Object Collision As Verb

At its simplest, collision detection asks whether two shapes overlap this frame, and collision response decides what that overlap means. A pair of coordinates and radii is enough: if the distance between torpedo and ship falls below a threshold, the ship is destroyed. Everything expressive about it lives in the response half. Same test, different verb, and you have a bullet, a footstep on solid ground, a locked door, or a catch.

Spacewar! arrived in 1962 on a PDP-1 at MIT, where Steve Russell and his collaborators had a vector display, scarce memory, and no precedent to copy. Two ships, two torpedoes, and a star: for any of it to be a game rather than an animation, the program had to notice contact and act on it. Hardware offered no help, so the test was arithmetic, run every frame on positions the program already tracked.

Because contact is cheap to compute and endlessly reinterpretable, it became the substrate rather than a feature. Genres separate mostly by what they do after the overlap: subtract hit points, stop movement, start a grab animation, deform a mesh. Costs followed. Collision is where cheating, lag compensation and hitbox arguments live, and a mismatch between the visible body and the tested shape is still one of the most common sources of felt unfairness.