Chapter 25 — Key Takeaways

1. Progression has two fundamental axes: player skill and character power. Pick your ratio deliberately. Celeste sits at one end (almost pure player-skill growth); Diablo sits at the other (almost pure character-power escalation); most games mix both. The ratio you choose determines how your game feels, what kind of player you attract, and how everything else in your progression system — XP curves, skill trees, caps — should be tuned. Write the ratio down before you tune a single number.

2. Your XP curve shape is a pacing instrument, not a math detail. Linear curves feel predictable. Exponential curves make late levels brag-worthy and late grinds brutal (ask any Vanilla WoW player). Polynomial curves (the modern default, exponent around 1.7) balance the two. Logarithmic curves flatten the grind and put progression weight elsewhere. Pick the family, then tune the parameters against your content pacing — not the other way around.

3. Skill trees produce build identity only if you force real choice. A tree where every optimal player takes the same path is an "illusion of choice" tree. A tree where everyone can take everything is a checklist. Identity comes from scarcity — skill-point scarcity, respec friction, mutual-exclusivity, or (as in Hollow Knight) a visible slot limit. Constraint is what makes a tree a tree. Without it, the tree is a cosmetic UI over a spreadsheet.

4. Unlock-based progression (the Metroidvania model) replaces numbers with spaces. Abilities work as keys that open new regions of the world. The Knight's world doubles in size when they get the wall jump; Ori's map opens up region by region; Samus's morph ball transforms her relationship to every passage she has seen. Unlock progression demands tight coupling between level design and ability design, but when done well, it produces the purest feeling of growth in game design.

5. Horizontal progression scales better than vertical progression over long content windows. Vertical progression (bigger numbers) works for a 40-hour campaign and collapses under live-service content releases that must keep raising the ceiling. Horizontal progression (new options at roughly the same power level) scales indefinitely but is harder to sell and harder to design. The long-running healthy games (Path of Exile, Magic: The Gathering) lean horizontal; the ones in constant squish-and-reset cycles lean vertical.

6. Soft caps encourage diversification; hard caps draw clean lines. A soft cap (diminishing returns past a threshold) nudges players to invest elsewhere without forbidding further investment in the same stat. A hard cap (an absolute ceiling) forbids further investment outright. Both are useful. Soft caps within a build, hard caps at the character-level ceiling. Avoid flat linear scaling — it produces single-stat specialists and destroys build diversity.

7. Power creep is the slow disease of every live-service game. Plan the response before the symptoms appear. New content releases tend to ship bigger numbers. Over time, old content is devalued, balance fractures, and new players cannot onboard. The industry's structural responses — standard rotation (Hearthstone, Magic), seasons (Diablo III, Path of Exile), prestige/reset systems (CoD, D2), and horizontal-heavy design — all work if committed to early. They rarely work if grafted on after five years of creep.

8. The "feeling of change" is what progression actually sells, not the numbers. The player at hour 30 should be meaningfully different from the player at hour 1. Sometimes that difference shows in numbers (HP, damage, stats). Sometimes it shows in abilities (what the character can do). Sometimes it shows in skill (what the player can do). The designer's job is to engineer the feeling of change, by whatever mix of mechanisms suits the game. Numbers alone without change produce hollow progression.

9. Playtest the middle of your progression hardest. Early-game progression is easy to tune because the content is dense and the reward rates are high. Late-game progression is often defensible because of the cachet of endgame. The middle — hours 10-25 in a 40-hour game — is where grinding valleys hide. This is where your XP curve, reward rates, and content pacing need the most scrutiny. If playtesters say "I got bored in the middle," the middle is where your numbers are wrong.

10. The skill tree, the XP curve, and the level design cannot be designed in isolation. They are interlocking systems. A skill tree that unlocks abilities the level designer has not planned spaces for is a skill tree whose unlocks feel random. An XP curve calibrated without content pacing is a curve that produces a grinding valley. A level designed without knowing the player's expected abilities is a level that will either assume too much or too little. Coordinate these three roles from day one, or rebuild them all at month six.

11. The single most common mistake is designing numbers for numbers' sake. You add a level system because RPGs have level systems; you add a skill tree because open-world games have skill trees. But if neither serves your game's specific promise to the player, cut them. Celeste does not need XP. Outer Wilds does not need a skill tree. Journey has neither. The best progression system is the one that serves what your game is actually selling — not the one convention requires.

12. First-draft progression is always wrong. Expect it. Iterate. Your first XP curve will produce a grinding valley. Your first skill tree will have an illusion of choice or a trap build or both. Your first set of ability unlocks will have one that feels pointless. None of this is a failure of design; it is the normal shape of design. The distinguishing skill is not getting it right on the first try — it is being willing to re-tune, to cut, to restructure after playtest. The designers who ship good progression systems are the ones who ship their tenth iteration, not their first.

One Sentence If You Remember Nothing Else

Progression is the engineered feeling of becoming someone the player was not when they started — and numbers, levels, unlocks, and trees are the tools you use to deliver that feeling, not the feeling itself.