Genetic algorithms are an excellent approach to optimization problems in electromagnetics. Their chief benefit is the simplicity of coding and operation of the engine, which emulates aspects of Darwinian evolution to decide on the suitability of a set of antenna model parameters. It executes a 'pass or chop' decision on each set by determining the value of the objective function and culling out all but the best candidates within a generation. Hybrid 'gene' parameters incorporating a crossover of parameters from two parents provide for new candidates, while random mutation provides additional diversity. Perhaps the biggest impediment to GAs in electromagnetics is the size of the jobs they seek to tackle. As more and more complex problems are considered, the number of parameters becomes large. In turn the gene size becomes large and crossover and mutation variations become exponentially large. Rather than being a simple way of achieving optimization, large genes become a potentially intractable problem. The author introduces and discusses a method of reducing the size of the parameter gene through a compressed, fractal coding. Although this method is bound to be robust under limited conditions of use, it does suggest promise in many applications.