it is not so terrible to look at... why do you dislike it?
First, we should always remember that
de gustibus et coloribus non est disputandum... I like Monica Belucci and I dis-like Nicole Kidman, for example.
However, I have thought about that, and I have decided that there is a
reason about it : In the one-collar version, one does not pay much attention to the asymmetry of the nipping structure, while, at the mirrored, duplicated two-collar version, the eye seeks the symmetry of the two ends ( around the collars ) at the middle of the knot s nub, too - and it remains disappointed, because it does not find it. Moreover, at the same time, this asymmetric double nipping structure occupies now the centre of the "stage", with the two collars at its two sides - so anything not perfectly smooth in it, even a minor discrepancy from absolute symmetry, is, in a sense, "magnified", regarding the amount of attention it attracts.
One might well ask : Why, then, you do not just use a "symmetric" Pretzel, where each leg leave the bight from different sides of its plane ? I feel that the "symmetric" Pretzel nipping structure is too "relaxed / open", its parts are not in contact, and do not cooperate with each other as much as they should, to keep itself in a closed, compact form, that would not need much aid from the collar structure. If, instead of the "symmetric" Pretzel, we use as a nipping structure one of the two asymmetric Pretzels, we can keep one of the legs underneath the other, so this nipping structure, although more asymmetric, would be more self-sufficient in ts effort to remain a closed form and do not degenerate into an open helix. Do not forget that we are searching for something at least as robust and efficient as the Girth hitch structure of the standard Mirrored bowline. I believe that the "symmetric" Pretzel, when it will be heavy loaded, it would have to rely to a very tightly interweaved collar structure, otherwise it will run the danger to 'open" up. I always prefer a nipping structure which is as self-sufficient as possible, and would be able to remain closed without transferring too much strain on the collar structure. The collar structure has one primary purpose, to remain attached on the standing part ! If a lot of its potential is dissipated, and "wasted", in its secondary purpose - which is to help the nipping structure remain in a closed form - I think that its primary purpose would have to be compromised a little bit...
I am also curious about this: in the images above, the pretzel's portion of the rope adjacent to the standing part of the mirrored version crosses passing over, while it passes under in the one-collar version... Is there a reason for this ?
Remain curious, keep looking at the details, and demand a reason !
In the one-collar version, we should try three things :
1. Drive the tail through the nipping loop at a location away from the crossing point - in the Pretzel double nipping structure case, away from the two crossing points of the Pretzel ( the contact points of the limbs of the Pretzel with its spine / belly )
2. Tie the Pretzel in a way the continutaion of the Standing end be in direct and extended ( helical) contact with the tail - so, it would better pass "under" the continuation of the eye-leg-of-the-Standing part, than "over" it.
3. As noticed by Dan lehman, before he had forgotten it (1), the curvature of the continuation of the Standing end into the knot s nub should better be as smooth as possible.
Now, the situation in the 2-collars version has been changed - so perhaps it would be prudent to change the way we tie the Pretzel, too !
I had tied it this way for a more concrete reason : The "middle" segment that goes through the nipping loop, the one that is collaring the two limbs of the Pretzel, should better pass through a path as close to the those two limbs as possible - and collar them in a symmetric way. Doing this, the "openings" of the two collars, between the two ends of this segment and the two limbs of the Pretzel, would be small. I do not want large openings, that give the impression of a loose knot, and may also be dangerous, too - because they are potential wide-open "handles" that can be caught up somewhere...
Let me correct some things : The only
pseudo-bowlines would be the bowlines that are intentionally
pretending that are true bowlines, while they are not. I have defined what I think a bowline is, and I follow my definition. Dan Lehman is free to follow his, but this does not makes me a liar and him a Chosen carrier of the truth...
Pseudo- means something that is a lie, and I think that I know what a pseudos ( ψευδος ) means, much better than Dan Lehman... His speciality is vegetables ( this
Turnip should be really tasty !
), but some words are all Greek to him.
1.
http://igkt.net/sm/index.php?topic=2050.0 
