Until recently no universal agreement as to a philosophical or scientific methodological framework has been proposed to guide the study of fluvial geomorphology. An understanding of river form and process requires an understanding of the principles that govern the behaviour and evolution of alluvial rivers at the most fundamental level. To date, the investigations of such principles have followed four approaches: develop qualitative unifying theories that are usually untested; collect and examine data visually and statistically to define semi-quantitative relationships among variables; apply Newtonian theoretical and empirical mechanics in a reductionist manner; resolve the primary flow equations theoretically by assuming maximum or minimum outputs. Here we recommend not a fifth but an overarching philosophy to embrace all four: clarifying and formalising an understanding of the evolution of river channels and iterative directional changes in the context of least action principle (LAP), the theoretical basis of variational mechanics. LAP is exemplified in rivers in the form of maximum flow efficiency (MFE). A sophisticated understanding of evolution in its broadest sense is essential to understand how rivers adjust towards an optimum state rather than towards some other. Because rivers, as dynamic contemporary systems, flow in valleys that are commonly historical landforms and often tectonically determined, we propose that most of the world's alluvial rivers are over-powered for the work they must do. To remain stable they commonly evolve to expend surplus energy via a variety of dynamic equilibrium forms that will further adjust, where possible, to maximise their stability as much less common MFE forms in stationary equilibrium. This paper: 1. Shows that the theory of evolution is derived from, and applicable to, both the physical and biological sciences; 2. Focusses the development of theory in geomorphology on the development of equilibrium theory; 3. Proposes that river channels, like organisms, evolve teleomatically (progression towards an end-state by following natural laws) and iteratively (one stage forming the basis for the next) towards an optimal end-state; 4. Describes LAP as the methodological basis for understanding the self-adjustment alluvial channels towards MFE. 5. Acknowledges that whereas river channels that form within their unmodified alluvium evolve into optimal minimum-energy systems, exogenic variables, such as riparian or aquatic vegetation, can cause significant variations in resultant river-styles. We specifically attempt to address Luna Leopold's lament in 1994 that no clearly expressed philosophy explains the remarkable self-adjustment of alluvial channels.