I'm sorry to say that your example is not quite properly defined. There's no good addition operation that takes a set of 3 elements and adds it to a set of 4 elements in the manner you've described. Most vector addition (in most algebraic rings) would take a(n,n,n) + b(n,n,n,n) to be (a+b)[(2n,2n,2n,n)], which makes the assumption that a can be written a(n,n,n,0). I definitely see what you're doing, but vector addition by tuples is component-based (first coordinate of a added to first coordinate of b, nth coordinate of a added to nth coordinate of b, etc.
Any time you have an additive inverse (a number gained by subtraction from 0 in one way of thinking about it), that number is in the ring itself, and therefore all the same properties (associativity, commutativity, etc.) hold.
a - b is the same as a + (-b), which is equal to (-b) + a. Addition is commutative. a - b is different from b - a because b - a is b + (-a). If you add a-b and b-a, you'll see that they add to zero, so (a - b) and (b - a) are themselves additive inverses. What about this doesn't work? Recall that additive inverses are the same number for 0 only.
The only things guaranteed by these properties are existence (within a ring), and consistency. It would lead to all manner of inconsistencies if (a - b) and (b - a) were the same number, because then they'd both be 0 for all values of a and b, which means that all values of a and b would have to be zero. That isn't a very interesting ring, I'm afraid.
EDIT: Sorry if this is a bit dense, I am coming from a background in abstract algebra. Let me know if there's something I could explain better.