Cancer is the most intensively studied biological phenomenon, yet it remains poorly understood. Mortality and morbidity rates for many major cancer types have scarcely changed in decades. We posit that this lack of progress stems from a flawed conceptual model for the nature of cancer. A novel NCI physical science and cancer initiative encouraged us to re-consider the conceptual foundations of the current cancer paradigm, and we present an outline of an alternative view here. We focus on the deep evolutionary roots of cancer, and hypothesize that at least some hallmarks of the cancer phenotype express ancient ancestral pathways that are highly-conserved. The inappropriate expression of these pathways may be triggered by, but are not created by, mutational changes.