What is a larva, if it is not what survives of an ancestor's adult, compressed into a transient pre-reproductive phase, as suggested by Haeckel's largely disreputed model of evolution by recapitulation? A recently published article hypothesizes that larva and adult of holometabolous insects are developmental expressions of two different genomes coexisting in the same animal as a result of an ancient hybridization event between an onychophoran and a primitive insect with eventless post-embryonic development. More likely, however, larvae originated from late embryonic or early post-embryonic stages of ancestors with direct development. Evolutionary novelties would thus be intercalary rather than terminal, with respect to the ancestor's ontogenetic schedule. This scenario, supported by current research on holometabolous insects and marine invertebrates with complex life cycles, offers a serious alternative to the traditional scenario ('what is early in ontogeny is also early in phylogeny') underlying the current perception of the evolution of genetic regulatory networks.