Really??!! 
How was it set, then?
Because I find it inevitable that loading will shift the knot
such that the hitching tail (blue cord, here) points more
towards its own S.Part's side of the fray than what here
seems to be even greater than 90deg to that,
which is more the sign of opposite-sides sheet bend
(which looks better, in a way, but seems to want to
slip beyond my acceptance/tolerance) !?
I find it very hard to get OTHER than this same-side
shifted result, now playing with small synthetic (shopping-bag
handles) braided cord, e.g.. Which is understandable, as the
blue S.Part is pulling first, hard, against the unloaded, orange
tail, so shifts that side *blue-wards*, pointing the blue tail so.
And the camera angle doesn't look to be deceiving?
(Did that puff of orange fibres come from the work, btw?
--bit of cut ends lower side of knot by blue ...)
As for the water bowline, it's not clear to me what the
bonus turn in the one leg around the other can do,
if not being a 2nd nip of the tail. Maybe ... "waterless".
On the sheet bend, not sure how to answer the how was it set question.
On the waterless bowline, as the photo shows, the bowline part of the knot remained fairly loose. Perhaps a classically tied water bowline would have performed better. Unorthodox as it was, this one worked effectively.
