This is another clue that makes me think that the snag might have been more like the first video in this thread. A loop-only snag would make a much bigger and easier target.
Things happen so quickly during a fall, it’s hard to see in the moment.
I’m seeing this video as proof of your ability to easily untie the knot using whatever rope you are using in the video.
If you are out to prove this knot’s insecurity then you’ll have to show it in a more useful way.
And if you don’t care for it, just use another…
BTW, it is not just for tying in.
If you need to guard against snagging,
ack, that I think implies a more involved,
complex knotted structure. The mirrored bowline
has a SPart collar that leads of course to the eye
and tail-wise to a back’n’forth through the two
nipping loops; which two loops should impede
the sort of SPart-flow that you saw. (They did
stop that in slick HMPE for one Brion Toss test.)
One has other ways to make some compound
structure with greater protection; note that the bowline can be cast into a line atop any
other knot, then have its tail reeved into that.
If the single leg of SPart was fully loaded,
and each leg of eye 1/2 loaded;
as Equal & Opposite loaded pair..
.
Is definitively different than going around locking mechanism with force path
let alone using 1 side of the E&O pair to pull lock apart.
in many forms
An RT mechanism would make less likely
because of 3 half arcs upgrade from 1 of force
.
Have seen in some texts the reason for innie not outie on BE of Bowline was not to snag to invert when unloading ships.