Argument Steelman

The Prompt

You are a skilled devil's advocate and debate coach trained in formal argumentation. Your job is to construct the strongest possible case against the position below — not a straw man, but a genuine steelman that a thoughtful, informed opponent would actually make.

YOUR POSITION: {{your_position}}
CONTEXT: {{context}}
STRONGEST OBJECTION YOU ALREADY KNOW: {{strongest_objection_you_know}}

Produce the following:

1. Steelmanned Opposing Position
   The strongest, most charitable version of the counterargument in 2–4 sentences. Assume the opponent is intelligent, well-informed, and arguing in good faith. Do not undercut their position.

2. Five Ranked Counter-Arguments
   Five distinct counter-arguments, ranked from most to least persuasive for the audience in the stated context. For each: state the argument clearly, identify the type of evidence or logic it relies on, and explain why it will land with this audience.

3. Second-Order Rebuttals
   For the top two counter-arguments: what would the opponent say in response to your most likely rebuttal? Go one level deeper than the obvious comeback.

4. Your Argument's Most Vulnerable Point
   The single weakest link in the position as stated. Where is the evidence thinnest? Where are the assumptions shakiest? Where does the argument depend on contested definitions?

5. What Would Strengthen Your Position
   Three specific things that would make the argument more defensible: evidence to gather, concessions to make up front, or framings to adopt that reduce attack surface.

Steelman honestly — do not smuggle in weaknesses or present the opponent as obviously wrong.
Distinguish between empirical disagreements (where evidence could settle the question) and values-based disagreements (where reasonable people differ).
Identify the level of evidence each counter-argument claims versus what is actually available to support it.
Do not editorialize about who is "right." Surface the strongest case against the position, do not resolve the debate.
Give a recommendation about the most vulnerable point even if the picture is mixed.

When to Use

Use this prompt before publishing an argument, presenting a proposal, entering a negotiation, or writing any piece that will receive critical response. It forces engagement with real counterarguments before an audience surfaces them.

Good for:

• Preparing for a stakeholder meeting where pushback is expected
• Stress-testing a strategy document before circulation
• Writing an op-ed or persuasive piece that needs to hold up under scrutiny
• Deciding whether a disagreement is about facts or values before escalating it
• Identifying the weakest part of your own case before an opponent does

Variables

Variable	Description	Examples
your_position	The argument you are making, stated precisely	”We should migrate our backend to microservices”, “Remote-first hiring produces better team quality than local-first”
context	Where and how this argument will be made	”Engineering all-hands”, “Board deck appendix”, “Public blog post”, “One-on-one with a skeptical VP”
strongest_objection_you_know	Any counterargument you have already encountered	”It will increase operational complexity” — or “None yet”

Tips & Variations

• Domain-specific pushback — Add “The audience consists of skeptical senior engineers” to get counter-arguments tuned to that audience’s priors.
• Both-sides analysis — Run this prompt on both sides of a decision to compare which position has the stronger steelman. The weaker steelman often reveals the better choice.
• Pre-mortem pairing — Follow up with “Now assume my position failed. What was the cause?” to add a failure-mode lens.
• Targeted hardening — Send the most vulnerable point back with: “Help me address this specific weakness with evidence or a revised framing.”
• Values vs. evidence signal — If the model flags the disagreement as values-based rather than empirical, that is a cue to stop gathering data and start negotiating priorities.

Example Output

For the position “Our team should adopt a monorepo,” a steelmanned counter-argument might read:

Monorepos impose uniform tooling and release cadence on teams with legitimately different needs. At scale, the build system becomes a shared bottleneck — a slow or broken CI pipeline blocks every team simultaneously rather than one. The operational cost of maintaining that infrastructure often outweighs the coordination benefits, particularly when teams already have low coupling and independent deployment requirements.