Many of the original insights of effective altruism are insights about diminishing returns, especially from concave functions. In particular, ideas about concavity motivate the concept of neglectedness, and the heuristic of giving to the most marginally cost-effective charity. But sometimes we instead confront increasing returns, often from convex functions. In such cases, I think some of the simpler EA heuristics break down in instructive ways. I also suggest thinking in terms of ‘impact functions’ for projects, analogous to production functions for firms. These are scattered points, many of which are kind of obvious.