It's not a double edged sword, Max. Your friendly compassionate rightwingers' argument is simply meaningless. Nothing useful can be said about changes in the poverty line solely by reference to the multiplier used to calculate it.
Basic food costs are multiplied by some number which is a proxy for the nonfood costs that are not measured. A change in the multiplier only has meaning with reference to a particular level of food costs. In fact, absent evidence that basic food costs declined in real terms, a rise in the multiplier *must* be due to an increase in nonfood spending (e.g., health costs), and the poverty line must be correspondingly higher (67 percent higher in the case of the multiplier rising fronm 3 to 5). However, if the multiplier rose mainly because real food costs declined, and if that decline was large enough, then a larger multiplier could result in a lower poverty line. But that's the only way that case can be made. Not by shabby data tricks as you describe.
It turns out that, as pointed out by Stone in the American Prospect article linked by Hank Leland (packed with good info, btw), in '67, soon after the poverty line was fabricated, Orshansky tried to convince the Johnson people to change the multiplier to 4, since "everyone knew" that's what it really was. Needless to say she got nowhere; everyone realized that would raise the line by 1/3 and cost a whole lot of money.
But at least she didn't have to put up with the ridiculous argument that if the multiplier was a larger number, that must mean the poverty line was really lower.
RO