The mathematical description of gene expression time profiles is generally a very hard task, given the overwhelming complexity of the underlying molecular components and their interaction network. In this talk I will try to approach this problem in a different perspective: in the presence of very stringent requirements (time, energy consumption,...) the behaviour of a biological entity may be forced to follow a stereotyped pattern amenable, as such, of mathematical description. For example, the crowd behavior induced by panic is well described by formal models able to accurately simulate group decisions. Interestingly, such models suggest actions for optimizing the flow toward the emergency exit. In this talk I will discuss the peculiar features of a generic "fast" response at the mRNA level and translate them in a straightforward mathematical model characterized by tunable parameters each having a precise biological meaning. The example of the SOS system in E. coli will be discussed together with experimental data provided by our laboratory.