So the question is, how good is this lower bound? Well, this lower bound is very strong. And if we compare the search tree for this strategy, you see that we get a solution with only this branching here. And the lower bound at the beginning is 14. And then 14 here which is the optimal solution, and 16, 17 and 17 on these other nodes. Whereas if you compute with the big M method, we get a lower bound of 0 at the beginning. And the lower bound doesn't increase until all the precedence we impose just make out a total order. Even for this small example the difference is striking.