This runs entirely in your browser: the simulation itself, not a canned result, at a fixed seed so two visits agree. Take the alarm line and move it: where the two ribbons overlap, no value of it separates noise from the shift. The population is synthetic. Source and method.