In relation to security, the most essential part of those bowline extensions is not the 3rd or the 4th diameter going through the central nipping loop, but the second collar around one of the three limbs of the initial knot. The positive contribution of the second collar is certain, while that of the 3rd ( and, to a greater degree, of the fourth) diameter is debatable, and has yet to be demonstrated. The nipping loop takes a more round, apparently more efficient nipping shape, that is true, but the nipping force on each rope diameter is reduced by 1/3… And we can be sure that the nipping action on each individual rope strand of a bundle of many strands will get less efficient after a certain number… but we do not know what this number is. I will not surprised if it turns out that 4 strands going through a nipping loop are in fact nipped not much more, or even less efficiently than 3… and I would love to be sure that 3 is better than 2, as I am sure that 2 collars are better than one.