Top-down 2D companion to the 3D version: each lane collapses the ball's flight to distance-traveled (x) vs. a single transverse offset (y) that stands in for the 3D drop-plus-curve from gravity and Magnus spin. The reactive paddle only knows the ball's offset with a small delay and chases it at a fixed top speed. The predictive paddle forward-integrates a = g₀ + k·spin the instant the ball launches, solves for the offset at the paddle plane, and heads straight there.
predicted: p(t) = 0.5·a·t², a = g0 + spinCoef(type)·mag
solve p at t = laneLen / vx → drive there now
reactive: target = offset delayed by Δt
paddle.y += clamp(target-paddle, maxSpeed·dt)
Watch the hit-rate gap widen as you push speed and spin up — that's exactly why real interception robots (ping-pong arms, catch robots) plan an intercept point instead of tracking the target reactively.