As for direction of groundpath, now I'm thinking that it's primarily down because it is established by the mind/body's direction to the ground, thus, path to the ground or ground path.
Second, according to my fuzzy understanding, you can compress yourself along that groundpath, like a big rubber shock absorber, and third, expand rather forcefully back into the source of the incoming force, along the same path.
So it seems that the groundpath direction is down but it can be used to direct outgoing force along the line where the body has maximum support...
I asked in the spear thrusting thread whether the push starts in the rear foot or in the pelvis. After reading CXW's comments, I thought a little more. When we practice with the bo, we always have someone pushing the other end and we set up the ground path so that we can relax while holding the other guy immobile, letting him push hard into the earth. Once that is established, we know that the push comes into the hips and goes into the ground foot to the earth. Then the question is, where does the push start? If you start with the idea of having a load against you, you can see that the hip then pushes down into the rear foot, which pushes back up with a corkscrew inward spiral to the hips. So that would be consistent with CXW's push coming from the hip to the foot.
I think all of this is sort of "given" in Tim's spear thrusting thread. (I think.) From my experience with Aunkai, the balance of the incoming push and the push from the hip to the foot, with the foot pushing back upward in a corkscrew is basic. In other words, they set up the groundpath first and work with it all the way through the thrust.
But then just how does the hip effect a push into the rear foot?
Well, even though I've exaggerated things in the classical manner, you need to take the below aspect into focus when you're talking about "groundpaths" and direction. Still, the picture should give you an idea why a linear description of groundpath/kokyu/jin can lead you off into a questionable direction. A linear representation of groundpath is actually only an incremental look at the vector-resultants:
Is a groundpath up or down? Suppose, in a static situation, the guy is pushing the elephant: is the push up or down? Look at the picture and imagine that the elephant is resting backward (but everything is static) onto the man's hand: is the push up or down?
BTW... isn't that an interesting pattern to the collision of lines at the guy's middle?