Why it matters

The easiest way to win an argument is to pick a fight with the weakest version of the other side — the careless phrasing, the overstated claim, the example that does not hold. That is a strawman, and beating it proves nothing, because the position you actually need to answer was never on the table. Steelman construction does the reverse. Before you criticize a position, you build the strongest version of it you can — repairing its weak premises, filling its gaps with the most charitable reading, marshalling the best evidence its defenders could muster — and then you engage that. If your objection still lands against the best version, it is a real objection; if it only worked against the weak version, you have learned that you do not yet have an answer.

For example: someone wants to argue against a $15 federal minimum wage. The lazy move is to mock the slogan and cite a corner store that might close. The steelman move is to first build the case as its sharpest proponent would — the velocity of money (poor households spend every dollar immediately, so the stimulus is mechanical, not ideological), the historical precedent (the wage once supported a family of three and was eroded by inaction, so this is restoration not invention), the live test case (California raised its floor without the predicted collapse), and the hidden subsidy (low wages push labor costs onto taxpayers through food and housing aid). Only after that case stands at full strength do you press on it — and you find the honest objection is narrow: a single national number strains low-cost regions, so the principle survives but the instrument needs regional tuning. That objection is worth something precisely because it survived contact with the best version. The mockery never would have found it.

  • What it reveals. The strongest defensible form of a position — its load-bearing premises made explicit, its gaps filled charitably, its best evidence assembled — and then exactly what survives a critique aimed at that strong form rather than at a caricature.
  • How it changes the read. You stop asking “what’s wrong with how they said it?” and start asking “what’s the best version of what they mean, and does my objection beat that?” — which is the only test that makes your own conclusion trustworthy.
  • When to foreground it. You are about to argue against a position, or you hold a strong view and want to pressure-test it honestly — and you want the conclusion to be defensible because it defeated the best case, not a convenient weak one.
  • What you’d miss without it. That your objection may only work against a misreading; beat the strawman and you keep a belief you have never actually earned, while the real argument stands untouched on the other side.
  • Where it misleads. Built dishonestly it produces a tinman — a version that looks strong but is quietly engineered to lose — or it drifts into a different, more agreeable argument than the one actually held; both feel like steelmanning while doing the opposite.

Realtime examples

See real, dated analyses where this mode built the strongest version of a position in the news before testing it → Steelman Construction on Main Street Independent

How to invoke it in Ora

You have a position you want reconstructed at its strongest before it is critiqued — either a view you oppose and want to engage honestly, or one you hold and want pressure-tested.

State the position concretely, name who holds it, and ask:

“Steelman this position: [the claim]. Best case for it, strongest version — then tell me what survives.”

The phrases steelman, best case for, and strongest version of are what route you here. Two things sharpen the result. First, name the proponent or the source — the reconstruction draws on the intellectual lineage that gives a position its weight, and “the strongest case a serious defender of X would make” is far better than a bare claim. Second, supply your own view if you have one: the mode looks for genuine common ground between the strengthened position and your view, and that shared ground is often where the real analytical leverage hides.

One boundary worth knowing. This mode is asymmetric by design — it strengthens before it critiques, and the critique only ever targets the strong version. If what you actually want is an even-handed verdict that weighs strengths and weaknesses without tilting toward the strongest form first, that is a different tool (balanced critique). And if you want the case argued for a position you will then act on, or the case broken by an adversary, those are advocate and red-team modes. Steelman construction strengthens, then tests what holds.

How it works

The trick has a precise opposite, and naming it is the fastest way in. A strawman is a deliberately flimsy stand-in for someone’s position — you knock down a version they never held and declare victory. A steelman inverts every step of that. Instead of the weakest reading, you take the strongest; instead of exploiting gaps, you fill them with the most charitable inference; instead of holding the other side to their clumsiest sentence, you rewrite it the way their best advocate would. Then — and only then — you argue against what you have built.

The discipline behind this has a name philosophers use: the principle of charity. When you interpret what someone said, you assume they are rational and you choose the reading that makes their argument as strong and as sensible as the words allow. The reasons are not politeness. If you pick the dumb reading, you mostly learn that you picked a dumb reading; you learn nothing about whether they are right. The charitable reading is the only one whose defeat tells you something.

