Lessons for my own games

Notes for Godot work, opinionated. Hollow Knight is a three-person studio's 2017 title that holds up better than most AAA contemporaries. The systems are small in count and precise in execution; the lessons mostly orbit constraint discipline.

1. A pure numeric loadout budget is enough

Sparklite has Tetris pieces. Warframe has polarities. Spire has hand-draw probability. Hollow Knight has integers: ~45 charms, costs 1–5, budget 11. The simplest possible loadout-as-budget shape; thousands of viable builds.

The discipline is in the charm design, not the system shape. Every charm has to be interesting at 1–5 notches, and interesting in combination with at least 5 other charms, and readable from the icon. Team Cherry hit this bar consistently across ~45 designs. That's the work.

For my own games: don't over-engineer the loadout system before the loadout items are designed. A pure integer budget can carry a metroidvania.

2. Healing should be a moment of vulnerability

Hollow Knight's Focus locks the Knight in place for ~1.5s. Players who try to heal at the wrong moment eat the boss attack. Players who learn boss patterns find Focus windows.

This is the opposite of how most action games handle healing. Most use potions you chug instantly. Hollow Knight makes the heal itself a decision and a risk.

For my own designs: if combat depth is your goal, make healing risky. The 1.5-second commitment is a feature, not a UX problem.

3. A single resource for offense and defense forces real decisions

Soul fuels both Focus (heal) and spells (damage). You can't pour resources into both. Every Soul gauge is an offense-vs-defense decision.

This is the same insight as Spire's tight-energy-budget — the resource is the design. Hollow Knight just applies it to action combat instead of turn-based combat.

For my own work: don't bifurcate health and mana if you want tense combat. The single-resource constraint is what makes encounters interesting.

4. Earned cartography changes the player's mental model

Without a quest marker or default map, Hollow Knight players spend the first ~5 hours disoriented. Then something flips: they're paying attention to the world itself. Memory of landmarks matters. Every region has a place where you found Cornifer.

The cost is real: ~5h of friction. The payoff is the rest of the game feels like exploration in a way maps-by-default games don't.

For my own designs: if exploration is your design pillar, ship without a map by default. Or at least make the map something the player earns. Quest markers are the death of exploration.

5. Active recovery > passive recovery

Hollow Knight's Shade fight is more memorable than Dark Souls's bloodstain because you have to play it, not just visit it. The recovery is its own little encounter.

For any game with a "lose-something-on-death" mechanic: make the recovery active. A short fight. A short puzzle. Anything that's played, not just walked through.

6. Free DLC builds long-tail community

Team Cherry shipped four free DLCs in the year following 1.0 (Hidden Dreams, Lifeblood, Grimm Troupe, Godmaster). Each was a multi-week event for the community. Goodwill: enormous. Critical reception: each was treated like a small new release.

For my own work: if I have post-launch content, free DLC over paid DLC builds dramatically more goodwill for a small studio. The marketing-as-content model.

What I'd be careful about

• Removing all quest markers is a polarizing call. Hollow Knight's first-impression review window was kind to it because the press knew. New IP without that reputation may not get the same patience.
• Charm balance is an enormous QA cost. ~45 × ~45 charm-pair interactions = ~1000 combinations to think about. If your loadout system is small, design it small enough that you can balance every combo.
• Atmospheric storytelling without a quest log requires exceptional writing. Hollow Knight's lore is genuinely good. A weaker writing team would have produced a confusing game where players don't engage with the lore at all. Don't ship "no quest log" without writing the team can carry the load.
• The 1.5-second Focus timing is a tuning constant. 1.0s would be too easy; 2.0s would be punishing. The number matters. Don't ship a healing-vulnerability system without playtesting the timing window into the right band.