Steel-Manning Can Be More Useful Than Fact-Checking
How engaging with the strongest version of an opposing argument builds a deeper defense than verification alone. A practical guide with a downloadable worksheet.
There’s a common experience of reading something that felt persuasive in the moment and then, hours or days later, realizing we couldn’t quite explain why we’d found it so convincing. The argument made sense while we were inside of the argument. The facts seemed solid and the reasoning flowed. But something about the conclusion didn’t sit right, and when we tried to push back, we couldn’t locate the specific place where the logic had gone sideways.
The standard advice for moments like this tends to be some version of the same instruction: check the facts and verify the claim. And that advice isn’t wrong, exactly. Fact-checking is a real and necessary skill. But it addresses only one kind of vulnerability: the risk of believing something that’s factually false. And the information environment we’re navigating now produces other kinds of vulnerability that fact-checking alone can’t reach.
The most common image of misinformation is the outright fabrication, a claim that can be tested against the evidence and shown to be false. When a fake statistic circulates or a public figure is quoted saying something they never said, fact-checking can trace the claim back to its source and measure it against the record. But the most persuasive arguments circulating right now, the ones that shift public opinion and reshape how we think about entire categories of issues, often don’t need false claims to be effective.
These arguments rely on framing. They use selective emphasis, emotional resonance, strategic omission, and structural logic that feels airtight because every individual piece checks out. The facts are real and the sources are legitimate. And the conclusion is still misleading, because the persuasion isn’t happening at the level of individual claims. It’s happening in the architecture of the argument: which facts were chosen, which were left out, how the remaining facts were sequenced, and what emotional context was wrapped around them.
This is a gap in our current information literacy toolkit. We’ve invested heavily in teaching people to spot fabricated claims, and that investment has produced real results: the Duke Reporters’ Lab counted 110 fact-checking projects worldwide in 2014 and more than 450 by 2022, and as of early 2026, 25 U.S. states have media literacy laws on their books. But framing-based manipulation, which is as old as persuasion itself, has received less attention as something we can learn to recognize and resist. The result is a lopsided defense: strong against outright falsehood, much weaker against carefully assembled truth. The imbalance becomes clearest in the specific situations where every fact checks out and the argument is still misleading.
The Limits of Fact-Checking
Fact-checking has become one of our most prominent tools against misinformation for good reason: when the problem is a false claim, fact-checking works. At its core, fact-checking is an exercise in verification: does this claim hold up against the available evidence? Some fact-checking organizations use nuanced rating scales to capture degrees of accuracy, but the fundamental question being asked is still whether the claim is supported by the facts. That question is useful when the problem is fabricated information, and fabricated information is still a real and widespread problem. But fabrication is also the form of manipulation that’s easiest to catch, because it can be disproven with evidence. The harder strategy to counter, and often the more effective one, is constructing narratives from true facts arranged to support a predetermined conclusion.
Consider how this works in practice. Two news outlets can cover the same event using the same verified data and produce stories that lead readers to completely different conclusions. Neither outlet is lying. Both can point to their sources. The difference is in what gets emphasized and what gets left on the cutting room floor. A fact-checker verifying the individual claims in either article would confirm them as accurate, even though the two stories point in different directions. The manipulation is structural, not factual.
Or consider how persuasive arguments circulate on social media. A post might cite a real study and draw a conclusion that sounds evidence-based. Fact-checking the citation confirms the study exists. Fact-checking the data confirms the numbers are real. But the post might omit that the study’s own authors cautioned against the exact conclusion being drawn, or that subsequent research contradicted the findings entirely. Every checkable fact passes inspection. The argument is still misleading.
When we train ourselves only to ask “Is this true?”, we’re left without tools for these situations. And these situations are where we’re most vulnerable, because the presence of verified facts creates a false sense of security. If every individual claim in an argument can be confirmed, we’re naturally inclined to trust the conclusion, even when the conclusion depends on framing choices and omissions that a standard verification check wouldn’t surface. Developing a defense against this kind of manipulation starts with understanding two very different ways of engaging with arguments we disagree with.
