My philosophy is that among the multitude of models of a complex system that one can build, only the very simple, robust solutions may have a chance to represent reality. Complex solutions with tens of fitting parameters may sometimes be of great practical use, but often misrepresent reality (the geocentric model of the Universe is one such example). The same philosophy applies to computational methods we develop: simple, robust solutions are preferred. As with any theoretical work, close contact with experimental groups is essential.