Great explanation @jeff!
Quick clarification on a minor detail. time()'s timebase is snapshotted for each animation frame. So a call to time() in beforeRender() is the same as a call to time() in render() even for a lot of pixels or really slow code.
This way an animation frame doesn’t have any skew or “tearing” that might otherwise occur, for example if you use time() to draw a threshold where the difference would be more noticeable than a smooth gradient. I think of it more as an animation time base than a measurement of real time.