A man in my Thursday group at the county library once spent eleven minutes constructing the strongest possible case for a position he does not hold. It was a good case. It was better organised than the one the author we were reading had managed. When he finished, a woman across the table who does hold that position said: that is not why.
Two years of Thursdays and it was the first time anybody had said it. We all heard it land. He asked her, reasonably, whether his version was not stronger than hers, and she said yes, probably, and that it still had nothing to do with why she believed what she believed.
I have been turning that over since. Steelmanning is supposed to be the courteous form of disagreement, and I have taught it, and I am no longer sure what it is doing.
Here is what troubles me. When I build the best version of an argument I do not hold, what am I actually using as raw material? Not the other person's reasons — I do not have those; that is the whole situation. I am using my own sense of what would be persuasive if I were in their position. So the artefact I produce is a report on my own susceptibilities, wearing their coat. It tells me what would have to be true for me to think that. It tells me nothing about what is true for them.
And then I defeat it, and I have the distinct sensation of having been fair.
Consider what has happened procedurally. I have removed the person from the exchange, replaced them with a model I built, argued with the model, won, and counted it as an encounter. If I had done that with a chess opponent nobody would call it a game. The courtesy is real — I did give the position my best effort — but courtesy to an argument is not the same as attention to a person, and the two have been quietly merged.
The objection is strong and I hold it myself half the time. What is the alternative? Engaging only with what people actually say means engaging with the worst available version — the half-thought, the slogan, the thing said in irritation at nine at night. That is how most public argument goes and it is plainly worse. A rule that says take them at their weakest is not an improvement on a rule that says take them at their strongest.
But notice that both rules have the same shape. Both take the position as the unit and the person as its container. The woman at my table was not asking to be steelmanned better. She was pointing out that the question of why she believed it had not been asked, by anyone, in eleven minutes of scrupulous fairness.
So what would asking look like? I do not think it is a technique, and I distrust my own urge to turn it into one, because a technique is exactly the thing you can perform while not listening. The nearest I have got is a question I now ask instead of building: what would have to change for you to stop believing this? The answers are almost never the ones my constructed version would have predicted. Frequently they are not propositions at all. They are a person, or a job, or something that happened in 1998.
Is that philosophy? I am genuinely unsure. It may be that I have replaced an intellectual exercise with a biographical one and called the trade a gain because it feels warmer.
What I can say is that the man came back the following Thursday and asked her the question, and she talked for twenty minutes, and at the end of it he said that he still disagreed and that he had been arguing with nobody for two years. Then he asked whether that meant the eleven minutes had been wasted. Nobody in the room could answer him, including me, and we have not raised it since, which I suspect is its own kind of answer.
The question you settled on is close to something in the commentarial tradition I grew up reading, where the opponent's position is stated by the opponent's school and not reconstructed. The difference is institutional rather than intellectual.