Anyone can knock down a straw version of an argument; it proves nothing but your aim. The honest critic first builds the strongest form of the case they mean to oppose — the steelman — and attacks that. Defeat the strawman and the idea walks away untouched; defeat the steelman and you have actually said something.
Model the target's strength as a function of charity: a strawman (low charity) is weak and easy to topple; the steelman (high charity) is the strongest defensible version. An attack of fixed force topples the target only when the target's strength is below it — so a successful attack on a strawman proves nothing (the real idea is stronger), while a successful attack on the steelman is real. The instrument shows the target strength rising with charity and whether the same attack still lands: only defeating the strong form counts.
'Argument strength' is a single scalar stand-in for a messy, multi-part thing; real steelmanning is qualitative craft, not a slider. The structural point — attacking the strong form is the only critique that means anything, and it is harder — is the honest content, and the method a fair critic owes.