I don't think it assumes that. If one is arguing against a fundamentally bad position, the Steel Man of that position will still be bad -- and by addressing that, you will have cut off evasions at the pass. Steel Man can be used to preempt Motte-and-Bailey fallacies by better defining the point being attacked.
All of this is true, and it's the roadmap to tautological soundness.
But, in these scenarios, it's rarely about tautology or perfecting the structure of your argument. Yes, you might win on a debate stage. But, you've not moved anyone.
