For decades, experimental work has laid the foundation for our understanding of the linear and branched pathways that are integrated to form the metabolic networks on which life is built. Genetic and biochemical approaches applied in model organisms generate empirical data that correlate genes, gene products and their biological activities. In the post-genomic era, these results have served as the basis for the genome annotation that is routinely used to infer the metabolic capabilities of an organism and mathematically model the presumed metabolic network structure. At large, genome annotation and metabolic network reconstructions have demystified genomic content of non-culturable microorganisms and allowed researchers to explore the breadth of metabolisms in silico. Mis-annotation aside, it is unclear whether in silico reconstructions of metabolic structure from component parts accurately captures the higher levels of network organization and flux distribution. For this approach to provide accurate predictions, one must assume that the conservation of metabolic components leads to conservation of metabolic network architecture and function. This assumption has not been rigorously tested. Here we describe the implications of a recent study (MBio 5;7(1): e01840-15), which demonstrated that conservation of metabolic components was not sufficient to predict network structure and function.