I've given this matter some more thought - I may publish an article about it - and I have figured out what bothers me about your 50-50 assumption. What you say in the article about a finite number of possible monetary amounts is true, but the implication that there can be a non-finite number of possible monetary amounts is false.

There is always a finite number of choices. There is no such thing as "all possible amounts of money" because "possible" contradicts "all." We have no reason to assume that every (or, for that matter, any) pair of envelopes is matched by an equal number of pairs where the larger amount in the first is equal to the smaller amount in the other. On the contrary, such a strict correspondence seems unlikely absent a declaration to that effect, and an declaration to that effect would entail an infinite number of envelopes, as each larger number seeks a pair where it is the smaller number.