Abstract
We argue that the important property of self-stabilization is, in principle, unstable across system classes. In particular, we show that, for a wide variety of system classes, there is no simulation that “preserves” or “enforces” self-stabilization.