Chapter 17 — Key Takeaways

1. The jump arc is the universe's constant. Before you place a single platform, lock in your jump height and horizontal jump distance. Every gap, platform, and challenge will be measured against this ruler. Changing it mid-development means rebuilding every level.

2. Coyote time and input buffering are non-negotiable. Forgiveness is not difficulty's enemy — it is its prerequisite. Even the hardest precision platformers (Celeste, Super Meat Boy) implement both. Without them, the game feels broken. With them, even hard sections feel fair.

3. Variable gravity makes jumps feel right. A real-world gravity constant produces a mechanical, abrupt jump. Use phased gravity: normal during rise, reduced near apex (for hang time), and increased during fall (for snappy descent). This single technique elevates a beginner platformer into something that feels professional.

4. Apply the introduce-test-twist-master pattern to every mechanic. Show the mechanic in safe context. Require it to progress. Combine it with another mechanic. Demand it under pressure. Skip any phase and the mechanic feels underutilized or unfair. This pattern, more than any other single principle, separates polished levels from amateur ones.

5. Enemies are level design, not decoration. Place enemies at decision points, not in empty corridors. Make them visible before they are dangerous. Layer two enemies into emergent puzzles. One new threat per screen in early levels — never two at once.

6. The camera is a character. Match the camera's behavior to your game's genre and speed. Aggressive lookahead for Sonic. Subtle lookahead for Mario. Smoothed dead zones for everything in between. Room boundaries prevent the camera from showing what shouldn't be seen.

7. Tilemap custom data layers are a design superpower. Annotate tiles with damage, climbability, biome, or any other property. Read the data at runtime to drive behavior. This shifts logic from code to authored content — easier to iterate, faster to design.

8. Hand-built and procedural are different crafts, not better/worse. Hand-built is denser, more memorable, finite. Procedural is varied, infinite, less tight. Pick based on whether you value re-experience or re-discovery. Hybrid (Spelunky-style) is harder than either alone.

9. Screen-based design forces clarity. A single-screen room must be readable in one glance. This constraint produces better challenges than scrolling levels because it eliminates clutter. Even continuous-scrolling games (like Celeste) benefit from designing each screen as a self-contained sentence.

10. 2D levels have rhythm. Platform spacing, enemy patterns, and hazard timing create implicit cadence. Tight rhythm = intensity. Looser spacing = breathing room. Vary intentionally; align to musical structure if it serves your aesthetic.

11. Speed and precision are opposing design philosophies. Sonic-style games trust the player to fly and design forgiving multi-path systems. Celeste-style games demand precision and reward execution with elegant single-screen puzzles. Pick a school. Don't try to serve both — you'll please neither audience.

12. Iteration beats inspiration. Your first draft of a level will be wrong in places you cannot predict. Playtest. Watch silently. Note where players hesitate, fail, or shine. Redesign the worst sections. Repeat. The polish you see in shipped games is almost entirely iteration, not first-pass brilliance.