An important and enduring discovery of generative syntax is that phrasal movements are not all alike. Since Chomsky 1981, they have been broadly classified into two types: A-movement and Ā-movement, with distinctive properties. In this talk, I pursue the view that the the A- vs Ā-distinction can be explained by considering how learners postulate movement rules based on the input that they are exposed, together with recent work on the geometry of Ā movements (Safir 2019 a.o.). The A- vs Ā-properties and the distinction itself are not innate, but are artifacts of the learning process.