It is widely believed that given the same labeling budget, active learning (AL)algorithms like margin-based active learning achieve better predictiveperformance than passive learning (PL), albeit at a higher computational cost.Recent empirical evidence suggests that this added cost might be in vain, asmargin-based AL can sometimes perform even worse than PL. While existing worksoffer different explanations in the low-dimensional regime, this paper showsthat the underlying mechanism is entirely d…