This document discusses generating networks with arbitrary properties. It begins by explaining the limitations of random graphs in capturing properties of real-world networks. Exponential random graph models are then introduced as a way to generate graphs with specific properties, but these models often produce degenerate graphs. The document proposes using a normal distribution over graph properties rather than just expected values. It describes using Monte Carlo Markov chain methods and integrating over Voronoi cells to sample graphs from this distribution. Finally, it shows an example of generating a network with the same properties as the real-world Karate Club network.