A strawman misrepresents an opponent's position as a weaker version, then defeats that. It's satisfying and useless — you've refuted something nobody believes. Steelmanning is the opposite discipline: reconstruct the strongest, most charitable version of the position, including arguments its actual advocates might not have made well, and engage with that.
This isn't politeness; it's self-interest. If your view can only survive against a caricature, you don't know whether it's correct — you just know you can beat a version nobody holds. The practical test is Rapoport's rule: you should be able to state the other side's position so accurately that they say 'yes, that's exactly what I think,' before you criticize it. Most people discover they can't. That failure is information: you were arguing against your model of them, not against them.
The philosopher Daniel Dennett popularized a set of rules for criticism originally formulated by the game theorist Anatol Rapoport, and they're deliberately demanding. Before you're permitted to criticize, Dennett writes, you must first re-express your target's position so clearly and fairly that they say 'Thanks, I wish I'd thought of putting it that way.' Second, list any points of agreement. Third, mention anything you've learned from them. Then you may rebut. Dennett's point was that this procedure works on the critic as much as the target — going through it honestly often reveals you'd misunderstood the position, or that your disagreement is narrower than you thought. John Stuart Mill made the sharper version of the case in 1859: 'He who knows only his own side of the case knows little of that.' Someone who has never seriously confronted the strongest opposing argument doesn't actually understand their own position — they've just never had it tested.