I used a very similar idea to create an indirect coding for neuroevolution that was successful in solving some tasks. It's based on a generalization of K^2 Trees and shares the same ideas of modularity and hierarchy.
I've used Z-order curves and related curves in Neuroevolution research. It allows conversion of an adjacency matrix to a spatial subtrate representation that preserves locality as it grows (Thus going from 1 dimension to 2 or 3). This technique is an alternative for HyperNEAT or ES-HyperNEAT and experiments demonstrate higher performance that ES-HyperNEAT on the modular retina task problem.