Steel-Manning: The Skill That Separates Good from Great
The ability to argue your opponent's case better than they can is the highest form of understanding. Here are specific scripts, exercises, and cultural practices to build this skill in yourself and your team.
John Connor
Technology Strategist
Steel-manning—articulating opposing views so well that opponents say "yes, that's exactly what I believe"—is the single most valuable intellectual skill. This post provides scripts for conversations, exercises for practice, and patterns for building a steel-manning culture in teams.
The Pattern That Predicts Everything
I've watched hundreds of debates and sat in thousands of meetings. The pattern is clear: the best thinkers aren't the ones with the strongest opinions. They're the ones who can articulate opposing views so well that opponents say, "Yes, that's exactly what I believe."
This skill is called steel-manning—the opposite of straw-manning. Instead of attacking a weak version of the other side, you strengthen it, then address the strongest version.
The concept comes from a tradition older than the term. Medieval scholars practiced disputatio—formal debates where you had to articulate your opponent's position before refuting it. Thomas Aquinas's Summa Theologica follows this structure: state the objection in its strongest form, then respond. The Talmudic tradition requires articulating minority opinions before rejecting them.
We've lost this. Modern discourse goes straight to refutation without understanding. The result is that we argue past each other, never actually engaging with what the other side believes.
The Steel-Man Test
Before you argue against any position, you must pass this test:
Can you present the opposing view so accurately and charitably that the person who holds it would say, "Yes, that's exactly what I believe—you've actually stated it better than I could"?
If not, you don't understand it well enough to argue against it.
Script 1: The Disagreement Opener
Use this script when you disagree with someone in a meeting or conversation:
Part 1: Demonstrate understanding
"Before I share my concerns, let me make sure I understand your position correctly. You're saying [their position stated charitably]. The strongest reasons for this are [their best arguments]. Is that right?"
[Wait for confirmation. If they correct you, update and try again.]
Part 2: Present your counter
"Given that understanding, here's where I see it differently: [your counter-argument]."
Example in Practice
Situation: Your colleague wants to delay the product launch by 3 months for more testing.
Without steel-manning: "We can't keep delaying. We need to ship."
With steel-manning: "Let me make sure I understand. You're concerned that launching now risks damaging user trust if we have bugs in production. The testing we've done is surface-level, and the edge cases could be significant. Given our brand position, the cost of a buggy launch might exceed the cost of delay. Is that right?"
[Wait for confirmation]
"I share that concern. Where I see it differently: I think we can mitigate risk through a staged rollout to 10% of users first, with rapid rollback capability. This lets us learn from real usage while limiting blast radius. Would that address the concern?"
Script 2: The Devil's Advocate Brief
Before any significant decision, write a Devil's Advocate brief. This is a document arguing against your intended decision.
Decision under consideration: [State the decision]
The case against this decision:
- [Strongest argument against] because [reasoning]
- [Second strongest argument] because [reasoning]
- [Third argument] because [reasoning]
What would have to be true for this decision to fail:
- [Assumption that could be wrong]
- [Risk that could materialize]
- [External factor that could change]
Why I'm proceeding anyway: [Your response to the above]
Example: Hiring Decision
Decision: Hire Sarah for the senior engineer role
The case against:
- She has no experience in our tech stack (Python/Django). Learning curve will slow her down for 3-6 months.
- Her references emphasized individual contributor work, not the team leadership this role requires.
- At $180K, she's at the top of our range, limiting flexibility for other hires.
What would have to be true for this to fail:
- She can't learn Python quickly enough and becomes a bottleneck
- She struggles with ambiguity and needs more direction than we can provide
- Team chemistry issues emerge that weren't visible in interviews
Why I'm proceeding: Her system design skills transfer across languages, and her track record of learning new stacks quickly (she picked up Rust in 8 weeks at her last job) suggests Python won't be a barrier. We're explicitly hiring for growth into leadership, and her self-awareness about this gap is actually a positive signal.
Exercise 1: The Belief Swap
Practice arguing for positions you disagree with.
Choose a topic
Pick something you have strong opinions about.
Write a 3-minute speech for the opposite position
It must be genuinely persuasive—not a straw-man you can easily knock down.
Deliver it
Present to someone and ask: "Did that sound like I believed it?"
Topics to try:
| If You Believe... | Argue For... |
|---|---|
| Remote work is best | Mandatory office attendance |
| Move fast, break things | Slow and careful |
| Work-life balance matters | Total work commitment |
| AI skepticism | AI maximalism |
| Regulation is good | Free markets solve everything |
Why this works: If you can't argue a position persuasively, you don't fully understand it. The exercise forces genuine engagement.
Exercise 2: The Pre-Mortem
Before any major initiative, run a pre-mortem:
Gather the team
Everyone involved in the initiative.
Frame failure as certain
"Imagine it's 6 months from now. This project has failed completely. What happened?"
Silent writing (2 minutes)
Everyone writes down reasons silently. This prevents groupthink.
Share and discuss
No defending the project—only surfacing concerns.
What you'll discover: Concerns people had but didn't voice. Assumptions nobody questioned. Risks everyone saw but nobody mentioned. The pre-mortem gives permission to articulate doubt.
Exercise 3: The Ideological Turing Test
Developed by economist Bryan Caplan, this is the gold standard for understanding:
Find a disagreement partner
Someone who disagrees with you on something important.
Both write the opposing position
They write their position. You write their position as you understand it.
Blind test
A neutral third party guesses which is which.
If the third party can easily identify which is the "real" believer, you failed. Your version was recognizably a caricature.
Lighter version: After any disagreement, summarize the other person's view and ask: "Did I get that right?" Track how often they say yes without correction.
Building a Culture of Steel-Manning
For teams, not just individuals:
Meeting Rule: Acknowledge Before Attacking
No one can argue against a proposal until they've summarized it to the proposer's satisfaction. Literally enforce this: "Before you respond, summarize what you heard."
Assigned Contrarians
In important meetings, assign someone to argue against the emerging consensus. Make it explicit: "Maria, your job today is to find the holes. Make the best case you can against what we're proposing."
Reward Updates
Publicly celebrate when people change their minds. "I want to acknowledge that James updated his position based on the data. That's exactly what we want to see." Make intellectual flexibility high-status.
Track Predictions
Keep records of predictions and their outcomes. This creates accountability for beliefs. "Last quarter you predicted X. It came out Y. What do you think happened?" No blame—just learning.
Common Failure Modes
Failure Mode 1: Performative Steel-Manning
Going through the motions without genuine engagement. "Let me acknowledge your point..." [immediately dismisses it]
Fix: The test is whether the other person feels understood, not whether you said the right words.
Failure Mode 2: Analysis Paralysis
Using steel-manning to avoid decisions. "We need to consider more perspectives..."
Fix: Time-box the exercise. Steel-man, then decide. Don't let the process become procrastination.
Failure Mode 3: False Equivalence
Treating all positions as equally valid. Some positions are wrong even when steel-manned.
Fix: Steel-manning is about understanding, not agreeing. You can fully understand a position and still conclude it's wrong.
The Deeper Point
Steel-manning isn't a debate trick. It's a thinking upgrade.
When you truly understand opposing views, you see more of the problem space. You anticipate objections. You build more robust solutions. You maintain relationships with people who disagree with you.
The world doesn't need more people who are good at arguing. It needs more people who are good at understanding.
- This week: Use Script 1 in one real disagreement
- Next week: Write a Devil's Advocate brief for a pending decision
- This month: Run a pre-mortem with your team
- Ongoing: Track how often people say "yes, that's right" when you summarize their position