You take a two-turn double nipping loop, and you "twist" the "higher" turn, and/or the "lower" turn, and/or both, 180 degrees. Then, you tie a common or an "Eskimo" bowline s collar. One of those variations is the ABoK#1013 - two others are the bowlines you show here. (*)
A first problem is that those "twists" force the
standing part to follow sharp curves, which absorb a large portion of the nipping loop s constricting power. Notice that the standing part s curves at the Tresse bowline, or at the 8-loop and the 8-8 loop, are much smoother.
A second problem is that the
nipping structure becomes much more complex, but the
collar structure remains as simple as it is in the standard bowline. There is no theoretical reason or experimental proof that a complex double nipping structure would nip the tail harder than a simple single one. So, if this complex nipping structure is just wrapped around itself, but, at the same time, does not induce some severe deflexion on the path of the tail that penetrates it, its contribution to the bowline s security would be negligible, or even nil. ( A
tail that follows sharp curves inside a complex nipping structure slips less easily than a straight one - we can say that, in a way, it is "hooked" inside the knot s nub...),
Sharp standing part s curves, smooth tail s curves, that is the problem !
(*) P.S. In just the same way, the 8 - loop is not but a "twisted" Eskimo bowline - so, before we
measure its behaviour, we can not say if it is more secure ( regarding a slippage of the tail ) than a simple "Eskimo" bowline, or not...