Starting with Savageau's pioneering work regarding demand rules for gene regulation from the 1970s, here, we choose the simplest transcription network and ask: how does the cell choose a particular regulatory topology from all available possibilities? According to the demand rules, a cell chooses an activator based regulation of a target if the target protein is required for most of the time. On the other hand, if the target protein is only required sporadically, its control tends to be via a repressor-based regulatory topology. We study the natural distribution of topologies at genome, systems, and micro-levels in E. coli and observe deviations from demand rules. Analyzing the regulation of amino acid biosynthesis, transport, and carbon utilization in E. coli and B. subtilis, and comparing choice of topology with demand, we observe an alternate pattern emerging. Simulations of networks are used to help explain the natural distribution of topologies in nature. Overall, our results indicate that the choice of topology is drawn randomly from a pool of all networks which satisfy the dynamic requirements of the cell, as dictated by physiology. In short, our results suggest that the cell picks "whatever works".