On this page
The shape of a good set
Across our three level-based games the same structure emerged independently, which suggests it is a property of the format rather than a preference:
- A free opening. One level that cannot be failed, which calibrates the controls and produces one guaranteed success.
- Single introductions. One level per mechanic, each demonstrating it alone.
- First combination. Two known mechanics together. Usually the first level people fail.
- A breather. Something easier immediately after, so the player re-establishes rhythm.
- Escalating combinations. Three mechanics, then four.
- A final level that combines and invents nothing.
That last point is the one most often broken. A final level introducing a new mechanic is asking the player to learn while under maximum pressure, which is the worst possible time.
The three sets we built
| Game | Levels | Mechanics | Ratio |
|---|---|---|---|
| Monster Castle | 8 | Gaps, wardens, crumbling stone, blades, keys, moving platforms | 4 teaching, 4 combining |
| Gravity Monster | 12 | Flip, crates, plates, doors, spikes | 5 teaching, 7 combining |
| Monster Cannon Challenge | 15 | Timber, stone, ice, stacking, shot allowance | 4 teaching, 11 combining |
The ratio shifts with how many interactions a mechanic has. Gravity flipping interacts with everything in its game, so it needed more chambers to explore than Monster Castle's largely independent hazards.
The breather is the piece that gets cut
After a difficult level, the next one should be easier. This is the piece that feels wasteful while you are building a set and is the piece that matters most.
A player who fails level six repeatedly and then immediately meets level seven at higher difficulty attempts it in a rattled state, and a rattled attempt reads as unfair even when it is not. An easier level seven lets them re-establish rhythm and meet level eight in a good state.
In Monster Castle, level five is the key level — mechanically gentle, structurally novel. It sits exactly after the first properly demanding level and exists mostly to be the breather.
Progression without levels
Endless games need the same structure and cannot use levels for it. They use time instead.
In Monster Tower, each chunk template has a minimum floor before it may appear, and is weighted by how far past that floor you are — so a mechanic is rare when it first becomes legal and normal later. That reproduces introduce-then-combine automatically.
In runners, the same job is done by the obstacle class schedule and the shrinking reaction allowance. A player meets each obstacle class alone early, when spacing is generous, and meets them combined later, when it is not.
Testing a set
- Can a first-time player finish level one? If not, it is not a free opening.
- Which level do people first fail? It should be the first combination, not an introduction.
- Do they know what killed them? If not, the problem is readability rather than difficulty.
- Does the failure rate dip after the hardest level? If not, the breather is missing or too hard.
- Does the final level introduce anything? It should not.
Difficulty is not monotonic
A good level set is not a straight line upward, and trying to make it one produces a worse experience.
The shape that works is a sawtooth: each level harder than the last within a group, then a drop at the start of the next group as a new mechanic is introduced gently, then climbing again from a higher floor.
That drop is what makes the introduction of a new mechanic safe. Teaching something new at the current difficulty ceiling means teaching it under maximum pressure, which is where learning is worst.
Pacing a set
A sequence of levels needs variation in intensity as much as it needs escalation in difficulty, and the two are different axes.
A set where every level is maximally demanding exhausts people regardless of whether each individual level is fair. What works instead is alternation: a demanding level followed by a shorter or more open one, then another demanding one, with the floor rising across the whole sequence.
The quieter levels are not filler. They are where a player consolidates what the previous one taught, and they are where the next escalation gets its contrast from — a hard level after an easy one reads as harder than the same level after another hard one.
The other pacing tool is length. Late levels should not be longer than early ones; difficulty should come from density rather than duration, because a long hard level punishes a late mistake by making you repeat everything that was already going well.
Where to put the hardest level
Not last. The hardest level of a set should sit around three quarters of the way through, with something slightly easier after it.
The reason is what a set feels like to finish. Ending on the hardest level means most players who reach the end are exhausted rather than triumphant, and many stop one level short of completion without ever seeing the ending they had earned.
A final level that is demanding but slightly more forgiving lets the set resolve. It also lets the player demonstrate everything they learned rather than be tested at their limit, which is a better final impression and does not make the set meaningfully easier.
Key takeaways
- A good set opens with a level that cannot be failed and closes with one that invents nothing.
- The teaching-to-combining ratio depends on how many interactions a mechanic has.
- The breather after the hardest level is the most-cut and most-important piece.
- Endless games reproduce the same structure using minimum-floor gating and weighted selection.
Frequently asked questions
Should the last level introduce a new mechanic?
No. A final level asking the player to learn under maximum pressure is the worst possible arrangement.
How many levels should a set have?
Enough for one introduction per mechanic plus combinations. That was eight for Monster Castle and twelve for Gravity Monster, whose central mechanic interacts with everything.
How do endless games do progression?
By gating each template behind a minimum level and weighting it by how far past that the player is, which reproduces introduce-then-combine automatically.
Should every level be harder than the last?
Should later levels be longer?
Should the last level be the hardest?
How do I introduce a new mechanic safely?
Alex Morgan
Browser Games Editor & Interactive Gaming Content Specialist
Alex Morgan is the editorial byline for everything published on Monster Arcade — the design notes behind all 30 original games, the strategy guides, and the tuning decisions that set every difficulty curve on the site. More about our editorial byline.
Published · Last updated
Games mentioned in this guide
Monster Castle
Climb a trap-filled castle as a small round monster. Eight hand-built levels with a real ending.
Gravity Monster
Flip gravity to walk on the ceiling. Twelve chambers where the safest floor is usually above you.
Monster Cannon Challenge
Launch a small fire imp at towers of blocks. Angle and power, fifteen stages, no time pressure.
Monster Jump Challenge
One button, twenty platforms. A pure jump-timing test that teaches the skill in ninety seconds.