Straw-Manning and Steel-Manning
Every day, we encounter arguments: in news articles, in social media posts, in conversations with coworkers and family, in political speeches, in marketing. And every day, the people making those arguments are making choices about how to present them. They’re choosing which evidence to include, which counterarguments to address, how to characterize the opposing view, and how to frame the conclusion. These choices are sometimes called rhetorical moves, which is just a way of saying that they’re strategic decisions about how to make an argument more persuasive. We all make these choices when we argue, whether we’re conscious of doing so or not.
Some of these choices are productive. When a speaker anticipates a strong objection and addresses it head-on, the overall argument gets stronger because the audience can see that the counterargument was considered and responded to. But other rhetorical moves work in the opposite direction, weakening the argument they’re responding to and making it easier to knock down a position by first making it flimsier than it really is. Recognizing the difference between these two kinds of moves changes how we evaluate every argument we encounter, including our own.
Two rhetorical moves in particular sit at opposite ends of this spectrum: straw-manning and steel-manning. A straw man weakens an opposing argument to make it easier to dismiss, while steel-manning does the reverse, strengthening an opposing argument to make our own engagement with it more rigorous. Both show up constantly in political debates, news coverage, social media arguments, and everyday conversations, and learning to identify each one changes how we process the information we’re taking in.
Straw-Manning
A straw man argument replaces someone’s real position with a distorted version that’s easier to attack. This distortion can be a deliberate rhetorical strategy, designed to win an argument without engaging with its substance. But it can also happen without any conscious intent, when we’re so certain of our own position that we stop trying to understand what the other side is saying and start arguing against what we assume they mean. The result, either way, is that we end up attacking a position that the people who hold the real view wouldn’t recognize as theirs.
Straw-manning tends to follow recognizable patterns:
Exaggerating to extremes.
A specific position gets inflated into the most extreme possible version of itself. The extreme version is easier to dismiss because it sounds unreasonable on its face, and dismissing the extreme version lets someone avoid engaging with the more moderate original position altogether. Someone who says AI tools can help students learn, for instance, gets characterized as wanting to replace teachers with machines.
Choosing the weakest opponent.
Every position has a range of people arguing for it, from careful and well-informed voices to inflammatory and poorly reasoned ones. This technique involves ignoring the careful voices entirely and responding only to the least credible one, as though that person speaks for everyone who holds a remotely similar view. This works because defeating a weak argument feels like defeating the whole position, even though the stronger and more common versions were never addressed. Instead of engaging with a detailed policy proposal about AI transparency, someone responds to the most inflammatory anti-AI post they can find and argues against that version instead.
Stripping away nuance.
A position built on conditional reasoning and careful distinctions gets flattened into an oversimplified line that loses the substance of the original argument. The simplified version can be knocked down quickly, while the full argument, with its reasoning and caveats intact, would require a much more serious response. A nuanced argument about how AI might displace certain types of jobs while creating others, for instance, gets reduced to "they think robots are coming for all our jobs."
Questioning motives instead of addressing the argument.
Rather than engaging with the substance of what someone argued, this technique redirects the conversation toward the presumed psychology of the person making the argument. If the conversation becomes about who someone is rather than what they said, the argument itself never has to be addressed on its merits. Someone raises a specific concern about AI-generated misinformation, and the response is "you're just afraid of new technology" rather than any engagement with the concern itself.
When we encounter a straw man, whether in media coverage or in a conversation, the most effective counter is to restate the original position accurately and identify specifically what was distorted. This shifts the discussion back to the real argument and makes the misrepresentation visible. Even when we disagree with the original position, naming the distortion keeps the conversation grounded in what people are arguing rather than what their opponents claim they’re arguing.
