Does following optimized routes for single cars improve car routing?

We study the impact of deserting a pre-established path, determined by a navigation software, on the overall city traffic. To do so, we consider a cellular automaton model for vehicular traffic, where the cars travel between two randomly assigned points in the city following three different navigation strategies based on the minimization of the individual paths or travel times. We found, in general, that, above a critical car density, the transport improves in all strategies if we decrease the time that the vehicles persist in trying to follow a particular strategy when a route is blocked, namely, the mean flux increases, the individual travel times decrease, and the fluctuations of density in the streets decrease; consequently, deserting helps prevent traffic jams.