The most useful operational version comes from the mathematician and peace researcher Anatol Rapoport, whose rules for honest disagreement were popularized by the philosopher Daniel Dennett. Before you are allowed to say a single word of criticism, you must do three things in order. First, re-express the other person’s position so clearly, vividly, and fairly that they say, “Thank you — I wish I’d put it that way myself.” Second, list every point on which you agree — especially anything not widely shared. Third, say what you have learned from them. Only after all three may you offer one word of rebuttal. The genius of the sequence is that it is nearly impossible to fake. You cannot pass the first test while secretly building a version designed to lose; a proponent reading your reconstruction would catch the sabotage instantly. That single check — would a thoughtful defender of this position endorse my version of it? — is the whole load-bearing test. Call it the mirror test: the reconstruction passes when the other side sees their own argument in it, strengthened, not weakened.

The second and third Rapoport steps matter more than they look. Listing genuine agreements is not a concession ritual; shared premises are usually where the real leverage is, because they mark the ground you do not have to fight over and often reveal that the actual disagreement is narrower and more specific than the shouting suggested. And forcing yourself to name what you learned breaks the reflex of treating the other side as an obstacle rather than a source.

Then comes the part that makes the whole exercise worth the trouble: the critique, aimed only at the strong version. Here the discipline is unforgiving in a particular way. If one of your objections turns out not to apply to the steelman — if it only worked against the original weak phrasing — you drop the objection; you do not quietly weaken the steelman to make your objection fit again. And when the dust settles you give an honest accounting of what is left standing: what survived your best attack, what got modified by it, and what (only if the critique was genuinely decisive) was actually defeated. The minimum-wage example shows the shape of an honest result — the core economic and moral case survives intact, the choice of a single national number gets qualified, and almost nothing is outright defeated. That is the payoff of the method: a conclusion you can trust precisely because it was reached against the best the other side has, not the worst.

Framework & implementation

This section uses Ora’s own terms for the parts of an analysis, so that if you open the actual mode file they line up. Each is glossed in plain language on first use.

Pipeline execution

Steelman Construction is an atomic mode in the artifact-evaluation-by-stance territory — a single evaluative pass that takes one position and works it from a fixed stance, rather than a composite of sub-analyses. The stance is the defining axis of the territory: each mode in it engages an artifact from a committed posture, and this mode’s posture is constructive-strong — strengthen the position, then critique only the strengthened form. There is no heavier sibling along that axis; this is the canonical constructive-strong mode.

It runs at Gear 4, Ora’s most thorough setting: a Depth analyst and a Breadth analyst work the position in parallel and then critique each other (cross-adversarial evaluation) before a consolidator integrates the result. The adversarial structure is doing real work here — the mode’s central risk is a reconstruction that is secretly weak (a tinman), and a second analyst whose job is to attack the first analyst’s reconstruction is exactly the check that catches a steelman built to lose.

The pass enforces one hard ordering rule: construction completes before critique begins. The reconstruction is built fully — strongest premises surfaced, gaps filled charitably, best evidence assembled — and then, as a separate step, the critique is aimed at it. Interleaving the two (building and attacking in the same breath) is treated as a defect, because it lets the analyst shape the construction around the attack it already has in mind.

The mode’s reasoning tools ride in its ANALYTICAL PERSPECTIVES block — the lenses it loads as it works. The constructive backbone comes from Rapoport’s rules (re-express, agree, learn, then critique) and a charitable-interpretation discipline drawn from Dennett. Two further lenses are the ones this mode hosts. The walton-schemes-and-critical-questions lens (its foundational reasoning aid) supplies a catalog of recognized argument forms and the “critical questions” that test each — so the reconstruction can be built to fit a real argument scheme rather than a strawman shape, and so the later critique tests the position’s actual defeasibility conditions instead of inventing weaknesses. The devils-advocacy lens supplies the protocol for mounting good-faith, strongest-form opposition — the disciplined way to run the critique step so the objections are real rather than theatrical. A bias catalog (Kahneman–Tversky) sits underneath as a foundational check against the analyst’s own distortions.

Output contract

The deliverable is a fixed set of sections, so the work is auditable rather than a free-form essay. Original Position restates the position as actually given — and is bounded (kept short relative to the reconstruction) so the strengthening dominates rather than the weak original being repeated. Steelmanned Reconstruction is the largest section: the position rebuilt at full strength across multiple paragraphs, hidden premises made explicit, gaps filled with the most charitable inference, best evidence marshalled. Strength Identification names the load-bearing elements — the specific claims hardest to dismiss — and says where each sits in the reconstruction and why it is hard to answer. Points of Agreement lists at least two pieces of genuine common ground (between the steelmanned position and the user’s own view, when supplied) and what leverage each opens — not concessions, but shared ground. Critique of the Steelman is prose that engages only the strongest version; an objection that does not apply to it is dropped rather than the steelman weakened to fit. Survival Assessment gives the honest accounting: what survives the critique, what is modified by it, and what is defeated (named only when the critique is genuinely decisive). The format is prose-only by design — the mode carries a no_visual flag, and any diagram or visual envelope is a structural failure; if the back-and-forth structure needs visual rendering, that is a different mode’s job.

