True -- but it is possible to make lightweight preemptively-multitasked threads that get scheduled onto multiple cores. It's just that this paper wasn't talking about it.