"Order" of rolls isn't quite the right way to think about that. You could throw 2 dice simultaneously and you'd come up with the same result (5 chances in 36) that you'd beat 5 bucks. This can be a particular confusing fallacy once you start layering in permutation and combination with probability.
In the case of the dice, order is no more important than color, location, or who throws it; order is just another way to help you think about 2 independent events.
Again possibilities are (A6,B4) (A6,B5) (A6,B6) (A5,B6) (A4,B6)
1/6*1/2 gives you the probability of the first three events occurring. To finish up you need to add 1/6*1/3 why not 1/2? why not (B6,A6) as well?
(B6,A6)=(A6,B6) and (A6,B4)=(B4, A6) order doesn't matter, discreteness does.
