Sounds like you and I at least are on the same page Erick.... I showed Dan how I normally handled ushiro ryotetori, and he very kindly showed me how I didn't need to play wristy-twisty aikido games ... broke many of the "rules" I taught my own students in the past - like avoid letting someone grab you on a 90 degree angle (where they can face you and you cannot face them).
I gathered the last from what you wrote, which was why, as it happened, I thought I would mention what we had just practiced -- where and how juji (90 degree orientation) can come up in ways that one doesn't necessarily see unless you are looking for it.
It is spatial (of which the sword exercise I described is one example) -- but also temporal, maai understood as interval -- a sine wave (spiral actually, but the same math applies). It applies to orientation in time as well as distance, but not merely who is first or last, closest or farthest away. If the spiral action is viewed as seeing that the opponent is screwed
juji represents a quarter turn of the screw, (and this is the important part -- or any odd number multiple thereof.
-- i.e.- 270 works as well, but not zero and definitely not 180). When Ikeda twitch-drops Joe Linebacker this is what he does -- really low amplitude -- REALLY HIGH angular velocity (m/2 *v^2), but that precise an interval at that short, fast period is (obviously) the product of much refinement, and he can do another multiple of the resonance interval in quick succession (like the whoop-de-do progressive sumi-otoshi, in miniature. Crack -- then shatter. (There is also a frequency issue here, but that is still only suggestive on my part) But I know where I am headed, in any event, with some idea how to get there.
As Chesterton said about falling (in a different context) -- there is only one angle at which we stand upright -- and every other angle is falling away. The same is true of dealing with the antagonistic force of gravity as with the antagonistic force of an opponent. With respect to gravity, 180 phase would be resisting gravity with counter-thrust thereby lifting into the air. I don't know about you but I have no rockets in my heels, and when I jump I fall back again. So this is not an option, "pushing back" or "pushing away" accomplishes net of zero and is doomed to failure by simple exhaustion if nothing else. 0/360 phase would be surrendering completely to the fall. The orientation to the attack must be similar, neither pushing back nor running away.
If I stay upright, keep juji, spatially, temporally, with respect to gravity AND the attack, because there is only one set of angles available with that quality (three space planes and one of time), and I occupy them with respect to both gravity and the attack, by definition he cannot -- and he is therefore "falling away." Conversely, if I do not occupy it he has little he needs to to do as I am already "falling away."
Seeing what juji is in any given interaction is a fascinating exercise.
Most people looking at attacks are looking at positive versus negative, 0 phase and 180 phase, in
and therefore the largest input or the earliest input "wins" or dominates the interaction. But if looked at and worked on from sideways, it doesn't work out that way. Phase difference really does work, and ahead or behind, a resonance interval (90 degrees offset, in time or space) royally screws up the input signal (attack). In military terms, the same principle operates in flanking action.
It leads to his attack momentum departing the stability limit of the structure that directs it, causing a buckling gyration in that structure (the mass of structure "trying" or "dragged" to follow the momentum path to recover it within the structural stability path, but consequently destroying the stability path progressively throughout the structure. (The head "snap" thing is this in action, and is actually very dangerous for an unprepared uke, because it can induce a whiplash injury very easily if done with too much energy.)
It also causes a discontinuity in the momentum (structure suddenly lagging instead of locked to momentum) which reverses the attack phase from the 0 to 180 or vice versa (and therefore the shifts the center of rotation -- or "statically," it shifts the center of the perceived moment).
The bio-mechanical aspect on involuntary reflexive action that may occur when this happens, (or when it happens in a proper way) is not trivial, in my opinion, but that is an area I am still exploring.
My opinion is you get a much better feel for this stuff when you apply it outside of aikido waza (where I'm less "comfortable").
I generally end class with a waza that I had in mind to begin it, after we have explored various elements of the interaction in "that" waza in isolation (which of course in a contingent real encounter -- "that" element of interaction may not lead to "that" waza at all. I usually do not do them in the order of progression of the waza. It is distinctly not linear.
We explore branch points that flow from each element, that do not lead to the waza I had in mind to begin with. The implicit atemi can be made explicit. The danger roads to kaeshi-waza can be noted to avoid (or exploit). The focus is on the reality of each immediate interaction -- not some pre-defined endpoint so it keeps both uke and nage honest in the immediate interaction. I typically do not mention for the class what the waza is that I am constructing with them and we just end up there at the end. It is a way to stay true to the received set of waza as valid "captured" studies in the principles (like a 4d recording), (and for their testing), without being them becoming bound to a given progression of interactions or events, so as to get caught in the kata trap.