Thanks to Reynolds' magnificently titled paper of 1883, Linear Stability Theory (LST) has become the standard method, in spite of repeated quantitative contradiction by experiments, to determine the conditions under which a laminar fluid flow will become turbulent. Some simple concepts from dynamical systems can help shed light on the shortcomings of LST. As early as 1908, Orr suggested the possibility of finite-time instability -- an idea that has now been expanded and incorporated into a broader concept known as Optimal Perturbation Theory (OPT). In this talk I will present some recent results of applying OPT to the difficult problem of Kelvin-Helmholtz instability in the atmosphere.