The idea for this paper came while attending ISMP12 in Berlin. Prof. Kim and my friend Carlile Lavor gave a presentation in the special session I had organized with Nelson Maculan. Then, this work was completely performed, unfortunately, without meeting in person. In the final version of this paper, there is an imprecision about MD-jeep, when it is claimed that BetaMDGP performs better. This is actually true only if the first solution found by MD-jeep is considered. MD-jeep, however, is potentially able to enumerate the whole solution set. Therefore, the quality of its solutions are likely to improve when the executions are not stopped as soon as the first solution is reached.