The challenge is that straw-manning works precisely because the distorted version tends to be more emotionally compelling than the real position. “They want to replace teachers with machines” provokes a stronger reaction than “They think AI tools can supplement classroom instruction.” In fast-moving conversations, especially in public forums and political debates, the straw man can take hold before anyone has a chance to correct the record, and once it takes hold it tends to become the version that sticks.
When we’re on the receiving end of a straw man, one of the more effective responses is to restate our actual position with enough specificity that the distortion becomes obvious by contrast. Rather than saying “That’s not what I said,” which sounds defensive and leaves the straw man intact, we can say something closer to: “My position is that AI tools can supplement what teachers are already doing in the classroom. The argument you’re responding to, that teachers should be replaced, isn’t one I’m making.” This forces the conversation back to the real disagreement and gives anyone listening the information they need to see the gap between the position being attacked and the position being held.
When we see straw-manning in media or political discourse, the counter is different because we’re usually observers rather than participants. In those situations, the skill is recognition: noticing when a commentator is responding to a position that nobody credible is making, or when a debate treats a caricature as though it represents the other side’s strongest case. That recognition doesn’t require us to respond publicly. It changes how much credibility we give the argument we’re consuming, because an argument that has only defeated a straw man hasn’t proven anything about the real position.
Steel-Manning
A steel man is the direct opposite: instead of weakening an argument to dismiss it more easily, we strengthen it to engage with it more seriously. When we steel-man a position we disagree with, we reconstruct the strongest possible version of that position, the version its smartest and most reasonable proponents would put forward if given the chance to make their best case. The method involves assuming good faith in the people who hold the opposing view and building the version of their argument that would be hardest for us to refute, setting aside the worst-case interpretations of their motives in order to engage with the substance of what they’re saying.
We steel-man an opposing argument for one of two reasons: either to discover that the opposing position has merit we hadn’t recognized (which makes our own understanding more accurate), or to be able to argue our own position from a much stronger and more grounded place (because we’ve already engaged with the best the other side has to offer rather than the worst). Both of these outcomes make us better thinkers. Neither requires us to abandon our original view.
There’s a practical test for whether a steel man has been done well: if we can restate the opposing position back to someone who holds it and they respond with something like, “Yes, that’s what I believe,” then we’ve succeeded. We’ve demonstrated that we understand their position on its own terms, not a distorted or simplified version of the position. Once we’ve reached that point of genuine understanding, the conversation changes. We know exactly which specific points we disagree on and which points the other side has right, which means we can focus our pushback on the real points of divergence instead of wasting energy on misunderstandings. And the person we’re engaging with is far more likely to listen to our counterarguments, because we’ve shown that we took the time to understand theirs first.
Steel-manning someone’s position doesn’t mean we’ve conceded anything. It means we’ve done the intellectual work of comprehending why they believe what they believe, so that if we still disagree, our disagreement is precise and informed rather than reactive. We’re not meeting them halfway; we’re making sure we know exactly where the halfway point is before we decide where we stand.
Steel-manning has its own recognizable characteristics:
Engaging with the strongest version of the argument.
When someone makes a point poorly or emotionally, our instinct is often to respond to the weakest interpretation of what they said. Steel-manning means doing the opposite: reconstructing the best version of what the person is trying to communicate, even when their phrasing makes it easy to dismiss. The weakest interpretation is rarely what the person means, and responding to it leads to an argument about words rather than ideas. If someone says "AI is going to destroy creativity," steel-manning asks us to consider what concern is driving that statement (perhaps that AI-generated content will crowd out original human expression, or that people will lose the motivation to develop creative skills if a machine can produce results instantly) rather than responding to the literal claim that creativity will cease to exist.
Taking the evidence seriously.
When we disagree with a conclusion, the natural impulse is to dismiss the evidence along with the conclusion. But the evidence and the conclusion are separate things: the research or observations someone cites in support of their argument are often real, even when the conclusion drawn from them is one we don't share. Steel-manning means engaging with that evidence on its own terms and asking what it does and doesn't support, rather than waving it away because we don't like the other person's conclusion. If someone arguing against AI in recruitment cites data showing that a specific AI screening tool rejected qualified candidates at higher rates from certain demographic groups, that data stands on its own, even if we disagree with their conclusion that AI shouldn't be used in hiring at all.
