Abstract. I will discuss three related topics in climate dynamics. First, what are some of the scientific problems? Second, how should we address these problems? And third, what is the relationship of what we do, or should do, to the classical scientific method? Regarding the science problems, I will argue that many of the outstanding problems involve nonlinear dynamics and its interaction with other physical processes, as opposed to purely thermodynamic or radiative problems. Thus the response of the atmospheric circulation patterns to global warming (a dynamical problem) is much more poorly understood than is the response of the temperature itself. Regarding the methodology, we often use large and complicated numerical models to address these questions. Is this the best way? Does 'theory' in the conventional sense -- elegant equations making testable predictions -- still have a role? Perhaps numerical models are our modern-day theory? Throughout the talk I'll pose questions as much as answer them and I'll be giving my opinion on the matters, not necessarily the consensus of the community.