The concept of adaptive myelination—myelin plasticity regulated by activity—is an important advance for the field. What signals set up the adaptable pattern in the first place? Here we review work that demonstrates an intrinsic pathway within oligodendrocytes requiring only an axon-shaped substrate to generate multilayered and compacted myelin sheaths of a physiological length. Based on this, we discuss a model we proposed in 2015 which argues that myelination has two phases—intrinsic and then adaptive—which together generate “smart wiring,” in which active axons become more myelinated. This model explains why prior studies have failed to identify a signal necessary for central nervous system myelination and argues that myelination, like synapses, might contribute to learning by the activity-dependent modification of an initially hard-wired pattern.