Locating the precise point of disagreement.
Once we've built the strongest version of the opposing argument and engaged with its evidence, we can pinpoint exactly where our reasoning diverges, and just as importantly, where it doesn't. Before steel-manning, a disagreement often feels like a broad sense that "they're wrong." After steel-manning, we can see which parts of their argument we find convincing and where exactly we stop being convinced. That turns a vague reaction into something specific and articulable, for example: "I accept their evidence about the economic pressure AI puts on creative professionals, but I weigh the benefits of broader access to creative tools differently." That kind of precision makes the disagreement productive rather than circular, because both sides can focus on the actual point of contention rather than talking past each other.
Engaging with the substance, not the person.
Steel-manning starts from the premise that the person making the argument has legitimate reasons for holding their position, even when we disagree with their conclusion. Rather than speculating about someone's motives or dismissing them as uninformed, steel-manning addresses what they said on its own terms. If someone argues that AI chatbots shouldn't be used for mental health counseling, steel-manning engages with their specific concerns about AI's ability to handle crisis situations and read the nonverbal cues that human therapists rely on, rather than dismissing the person as "someone who doesn't care about the millions of people who can't afford a therapist."
Steel-manning isn’t necessarily about “seeing the other side” or conceding ground. We’re not looking for compromise or trying to meet in the middle. We’re stress-testing our own reasoning against the best available challenge, and the result is a position that has been refined by contact with the strongest counterarguments rather than insulated from them.
The adversarial legal system works on a similar principle. A defense attorney who builds the strongest possible case for their client isn’t declaring a personal belief in the client’s innocence; they’re providing the vigorous advocacy that ensures the prosecution’s case gets properly tested before a verdict is reached. When we steel-man an argument, we’re doing something similar: ensuring that our own position has been tested against the best available opposition, not the weakest.
Steel-manning is also distinct from playing devil’s advocate, which tends to be a contrarian exercise performed without a specific analytical goal. Devil’s advocacy often means arguing a position we don’t hold just to see where the conversation goes, while steel-manning has a more focused purpose: locating the precise point of disagreement between two positions by engaging with both at their strongest. And steel-manning is different from the “both sides” reflex that treats every disagreement as though the truth must be somewhere in the middle. Sometimes, after steel-manning an opposing argument, we come away more convinced of our original position, because we’ve identified exactly where the opposing argument’s reasoning breaks down and can articulate why.
Learning to steel-man also sharpens our ability to recognize when someone else isn’t doing it: when a commentator is responding to a caricature rather than a real position, or when an argument seems persuasive primarily because it never engages with the strongest version of its opposition. That recognition helps us evaluate not just whether a claim is true, but whether the reasoning behind the claim has been tested against its best counterarguments. When we can see that an argument has only engaged with the weakest version of its opposition, we know that the conclusion, however confident it sounds, hasn’t been earned.
How to Steel-Man in Practice
Knowing what steel-manning is and recognizing it in the abstract is one thing; doing it in real time is another. The practice follows a concrete set of steps, but it does require some discipline, because when we encounter an argument we disagree with, our first impulse is usually to look for the weak point, the thing we can refute. Steel-manning asks us to resist that impulse temporarily.
Restate the argument in the most generous terms possible.
We can start by stripping away the inflammatory language and the bad-faith framing. What is the core claim? If the person making this argument were thoughtful and arguing in good faith, how would they phrase their position?
Identify the strongest evidence supporting the argument.
Even if we think the conclusion is wrong, what real-world observations or data points would a reasonable person cite in its favor? Where does the argument connect to something observable and verifiable? This step often reveals that opposing positions are grounded in real evidence we hadn’t considered, which is useful information whether or not it changes our mind.
