The approach developed here is complementary to much of the research in theoretical and computational evolution. In short, skills are complex structures that have to be combined from the parents in meiosis, and that have to support mutation.
Research into skills, as envisaged here, has to produce a reasonable information representation that is sufficiently compressed to fit into genes.
A second issue is chunking and alignment. The recombined genes have to introduce and preserve variability in skills while producing mostly viable skills, e.g. that do not produce dogs that walk with the front legs while dog-paddling with the hind legs.
Learned skills also have to improve over time and generations, so mimicry is explored for apprenticeship-based skill passing that might be functionally similar to genetic evolution.