Pathways that control complex biological phenomena must be robust enough to function reliably yet retain flexibility if they are to evolve. The ability of bacteria to form spores or move toward food clearly confers a fitness advantage, but it is unclear how the complex networks that control such capabilities evolve. Singh et al. performed a large-scale phylogenetic analysis of three microbial stress response gene networks--those involved in chemotaxis, sporulation, and competence to take up DNA--and found, across various species, that some networks are composed of distinct groups of genes that share similarities in function, phenotype, and sequence evolution rate, groups that they termed "modules." Components of the chemotaxis network clustered into five modules, which evolved independently from one another and often occurred in different combinations that did not correlate with species relatedness. The sporulation network comprised three modules that were very tightly linked to one another, so they were far more likely to be found in an all-or-none manner. In both the chemotaxis and sporulation networks, the combination of modules present in a species did not correlate with its phylogeny but did correlate with its phenotype. The authors also identified several previously undescribed genes in chemotaxis and sporulation modules and suggested that their inclusion in a particular module might be used to predict the function of these genes. Components of the competence network, however, did not form modules. For example, a gene that is present in two competent species may be required for conferring competence in one species but have no function in competence in the other. The authors hypothesize that differences in how these three networks have evolved are a consequence of how the network components are reused in other processes. Organization into modules may simplify repurposing of the components but also limits the points at which new inputs and outputs can feed into the system.