One of the most resounding, counter-intuitive points from the late Brion Toss training vid to me is,
to flatten the fibers on the (what i call) the control side of the seam.
.
As always and always target keeping what would be the Standing Part(SPart)pristine w/o deformity as possible.
If the 'off-side'(Dent term for side opposing side lean as face target, making it the control side against that sidelean)/control side of the bight are what he called 'flat ribbons' they disturb the primary side of the bight less than if those 'secondaries' were left round.
ALSO, the now flat 'ribbons' have more surface area of contact thru the splicing than round would, for a double win!
This one very simple concept helped me look at all knots and rigs much harder from the maintaining the sanctity of the pristine SPart and then providing collective frictions, arcs and nips as best could .
.
A Bowline eye is weaker than splice because of the deformity in the SPart as a support column is increased many times in Bowline by comparison/ pristine sanctitiy of the SPart as a support column more fully violated in Bowline, than proper splice.
Both strategies use the primary force to lock secondary, Bowline has to deform more to make this lock to not burst 'seam'.
.
i wasn't concerned so much about pretty finish, so feeding the tails out of the end of the splice, and giving a Brion Toss patented
'butane backsplice' to make a bell on the end that could not possibly pull back thru the opening, let alone the gauntlet of arcs in the weave .
The sailor twine finish to appropriate same end is cleaner, and more finesse; but probably never went to class enough to have much class; just went with what worked! Generally very tight and bright taped over that end for identification and internal pressure on splice, leaving the bells out as inspection point.