My formal training is as a theorist, specifically in orbital dynamics, but I quickly came to appreciate the need to tie theory to observations, and likewise to direct the course of observations by theoretical considerations. That is, observations need to be motivated not just by what is possible to observe, but as well by what is useful to observe: what observations are needed to critically test a theory or hypothesis, or to improve the determination of a physically meaningful quantity? My first application of this philosophy came early in my career at JPL, when I developed a theoretical model for the collisional evolution of asteroid spins (Harris 1979). While studies of asteroid rotation have been the central focus of my research activities, I have continued to pursue theoretical studies as well. In the past few years I have indulged in several studies that I would characterize more as amusements than rigorous science, but some nevertheless are quite important, and indeed have generated more public interest than more serious works. The most substantial of these has been a broad study of the hazard of impacts on the Earth by asteroids or comets, and what to do about it.