Name the values driving the argument.
Many disagreements, especially polarized ones, have a values component that runs deeper than the factual claims on the surface. Two people can agree on the same data and still reach different conclusions because they’re weighing competing priorities. Someone arguing against AI regulation might be prioritizing the speed of innovation and the risk that regulation would entrench dominant companies. Someone arguing for AI regulation might be prioritizing public accountability and protection for people affected by automated decisions. Both sets of priorities are legitimate. The disagreement is about which one takes precedence in a specific context.
Find the precise point where we diverge.
Once we’ve built the strongest version of the opposing argument, the next step is to go through it piece by piece. Which parts of the argument do we find convincing, and where exactly does our reasoning stop following theirs? Working through the steel-manned argument this way narrows the disagreement from a broad reaction to a specific location. Maybe we accept the same facts but weigh the risks differently. Maybe we share the same goal but disagree about the mechanism for reaching the goal. Locating this point transforms a vague sense of “that’s wrong” into a specific, articulable position.
Working Through It
The downloadable worksheet below walks through the complete steel-manning process as a guided exercise. It starts with choosing a specific position we disagree with, then breaks the reconstruction into four prompts: isolating the core claim, identifying the strongest evidence, understanding the values driving the position, and assembling all of that into the complete argument. From there, the worksheet helps us separate the parts of the opposing argument we accept from the parts we don’t, so that by the end, we’ve moved from a vague sense of disagreement to a precise understanding of what we believe and why. The final step turns the same lens on our own view. Each stage includes writing space and prompts designed to make the process concrete rather than abstract.
The Compounding Effect
Each of the steps above works on its own, applied to a single argument in a single moment. But the practice compounds. Every time we steel-man an opposing argument, we get a little faster at finding the strongest version rather than reacting to the surface-level version, and a little more comfortable sitting with a position we disagree with long enough to understand it on its own terms. The positions we arrive at on the other side of that process get progressively more refined, because each one has been tested against stronger opposition than the last.
Over time, steel-manning changes the default way we approach disagreement. Instead of hearing an opposing argument and immediately looking for the flaw, we develop a habit of asking, “What could the strongest version of this look like?” That shift is small, but its effects are wide. Conversations become more productive because we’re engaging with what people mean rather than the worst interpretation of what they said. Disagreements get more focused because we’ve already identified exactly where we diverge, rather than arguing in circles around a vague sense that the other side is wrong. And, hopefully, the people we’re engaging with are more likely to hear our counterarguments because we’ve demonstrated that we heard theirs first.
There’s also something that changes about how we hold our own positions. After practicing steel-manning for a while, we tend to hold our views with more precision and less defensiveness, because we’ve already reckoned with the best arguments against those views. We’ve developed a clear sense of where we stand and the specific reasons we stand there, because we’ve located the exact point of divergence between our position and its strongest opposition. That’s a different kind of confidence than the kind that comes from never having engaged with a serious counterargument. It’s more durable, because it’s already been stress-tested.
We live in an information environment where some of the hardest arguments to evaluate aren’t the ones built on lies. They’re the ones built on carefully selected truths, arranged with precision to produce a specific emotional and intellectual response. Checking whether the individual facts are real is a good and necessary starting point. But the skill that protects us when the facts are real, when every citation checks out and the conclusion still leads us somewhere misleading, is the ability to engage with why the argument is persuasive in the first place. Steel-manning builds that ability. And unlike a single act of fact-checking, which addresses one claim at a time, the habit of engaging with the strongest version of opposing arguments reshapes how we think about every argument we encounter going forward.
A lie runs halfway around the world before the truth laces up its sneakers. Fact-checking propaganda (lies) is a fool’s errand.
I like your “steel-manning” approach as a way of engaging in a good-faith debate. I always like to start a conversation with finding something to agree with. Builds trust and openness.
Your essays are terrific, Hana. Your rising stats on this platform are well earned! 👏
Love this, thank you. Civility = steel-manning?