If you are wondering why a close up gravity source can interrupt a jump and drop them out right there, while others won't, think about this.
Imagine travel along a flat plane, like a giant piece of paper. Now imagine jump ships as fleas that jump a good 14' into the air. But you rate them based on how far from their starting point they land on the paper at. So a jump 1 only goes 1', while a jump 6 lands 6' from where it started.
Now, we add gravity wells. They are like jello blocks, cones, and cylinders floating just above plane. (Pick a flavor) If a flea is under jello when it jumps, it goes nowhere. If it tries to jump into one, it smacks into it, and slides down the gooey sides. If it's far enough away, it jumps just fine in a nice little parabolic arc.
In this silly analogy, you could even imagine a misjump as unexpectedly smacking into a wad of finger jello and bouncing off randomly.
If you are curious, the paper represents N-Space, while the area above it is a non-euclidean multi-dimensional (> 4) jump space. What would be a parabolic arc through that mess would probably be a crazy straw on acid trip for us 3 dimension types. That's why it needs big computers, and common sense has nothing to do with navigating through jump space.
That fits what has been explained in the books. It feels pretty good to me on both a science, and a technobabble front. And if you don't like like it, just come up with your own explanation for those pesky players that demand a copy of 'Jumpspace for N00bs'.