actually, it doesn't even matter whether or not the a bottle that's half filled with water must have the other half filled with air or vice versa. in fact, it doesn't matter whatever the hell 'w' or 'a' equals to, just as long as one doesn't get confused by the difference between the abstraction of algrebra and the physical context in which it is applied here.
'(1/2)w' or '(1/2)a' can simply be intepreted as (literally) half a bottle of water or air [ie. a full bottle of water of air cut in half by a knife].
in this intepretation, the half bottle has half the initial volume but is still filled up with water or air. so it only makes sense that 2 such half bottles full of air will equal to 2 such half bottles full of water IF it is initially true, for whatever physical reason (which is not important here), to say that half a bottle of water = half a bottle of air.
the equality sign means nothing, it's not necessary to confuse it with any physical manifestations of equalities between a bottle of water and a bottle of air. they can in fact be made to be worth the same - eg. since it's not specified, then the size of the bottle containing the air can be significantly larger than the bottle of water, such that perhaps the cost of the larger bottle + the air is = the cost of the smaller bottle + the water. (only if price is at stake here).
if it's the water that counts, then the bottle containing the air can be large enough such that the number of water molecules in the water vapour of the air is equal to the number of water molecules in the smaller bottle with the liquid water...
(1/2)w = (1/2)a -----> w/a =1 ; ie. if a physical intepretation is needed, then it simply implies that the ratio of a certain property of the bottle with water and a property of the bottle with air is 1:1, ie. in
equivalent proportions.
hence, w =a is then perfectly correct if the variables 'w' and 'a' represent the same quantities for before and after.
so if the initial equality of (1/2)a = (1/2)w is true, then there is no paradox here at all. if it's not, then the question is non-existent via self-contradiction.
the whole "paradox" plays on the fact that we are supposed to accept (1/2)w = (1/2)a as a true inequality and at the same time give physical reality to the pronumerals 'w' and 'a':
let w = true ; and, a = false...
thus, (1/2)w = (1/2)a -----> x2, w = a -----> true = false
hence, it follows that if the argument for a bottle of water being equal to a bottle of air is valid, then what is true is also false, meaning that a bottle of water is then NOT equal to a bottle of air, self-contradicting the initial "paradox" - then the incompleteness theorem dictates that the "paradox" be 'indeterminate'.
it all falls upon interpretation