The question wasn't changing the murder rate now, but changing "the population over a few centuries". If it doesn't change the population genetics significantly it won't do that.
If 1% of men are potential murderers, and we execute 10% of them in each generation, it will have huge a impact on the murder rate over a few centuries, even though not a lot of people got executed, and the overall genetics of the population hasn't changed much.
Well, no, that presumes "murderosity" is due to rare genes concentrated in murderers, not unfortunate combinations of genes widely spread in the population. Experience with "disease genes" has been they mostly of the latter type, with each gene having a minor effect.
