Straw man

Exaggerating a proposal

Jordan: I think we should cut the weekly team meeting to every other week. Most of what we cover could go in a shared update.

Casey: So you don’t think we need to talk to each other at all? A team that never meets falls apart.

Jordan proposed meeting less often, not never. Casey’s point about teams that never meet may be true, but it answers a proposal nobody made. This is the clear-cut case: the distortion is large enough that Jordan will object at once.

Quietly turning “most” into “all”

Priya: For most weeknight cooking, dried herbs work about as well as fresh ones.

Sam: Really? You think pesto made with dried basil is as good as the real thing?

Priya’s claim had a limit (“most weeknight cooking”) and Sam’s reply drops it, then picks the one dish where the unlimited version fails most obviously. Nothing Sam says is false, and the reply sounds like a fair test. But pesto is exactly the kind of case Priya’s “most” left room for, so it doesn’t count against what she said. Dropping a quantifier (a word like “some”, “most” or “usually” that sets how far a claim reaches) is one of the easiest ways to build a straw man without noticing.

Answering the weakest advocate

A running blog post is titled “Why the case for minimalist shoes falls apart.” It quotes a forum commenter who claims that cushioned shoes cause every running injury and that nobody got hurt before shoes were invented, shows why both claims are false, and concludes that there’s no reason to consider minimalist shoes.

The commenter’s words are quoted accurately, so this isn’t a distortion of anyone’s statement. The problem is selection: the post treats the least careful advocate as if they were the whole case, and never looks for a narrower, more careful argument for minimalist shoes. Refuting the weakest version and presenting it as the best available is the form Aikin and Casey call the weak man (see Variants).

• Straw man (representation form): misstating what a particular opponent said, usually by exaggerating it or removing its qualifications. “You want less screen time for the kids? So you want to throw out the TV.”
• Weak man (selection form): accurately quoting a weak version or a poor spokesperson and implying it is the strongest case the other side has. Talisse and Aikin (2006) called this the selection form; Aikin and Casey (2011) call it the weak man.
• Hollow man: attacking a position that nobody in the discussion holds, often attributed to unnamed opponents. “Some people think you should never back up your files. They’re wrong.”

Aikin and Casey argue that each of the three can occur in non-fallacious cases as well, which is why the conditions below matter.

The opposite habit, restating an opponent’s view in its strongest form before criticizing it, is often called steelmanning, and is related to the principle of charity: when a statement can be read more than one way, assume the most reasonable reading.

The term may go back to Isaac Watts’s Logick (1724), which describes disputers who dress up their opponents’ opinions as “images of straw” and then triumph over them.

A reply isn’t a straw man just because it restates the other view, simplifies it, or finds a problem with it.

• A blunt but accurate summary. Putting someone’s view in fewer, plainer words is fine as long as the scope and the qualifications survive.
• Drawing out real consequences. Showing that a position, as stated, commits its holder to something they’d reject is a legitimate objection, called a reductio (short for reductio ad absurdum, “reduction to absurdity”). It becomes a straw man only if the consequence depends on words they didn’t use.
• Criticizing a weak version that people actually hold. If a weak version is widespread or is the one being advocated in front of you, it’s fair to answer it. What isn’t fair is presenting it as the best the other side can do. Talisse and Aikin point out that nobody is obliged to answer every objection, but someone who claims to have defended a view should take on the best objections they can find, not only the weakest.
• Checking what someone means. “Do you mean all weeknight cooking, or just most?” is a question, not a refutation, and it gives the other person the chance to correct it.
• Openly discussing a hypothetical view. Examining a position no one present holds is fine if you say that’s what you’re doing, rather than attributing it to someone.

The test: would the person you’re answering accept your version as a fair statement of their view, and if you chose one version among several, did you say so?

Following a rule to where it leads

Alex: Anything in the shared fridge without a name on it is fair game.

Morgan: Then the milk I bought yesterday is fair game, and so is the jar of mustard, since neither has a name on it. Is that what you mean?

Morgan’s reply sounds like an exaggeration, but every step follows from Alex’s rule as Alex stated it. Morgan hasn’t changed the rule; she’s shown what it covers, and asked. This is the real consequences condition above: if Alex meant something narrower (leftovers, say), the objection shows the rule needs rewording.

A short summary that keeps the claim intact

Jordan sends a long email arguing that the weekly team meeting costs more time than it saves, with a breakdown of the agenda, and proposes replacing it with a written update. A colleague replies: “So, in short, you want to replace the weekly meeting with a written update. I disagree, because the meeting is where problems get noticed early.”

The summary drops nearly all of Jordan’s detail, but it keeps what Jordan actually proposed, and Jordan would sign it. The rebuttal then answers that proposal. This is the blunt but accurate summary condition: shortening a view is only a straw man when the shortened version is a different claim.

A position with limits and qualifications is harder to refute than one without them, so every restatement is pulled toward the version that is easiest to argue with. Qualifiers like “most” or “in this case” are also easy to drop without noticing when retelling someone’s view, which means a straw man can be built in good faith.

It persuades when the audience doesn’t catch the substitution. Talisse and Aikin describe two routes: the representation form works on an audience that isn’t tracking what the opponent actually said, and the selection form works on an audience that doesn’t know the better arguments on the other side, even if it has followed the quoted ones closely.

There is a little experimental evidence on when it persuades. In two experiments, Bizer, Kozak and Holterman (2009) had participants read statements from two people competing for a public office, one of whom did or didn’t use a straw man argument. The technique was more persuasive when participants were led to believe the office mattered little to them personally, and less persuasive among participants low in need for structure (a preference for simple, clear-cut structure in information). The authors suggest a straw man may work when people aren’t motivated to think the arguments through, and may fail or backfire when they are.

• Restate before you rebut. Say back what you take the other view to be, and ask whether that’s right. If they correct you, answer the corrected version.
• Check the quantifiers and qualifiers. Compare “some”, “most”, “usually” and “in this case” in the original with what’s being attacked.
• When it’s aimed at you, restate your actual claim plainly instead of defending the exaggerated one. It’s easy to end up arguing for a position you never held.
• When you’re the one arguing, ask whether the version you’re answering is the one its most careful defenders would give, and if it isn’t, say so.