This question considers randomized algorithms for approximating the vertex-cover problem. What are some advantages and disadvantages of using randomized algorithms for this problem?