Validation that one can design robust networks regardless of the degree distribution and the system size. The relative robustness improvement R/R0 - 1 vs network size N for (A) scale-free networks with degree exponent γ = 2.5 and 3 and (B) Erdős-Rényi networks with 〈k〉 = 3.5 and 4. Starting from a given network, we swap two randomly chosen connections, that is, eij, which connects node i with node j, and ekl become eik and ejl, only if the robustness of the network is increased. This procedure is repeated until during the last 10,000 attempts no further improvement could be achieved. Note that the swapping keeps the degree of each node unchanged. Results are averaged over at least five independent initial networks. We do not show error bars, because they are smaller than the symbol sizes.