It has recently been proposed that the model plant, Arabidopsis thaliana (thale cress), uses a newly discovered genetic repair system to repair errors at the genetic level. A. thaliana uses information from the grandparent's genes as a basis for this correction - so genetic information appears to skip a generation. We apply this gene repair strategy to a combinatory optimization problem, firstly comparing the performance of parent and grandparent based repair. Subsequent experiments expand our understanding of the GeneRepair algorithm, by examining the parameters of fitness and direction involved in the generepair process. Our results point to a tentative explanation as to why A. thaliana might have evolved such an apparently complex inheritance process. © 2008 Springer-Verlag Berlin Heidelberg.