Walsh (2007, 2010) was correct to conclude that the way a biological population is described should affect conclusions about whether natural selection occurs, but wrong to conclude that natural selection is therefore not a cause. After providing a new argument that (Walsh, 2007) ignored crucial biological details, I give a biological illustration that motivates a fairly extreme dependence on description. I argue that contrary to an implication of (Otsuka, Turner, Allen, & Lloyd, 2011), biologists allow much flexibility in describing populations, as contemporary research on recent human evolution shows. Properly understood, such description-dependence is consistent with descriptions capturing different causal relations involving the same population. I thus show that Walsh's (2007) arguments fail for reasons that have not previously been understood; I argue that Walsh's (2010) more recent "Sure Thing" argument fails for similar reasons. The resulting view provides a new perspective on causation in evolutionary processes. © 2013 Elsevier Ltd.