Ford$-$Fulkerson algorithm is a greedy algorithm which finds a maximum flow on a network $G=(V,E,s,t,c)$ by iteratively picking augmenting paths on a residual network $G_f=(V,E_f,s,t,c_f)$ where $E_f=\{e$:$f(e)0\}$ and $c_f(e)=\{\begin{array}{ccc}c(e)-f(e)& if& \text{e}\text{i}\text{n}\text{E}\\ f(e)& if& e^{\text{r}\text{e}\text{v}\text{e}\text{r}\text{s}\text{e}}\text{i}\text{n}\text{E}\end{array}.$ What is the main motivation of augmenting flow iteratively on the residual network $G_f$ rather than on the original flow network $G.$