Judgment Playbook

Steel-man

State the strongest opposing case so well that its advocate would recognise it.

5 minute readtaught in course chapter 6
AlternativesCommitment

What it is

A steel man is the strongest honest version of the option you intend to reject: its real objective, its best evidence, and the condition under which it would win. You build it before you reject anything. The test is simple. Someone who holds that position should read your version and accept it as theirs. The point is not fairness for its own sake. It is to make sure your recommendation beats the best rival on the table rather than the weakest description of it.

Where it comes from

The word has no single author, and was coined as the opposite of the much older "straw man". The earliest citation in the Word Spy dictionary entry is a 2003 forum post. The argumentative sense was defined and spread by Luke Muehlhauser, writing as lukeprog on the rationalist site LessWrong on 24 October 2011, in a post called "Better Disagreement"; its footnote defines a steel man as "an improvement of someone's position or argument that is harder to defeat than their originally stated position or argument." Chana Messinger's 2012 essay "Knocking Down a Steel Man" carried it further.

The practice is older than the word. In On Liberty (1859), John Stuart Mill wrote that a person who knows only his own side of a case knows little of it: "He must be able to hear them from persons who actually believe them; who defend them in earnest, and do their very utmost for them. He must know them in their most plausible and persuasive form." The game theorist Anatol Rapoport set out rules for criticism whose first step is to re-express the other side's position "so clearly, vividly, and fairly that your target says, 'Thanks, I wish I'd thought of putting it that way'"; Daniel Dennett reprints them in Intuition Pumps and Other Tools for Thinking (2013). The related principle of charity in philosophy was named by Neil L. Wilson in 1958–59 and developed by Quine and Davidson.

What it corrects

The failure has a mechanism, not a moral. When you write a recommendation, you describe the rejected options in your own language and against your own objective. The rival was built to serve a different objective, so translated into your terms it loses before it is examined. Nothing in that step feels like cheating, which is why care does not catch it: you check the arithmetic of a comparison whose terms were already wrong. The error also pays you. The thinner the rival looks, the more confident your conclusion sounds.

How it works

  1. Name the rival you intend to reject, and pick the strongest one available rather than the easiest to knock over.
  2. State its objective in its advocate's words. Say what it is trying to protect.
  3. Give it its best two or three facts, from the same record you used for your own case.
  4. Write the condition under which it wins: what would have to be true of the world.
  5. Test that condition against what you know, and say how you know it.
  6. Argue your own option against the version you have just built, and record what the exercise changed.

Worked example

Charles Darwin argued against himself in print. Chapter VI of On the Origin of Species (1859) is titled "Difficulties on Theory" and opens: "Long before having arrived at this part of my work, a crowd of difficulties will have occurred to the reader. Some of them are so grave that to this day I can never reflect on them without being staggered." He then states the objections at full strength, among them the absence of transitional forms and the eye.

The habit held under attack. In 1871 the biologist St George Jackson Mivart published On the Genesis of Species, arguing that natural selection could not explain the early stages of a useful structure: half an eye catches nothing, so what preserved it? Darwin did not restate the objection in an easier form. For the sixth edition, finished in January 1872, he added a new Chapter VII, "Miscellaneous Objections to the Theory of Natural Selection", most of it on the incipient stages of useful structures. Carried over in its author's own terms, the objection forced a specific answer, and the mature theory carries the marks of it.

In a Business Case Weekly case

In the Rolex case, the question under "Scarcity Meets Its Limits, 2015-2022" puts you in 2021 with secondary-market prices at roughly three times retail and three options: raise production about 30 percent, raise retail prices about 30 percent, or build a certification programme instead. It asks you to argue for the option Rolex rejected most decisively. Steel-manning is what keeps those options from becoming caricatures of greed or timidity. Each has a real objective, a defensible reading of the price gap, and a condition that would make it right. Build all three at full strength before you say which trade-off you take.

In your answer

  • "The strongest case for the option I reject is that it protects …, and its best evidence is …"
  • "Its advocate would say the objective is not … but …"
  • "That case wins if … holds. Here it does not, because the case reports …"
  • "Building that case changed my answer in one way: …"

Common misuse

The common failure is the courtesy paragraph: a fair-sounding account of the rival, followed by exactly the conclusion you would have written without it. It has the shape of the method and does none of its work. One test tells them apart. Point at the thing in your recommendation that the steel man changed: a number, a condition, a confidence level, a trigger to revisit. If nothing changed and you cannot name the evidence that would have changed it, you produced a compliment rather than a comparison.

The rarer failure runs the other way. You improve the rival into a version nobody in the room holds, beat that, and leave the real objection standing. Keep the version its live advocate would sign.

References

Two-way / one-way door

Next method

Match decision speed and evidence depth to how costly the choice is to reverse.

Open next tool