on retention, and nothing kills retention faster than a level that makes a player feel stuck and stupid.
We needed puzzles that were actually difficult but never felt unfair. That's a very different design target than either "easy to complete" or "maximally challenging."
Our first approach was the obvious one: assign each puzzle a single difficulty score from 1 to 10, then arrange them in ascending order. Linear ramp, done.
It didn't work. The problem is that difficulty in logic puzzles isn't one-dimensional. A puzzle can be hard because it has many steps, or because one step requires a non-obvious insight, or because it applies a familiar rule in a genuinely unfamiliar context. These are different types of hard - and a linear score flattens all of them into a single number that doesn't predict the actual player experience.
In playtesting, players sailed through some "level 7" puzzles on the first try, then got completely stuck on "level 5" puzzles they should have found easier. The numbers weren't measuring what we thought they were.
We moved from a single difficulty axis to three separate dimensions:
A high-complexity, low-novelty, low-insight puzzle is hard but learnable through persistence. A low-complexity, high-novelty, high-insight puzzle can feel impossible approached the wrong way - but trivially easy once you see it. These two types require completely different player states and completely different pacing around them.
We now design the level order so that insight-dependent puzzles always come after a player has absorbed the underlying pattern in an easier context first. You don't hit the "aha" puzzle until the puzzle before it has already started building the mental model you'll need.
Csikszentmihalyi's flow model describes the zone between boredom and anxiety - where challenge is high enough to require full attention, but skill is high enough that success feels possible. That's the experience we're designing for in every MyRin level.
In practice, our target is this: a player should spend about 60 to 70% of a session in flow, with the rest split between "slightly too easy" (recovery time) and "slightly too hard" (the productive frustration that drives insight moments).
Getting there required extensive playtesting - not with team members who'd spent months staring at the puzzles, but with players encountering them completely fresh. The fresh-eyes experience is the only one that matters. Our assumptions about what was "obviously solvable" were wrong more often than we expected.
The full 1000+ level set is in MyRin now - each one placed where it is for a specific reason. If a level is frustrating you, the puzzle before it was supposed to give you the tool you need. Go back and look at it again.
Download MyRin on iOS · Download on Android
We'd instrument earlier. We started collecting meaningful session data - completion time, hint usage, retry count per level - later in development than we should have. By the time we had data showing which specific levels had abnormal drop-off rates, those levels had already shipped and revisiting them was expensive.
If we were starting MyRin today, we'd instrument every puzzle from the first external playtest. Not to optimize for completion rate (that's a trap - the hardest puzzles will always have higher drop-off, and some drop-off is healthy), but to detect levels that are failing for the wrong reasons: confusing rule presentation, unfair insight gaps, or unintentional difficulty spikes.
The difference between a player quitting because a puzzle is hard and a player quitting because a puzzle is unfair is invisible in aggregate retention data. You need per-level instrumentation to see it.
Every level in MyRin exists because it passed playtesting with real players who weren't on our team. If a level made it in, players who genuinely engaged with it solved it - not by guessing, but by working through the logic.
When you get stuck, that's the design working. The insight you need is in there. The puzzle before it was building toward this one. If you use a hint, you're not failing - you're seeing the solution that makes the next level click faster.
MyRin's 1000+ levels aren't a difficulty ramp. They're a cognitive curriculum. And like any curriculum, some lessons take longer than others.
You can explore them at stikastudio.com or download the app directly on iOS and Android.