The role of nature versus nurture, and the relative strengths and weaknesses of evolutionary and gradient-based approaches have been the subject of fiery debate in the AI community for quite some time. As outlined in the paper, these approaches have unique advantages and at the same time by no means exhaust the list of tools necessary for designing artificially intelligent behavior. In particular, the importance of parameter initialization is often underestimated, as demonstrated by the widely publicized 'lottery ticket hypothesis'. Following a discussion on why parameter initialization seems to be more important than previously acknowledged, a grid-based model is introduced which combines certain characteristics of evolutionary and gradient-based approaches with the goal of supporting robust parameter initialization.