Origin and evidence

The method’s spine is the principle of charity, a doctrine of interpretation developed in twentieth-century analytic philosophy. The term is associated with Neil Wilson and with W. V. O. Quine, and was developed most fully by Donald Davidson, who argued that understanding anyone at all requires assuming they are largely rational and choosing interpretations that maximize the sense their words make — you cannot even begin to interpret a speaker you assume to be mostly wrong. The operational form most people actually use comes from Anatol Rapoport’s rules for productive disagreement, popularized by Daniel Dennett in Intuition Pumps and Other Tools for Thinking (2013), which gives the four-step sequence — re-express until the other side endorses your version, list agreements, say what you learned, then critique — that the mode’s construction stage implements directly. The discipline of building the reconstruction to fit a recognized argument scheme, and of testing it with the matched “critical questions” rather than ad-hoc doubts, comes from Douglas Walton’s work on argumentation schemes, set out with Chris Reed and Fabrizio Macagno in Argumentation Schemes (2008) — the catalog the hosted walton lens draws on. The contrast case the whole method defines itself against — the strawman fallacy — is one of the oldest entries in informal logic.

Applications and common uses

  • Op-ed and argument evaluation. Reconstruct a published position at its strongest before judging it — the use this mode is built for: best case for the policy or thesis, then the objection that survives it.
  • Pre-mortem on your own view. Steelman the position you already hold (or its opposite) to find out whether your conviction survives contact with the best counter-case, before you commit to it publicly.
  • Cross-partisan reading. Build the strongest version of the other side’s argument so a disagreement turns on the real load-bearing claim rather than on each side’s caricature of the other.
  • Negotiation and mediation preparation. Reconstruct the counterparty’s position so well they would endorse it — which surfaces the genuine common ground (the points-of-agreement step) that a deal can be built on.
  • Scholarship and review. Engage a rival theory or a paper under review at full strength, so the critique is one the author would recognize as fair and serious rather than a misreading.

Failure modes and when not to use it

  • The tinman trap. A reconstruction that looks strong but is quietly engineered to be knocked down by the critique that follows. The guard is the mirror test — would a thoughtful proponent endorse this version? — enforced by the cross-adversarial second analyst whose job is to catch exactly this.
  • Identity loss (drift). The steelman wanders into a different, more agreeable argument than the one actually held; it is strong, but it is no longer their position. The guard is that the core claim of the original must remain present throughout — strengthened, not swapped.
  • Retreat to the original. The critique scores its points by quietly aiming at the weak original phrasing instead of the strengthened version. The rule is absolute: the critique addresses only the steelman; an objection that fits only the weak original is dropped.
  • Projection. The reconstruction is filtered through the analyst’s own worldview rather than the proponent’s values, so the “charitable” inferences secretly serve the analyst. The guard is that charitable inferences must favour the proponent’s frame.
  • Eliding what survives. A verdict of “and so the steelman fails” that quietly skips the honest accounting of what holds. The survival assessment exists precisely to force that accounting — what survives, what is modified, what is genuinely defeated.

When not to reach for it. When you want the case argued for a position you intend to adopt or advance — that is advocacy of the opposing case, which routes to a red-team-advocate mode, not a strengthen-then-critique pass. When you want an adversary to break an artifact — to find how it fails under hostile pressure rather than how it stands at its best — that is a red-team-assessment. And when you want an even-handed verdict that weighs strengths against weaknesses without first tilting toward the strongest form, that is balanced-critique; steelman construction is asymmetric by design, and forcing it toward balance dissolves it into a different operation.

  • Balanced Critique — the neutral sibling in the same territory: when you want strengths and weaknesses weighed even-handedly rather than the position first tilted to its strongest form, this is the sideways-route.
  • Red-Team Assessment — the direct opposite stance: an adversary trying to break the artifact under hostile pressure, where this mode tries to build it at its best.
  • Red-Team Advocate — the mode for arguing the case for a position you will then act on, rather than strengthening it only to test what survives.
  • Walton Schemes and Critical Questions and Devil’s Advocacy — the two lenses this mode hosts: the catalog of real argument forms that keeps the reconstruction honest, and the protocol for mounting strongest-form, good-faith opposition in the critique step.