What Is a Steelman Argument (and Why You've Never Heard One)

A steelman argument is the strongest possible version of a position you disagree with: the version its smartest advocate would make, not the version that's easiest to knock down. It's the opposite of a strawman. And here's the uncomfortable part: on most contested topics, you have probably never heard the steelman of the other side. You've heard a caricature, argued against the caricature, won, and concluded your beliefs are strong.

Economist Bryan Caplan proposed a test for this, the Ideological Turing Test: can you explain the opposing side's position so well that someone on that side can't tell you're faking it? Most people fail immediately. Not because they're stupid. Because they've never encountered the real argument.

Why have you never heard the other side's best case?

Look at how information reaches you on any contested topic. Arguments for your position arrive as carefully reasoned essays, recommended books, and long conversations with people you respect. You get the best version of your own side. Arguments against your position arrive as headlines you scroll past, idiotic tweets your side mocks, and the worst spokespeople saying the dumbest version on cable news. You get the worst version of the other side.

This asymmetry isn't accidental. It's what happens when algorithms optimize for engagement and networks cluster by ideology. You get a high-resolution picture of your own side and a funhouse-mirror distortion of the other. That's training for a fight by punching air.

Does hearing the other side actually change minds?

Mere exposure doesn't. In a 2018 study, Christopher Bail and colleagues at Duke paid liberals to follow a conservative Twitter bot and conservatives a liberal one for a month. Nobody moderated. Conservatives became more conservative. Low-quality, context-free, adversarial exposure isn't argument. It's performance.

Structured disagreement tells a different story. When people engage opposing arguments in formats designed for reasoning rather than winning, like formal debate or deliberative polling, positions do shift. Not always toward the center, but toward more nuance and accuracy. Daniel Kahneman found the same thing at the individual level with the "consider the opposite" technique: generating reasons you might be wrong improves judgment, and people almost never do it spontaneously.

The variable isn't whether you hear the other side. It's whether you hear the best version of it, in a context that rewards thinking over performing.

What is the steelman gap?

Philosopher Daniel Dennett popularized rules for productive disagreement, attributed to Anatol Rapoport. The first: before criticizing a position, re-express it so clearly and charitably that your opponent says, "Thanks, I wish I'd put it that way."

Almost nobody does this. It's expensive: building the best version of an argument you find wrong means inhabiting a worldview that feels alien. So people argue against the dumb version, win easily, and learn nothing.

The steelman gap is the distance between the argument you think the other side is making and the argument they'd make if they were rigorous. Take gun control. If you're pro-regulation, the opposing argument you've heard is "I like guns and the government can't tell me what to do." The actual strongest case involves Swiss gun culture, defensive gun use, enforcement problems, and the disparate racial impact of gun laws. If you're anti-regulation, the version you've heard is "ban all guns like Australia." The strongest case involves public health framing and the distinction between ownership regulation and access regulation.

Both sides are arguing against ghosts. The real opponent never showed up.

Why are smart people the most vulnerable?

Intelligence doesn't protect you from motivated reasoning. Dan Kahan's Yale research found that people with higher scientific literacy were more polarized on politically charged science topics, not less. They're better at constructing sophisticated rationalizations for what they already believe.

This is the "soldier mindset" Julia Galef describes in The Scout Mindset: reasoning as a weapon to defend territory rather than a flashlight to explore. Smart people make better soldiers, which makes them worse scouts. It's part of why political polarization keeps deepening among the most informed. If you've never genuinely struggled to counter an argument against your position, you've never been in the room with the real counterargument.

How do you actually find the steelman?

The honest answer is that the infrastructure barely exists. Formal debate is niche. Deliberative polling happens in academic studies. Nobody at dinner wants to steelman a position the table finds repugnant. And the internet theoretically offers every argument ever made while in practice serving the worst ones in the most inflammatory packaging, a big part of why nobody argues in good faith anymore.

AI debate is useful here, not because AI is smarter than a human opponent, but because it has no ego investment in either side. It builds the steelman of any position without the social cost that stops humans. Strangely, not understanding its own arguments is exactly what makes it good at this.

Pick the position you hold most confidently. Could you explain the opposing view so well that someone who holds it would say "yes, that's fair"? If not, you've found a blind spot. The fix isn't to change your mind. Your position might survive contact with the strongest opposition. But you won't know until it has.

FAQ

What's the difference between a strawman and a steelman?

A strawman misrepresents an opposing position to make it easier to defeat. A steelman strengthens it, even beyond how it was originally stated, to make your own reasoning face the hardest test available.

How do you practice steelmanning?

Write out the opposing position before criticizing it, and check it against what advocates actually say. Or debate an opponent whose job is to give you the strongest case.