Principle of charity

Two readings of “nobody”

At a team meeting, Jordan says: “We should stop using the shared calendar. Nobody keeps it up to date.”

Alex: I update it every day, so that’s just not true.

Morgan: I think you mean most people don’t, so nobody can rely on it. That’s true for meeting rooms, but vacation dates are almost all in there. Could we keep it just for those?

Read literally, “nobody” is refuted by a single counterexample, and Alex supplies one. But Jordan’s proposal never depended on literally nobody: it depends on the calendar being unreliable. Alex has refuted the most extreme reading of “nobody”, which is more than Jordan’s argument needs, so the proposal still stands. Morgan answers the reading the argument actually depends on, which is why her reply can change the plan. Jordan would also recognize Morgan’s version as what they meant, which is the check that keeps charity honest.

Charity that goes too far

Riley joins in: “I think what Jordan’s really getting at is that our whole approach to scheduling is broken, and we need a proper project-management system.”

Jordan said nothing of the kind. Riley’s version may even be a better argument, but it’s Riley’s, not Jordan’s. It also has a side effect: Alex’s and Morgan’s replies now look as if they missed the point, when they answered what was actually said. That’s the problem Scott Aikin and John Casey call the iron man: improving an argument can misrepresent both the arguer and their critics.

The same name covers two related but different ideas. Marcin Lewiński describes the principle as a “loan-concept” that argumentation theory borrowed from the philosophy of language, and reports Trudy Govier’s judgment that there seems “only a tenuous connection” between the two.

Interpretive charity (philosophy of language) is a condition for understanding anyone at all. W. V. Quine, discussing how to translate the logical words of an unfamiliar language, wrote that “one’s interlocutor’s silliness, beyond a certain point, is less likely than bad translation—or, in the domestic case, linguistic divergence.” If a translation has people cheerfully asserting contradictions, the translation is probably wrong. Donald Davidson made charity central to his theory of radical interpretation, the task of understanding a language from scratch with no translator. Jeff Malpas’s encyclopedia entry on Davidson summarizes the principle as advice “to interpret speakers as holding true beliefs (true by our lights at least) wherever it is plausible to do”, and notes that it “cannot be rendered in any completely precise form”. It doesn’t assume people are never wrong: particular attributions of belief can always be revised. As Lewiński quotes Davidson, the method “is not designed to eliminate disagreement”; disagreement only makes sense against a large background of agreement.

Argumentative charity (critical thinking and argumentation theory) is advice for evaluating arguments. Its classic statement, which Lewiński quotes, is from Michael Scriven’s textbook Reasoning (1976): “The Principle of Charity requires that we try to make the best, rather than the worst, possible interpretation of the material we’re studying.” This is the version taught in critical thinking courses, and the one that matters for spotting a straw man. Unlike interpretive charity, it isn’t needed to understand the words at all; it’s a choice about which of several understandable readings to answer.

Katharina Stevens, introducing a 2025 journal issue on the subject, sums up three kinds of argument for argumentative charity, all present in Scriven’s early account:

• Prudential. Objections to a weak reading are easy for the other person to escape by clarifying. Answering the strong reading finds objections they can’t so easily avoid.
• Epistemic. Taking the best version of each side seriously is how the best reasons come to light.
• Moral. Reading someone charitably is a matter of fairness or goodwill toward them.

Charity also changes what counts as an error. Read uncharitably, “you don’t follow your own advice” sounds like a claim that the advice is false, a Tu quoque; read charitably, it’s often a reasonable doubt about whether to trust the adviser. Before naming a fallacy, it’s worth checking whether a more plausible reading commits it at all.

Steelmanning means restating an opposing argument in the strongest form you can, stronger than the way it was put, before responding. It’s an informal term, formed as the opposite of “straw man”, but it has entered the scholarly literature. R. A. J. Shields (2025) notes that “it has become commonplace to ‘steelman’ or ‘iron man’ another’s arguments” in philosophy, and defines steelmanning as making an argument “stronger than its original author either intended or communicated”. Argumentation theorists Aikin and Casey use the term iron manning for the same practice.

That definition marks the difference from charity. Charity chooses among readings the words plausibly allow. Steelmanning can go further and build an argument the author didn’t give. Both have real uses: Aikin and Casey point out that the most robust opponent is the most useful one if you want the truth, and that a teacher may need to improve a student’s half-formed point to show how it could work. The risk is in presenting the improved version as the other person’s own.

• Over-charity invents arguments. Ralph Johnson, as quoted by Lewiński, objected to reconstructions that quietly repair bad reasoning: “Benefit of the doubt, si; whitewash, no! [...] This is not charity; it is welfare!” Govier argued for moderate charity instead, and Stevens summarizes her worry: too much charity can turn an argument into something the interpreter finds convincing but that has little to do with what the arguer offered. Lewiński reports a related criticism by Douglas Walton and C. A. Reed, that routinely supplying charitable unstated premises can produce a straw man of sorts, since the arguer is credited with a premise they never endorsed.
• It pulls against accuracy. Shields notes that the author of an argument can reject a steelmanned version as not really theirs. Stevens reports that Geoffrey Goddu, in the same 2025 issue, argues most of what is credited to charity is better explained by principles of accuracy and respect in interpretation.
• “Strongest” is ambiguous. Lewiński points out that the easiest premise to defend isn’t always the reading that makes the best argument. “Some of us got tickets” is easiest to defend if “some” means “at least one”, but then it supports much less. Charity doesn’t say which to choose.
• Charity to one side can be unfair to the other. In a real disagreement, Lewiński argues, an analyst who gives one party the most defensible reading of a disputed statement gives the other party a weaker version of their objection. He calls this the paradox of charity, and proposes limiting charity to cases of irresolvable interpretive doubt, extended to one side or the other depending on the context. Aikin and Casey make the same point from the other direction: iron-manning one arguer can straw-man their critic.
• It isn’t always owed. Johnson, as Lewiński quotes him, limited charity to “a fully expressed argument” from “a serious arguer” on “a serious matter”. Aikin and Casey argue that improving an argument is a mistake when it isn’t likely to be relevant or successful, when there isn’t time for it, or when taking the speaker seriously encourages more badly formed arguments. Their paper is titled “Don’t feed the trolls”.
• It can be used as a shield. Stevens describes an argument by Aikin and Casey, in the same issue, that arguers tend to impose stronger-than-warranted readings on their own arguments after the fact, and then accuse anyone who reads them otherwise of being uncharitable. A demand for charity can become a way of avoiding criticism, much like retreating to a safer claim in a Motte and bailey.

• Among readings that fit what was said and the context, answer the strongest. Don’t pick a reading the words don’t support, in either direction.
• Check with the person when you can. “Do you mean nobody, or most people?” settles what charity can only guess at.
• Say when you’re improving an argument. “The best version of this I can think of is...” lets you engage the strongest case without putting words in anyone’s mouth.
• The test: would the person recognize your version as what they meant, and does your reply still address the reading the argument actually depends on?