Into the Breach

Subset kept the most fun enemy and rebuilt the game around it. Into the Breach started as an Advance Wars-style tactics game with random, hidden enemy attacks. One enemy that highlighted the tile it would attack was the most enjoyable part, so Subset made telegraphed attacks the core of the game.

Each choice led to the next. If players can see and dodge attacks, the game becomes about protecting static buildings. Then pushing enemies makes their attacks miss or hit each other, which leads to tricking enemies into killing each other.

Build a rough prototype and follow what is fun. The whole design of Into the Breach followed from that one telegraphed foe, an example of a game that seems to design itself once the team listens to the prototype.

This way of working makes schedules hard to predict, so it depends on failing fast: cheap attempts, quick tools and placeholder art. Sam Coster of Butterscotch Shenanigans describes the process as "the game discovering itself over time."

Mechanics that couldn't be shown clearly were cut. Into the Breach displays every enemy action at once on a small grid as a web of arrows and icons. Subset removed enemies and weapons with long cause-and-effect chains or large-area attacks because they were too hard to display.

The HUD was designed together with the mechanics, not added late to patch them. Co-designer Justin Ma: "we would sacrifice cool ideas for the sake of clarity every time."

Animated previews explain weapons better than text. Weapon effects are shown as small animations of tiles moving instead of written descriptions.

Justin Ma: "You could type out a hundred times, 'Damages a tile and pushes adjacent tiles,' but showing that little animation of them moving is a thousand times more effective." Where text is still needed, use fewer words, stay consistent and avoid jargon.

Roll randomness before the player decides, not after. After FTL, which was packed with swingy output randomness, Subset built Into the Breach almost entirely on input randomness, randomness the player sees before acting. The result is fairer and more strategic.

Output randomness, like XCOM hit chances, breaks plans through bad luck rather than poor play, which is why it draws most of the anger. Geoff Engelstein calls this "the fundamental difference between randomness that supports strategy, and randomness that undercuts strategy."

Keep only output luck that can save the player. The one output roll left in Into the Breach is a small chance that an enemy attack on a building misses, which can rescue you from certain defeat. It is too small to rely on but feels great when it lands.

Justin Ma: "if there's randomness where you're expecting something bad and then you get something good, no one ever ever complains."

Full information has a cost: long, analytical turns. Because Into the Breach shows every enemy plan, a single turn can take 10 to 20 minutes of analysis. Perfect information invites paralysis and airtight plans.

Keith Burgun calls the limit on what a player can foresee the information horizon, set by complexity, execution, hidden information or randomness. In Into the Breach, not knowing enemy actions for the next turn is what keeps the horizon from being total.

Pair movement and chain effects so units combo. One unit chains electric shock damage through adjacent targets, while another grapples enemies and moves them around the map. Pulling enemies together sets up the chain.

A temporary carry-over gives a reason for one more run. When a run ends, one levelled-up soldier survives and can be used in the next attempt, but not permanently. It gives a leg-up without a lasting change to difficulty.

"It just gives you a bit of a leg-up, and also an excuse for 'just one more go'."