For me the 1/3 guideline always mostly meant hinge fibers then start further than 1/3 back , and their tensioned backfield even further back, more into 1/2 way range for hinge of most width.
.
Most width affording more side to side control, and a thinner/ more flexible hinge. Especially in the tension backfield vs. compressed pivot to CoG side of front fibers.
.
Bluff hinge foward on hard forward lean, back on soft or back lean. On pure round each is probably at a loss of side to side control and further back makes CoG more towards target, but lose wedge leverage as lift or even safety backstop in trade/as stated. Each time chasing the CoG with compressed part of hinge as pivot. Stump is finite playground to align these choices on, capitalizing on the pluses as cap the losses
that occur in trade, or even capitialize on losses and or their byproduct(s).
.
If using shallow face in large tree would precut to triangle or strip back so final backcut is not too slow. May center punch face too, which also affords e-apportioning those fibers to sides for more steering control.
.
Always favor Tapered Hinge as built in self adjusting ballast against side force until tearoff. Simply exemplifying the side control to a single 'off side'(Dent). Side force ballasted out of formulae allowing remaining forward force to more deftly carry the whole package home to target. Tapered vs Standard both fold at same forward resistance to same load, only Tapered reapportions to greater resistnce against cross axis pull of side lean. Even Tapered pointing downward against side load to mostly horizontal path on long horizontal limb can work this way. Dent's lil'drawings persist to be true, even at these angles.
It seems BCs happen when internal stress overcomes internal strength.
i characterize as massive internal competing forces overflowing their container, over ruling the constitution of the spar. Massive train force lunging forward vs. a full stop pushing back, giving split decision of BC of neither winning nor losing the fight. Usually more from full across Dutching/early close I think. Because a step on 1side to close offers the other side as potential path of relief. Slow fall can allow force to build to this too, rather than more flowing forward fall. Velocity gives squared increase in E=MCsquared, carrying the tree more cleanly, deftly .
My point is stick to the rules until you know different. Look at the fibers, look at the tapered hinges.

Most best is L-earning to forensically read own stumps/hinges well enough to reverse engineer to consistent model of what is going on . What made what part go well or not, well enough to l-earn a ton from even other folk's hinges. i even remembering old hinges to gain unharvested lessons and confirmations as figured things out. Then even carrying same examination aloft to work more horizontally in tree, to then carry those lessons even more confidently back down to the base of the tree, as Dent showed them originally.