I agree that challenging the axioms of a hypothetical can be understood as equivalent to changing the topic. But it simply doesn't follow from this that the proper way to respond to a hypothetical is never to change the topic! That's a patent logical fallacy. Your article is correct in most of its analysis, but the normative conclusion in the title is just not justified.

Do you really propose that the way to respond to Chalmers' Zombie hypothetical is to accept the axioms? But the axioms are ridiculous, and that's the problem!

In general, given a sufficiently strong set of axioms, a hypothetical can be constructed to irrefutably argue for any conclusion at all, with complete logical validity. Outlandish conclusions will require outlandish axioms, but they will be available. This might be ok if axioms were always clearly marked and explicit, but such is an impossibility with a language like English and its many hidden assumptions, and is doubly impossible given the need to reason with our brains which incorporate many axioms (read: biases) to which we have no conscious access.