Increasing returns to scale is the basis for many powerful results in economics and economic geography. But the limitations of assumptions about returns to scale in economic growth theories are often ignored when applied to geography. This leads to an unintentional bias favoring scale and mistaken conclusions about geography, scale and growth. Alternatively, this bias is used as a convenient modelling trick by urban economists to describe agglomeration economies for innovation without examining the spatial mechanisms that actually create agglomeration economies. I discuss techniques to focus on the distinctly geographic mechanisms that define returns to scale at appropriate spatial scales.