How to Argue

Before any argument, one question decides everything after it: are you trying to win, or trying to end up correct? In a genuinely zero-sum contest — a negotiation over the one hat — pressing your case regardless of the other side’s position can be rational. In every other case, the goal of communication is learning, not winning, and everything below follows from taking that seriously.

First, find the actual disagreement, because most arguments are fought over a misconception of the other side rather than the other side. The tempting version is the straw man — answer the weakest reading of their view and declare victory. The productive version is the steel man: state their position so well that they endorse the statement, and only then answer it. Half the time this step ends the argument by revealing there wasn’t one; the rest of the time it locates the true disagreement, which is the only place progress was ever available.

Second — and this is the citizens’ own move — when the person across from you shares the common ground, say so to yourself before speaking. Both of you hold the values; neither of you wants the suffering of the other; therefore what you have is a disagreement about direction — about depths, weights, and probabilities, exactly the three honest sources from Part one. Locating which one transforms the conversation: a weights dispute argues priorities, a probability dispute argues evidence, a depth dispute teaches — and none of them requires an enemy. This is also, not coincidentally, the same dispute model this book’s legal doctrine will run in Part four: argue over the facts, their ranges, and their weights — never over the character of the person you are debating. What works in a courtroom works at a dinner table, and for the same reason. The common ground lets you see the good in the other, not an enemy.

Third, manage the machine you argued with, because stage one never stops applying. Opposing views trigger the body — the impulse arrives before the reasoning does — and the person who has not corrected their own thinking first will mistake their defensiveness for insight. The ancients put it flatly: the rational faculty is the distinguishing one, and failing to govern the body with it is a failure to communicate that everyone is guilty of, and its how often and to what degree that we work on. So the working order is fixed: understand them, even where you disagree; then try honestly to knock down your own belief — and understand what the attempt actually does, because it works four ways at once. It bridges: attacking your position from their side forces you to build their view inside your own head. It sorts: the places your belief resists the attack are your actual objections, finally separated from the reflexive ones — you learn what your problem with their position really is. It signals: the other person watches you try their side on, which tells them you are reachable and shows them exactly what they would need to overcome. And it obliges: an opponent watching you test your own position honestly is pushed to defend theirs the same way. All four run on one condition — sincerity. Performed openness convinces few, the exercise pays only if you are genuinely willing to lose it. Then reason together with learning as the goal. Recall Jake and the bag of oranges from Chapter Eleven — the gift, the joke, the courtship, and then a whole town in which oranges mean something they mean nowhere else. That was social truth caught in the act of being born, and the deepest reason arguments miss each other sits underneath all three sources: the words themselves, made by communities, real by agreement, portable nowhere without it. And notice what just happened as you read: the word oranges arrived already loaded, carrying a meaning it holds only among the readers of this book, because a convention formed between these pages and you were in the town where it formed. That is language, whole. Every word you have ever read or spoken is a bag of oranges — freight assigned by some community, somewhere, at some time — which is why this book defined its terms as it went, and why honest communication begins with the humility that the other person’s town may assign your words different freight. We manage at all because both sides assume cooperation — a finding with a name too few are ever taught: Grice’s maxims, the observation that conversation runs on four standing assumptions extended on credit. Say as much as is needed and no more; say what you believe true; say what is relevant; say it clearly. Every listener assumes the speaker is following them, which is why a broken maxim reads as a message in itself — the too-short answer heard as coldness, the truthful-but-irrelevant reply heard as evasion, the excessive detail heard as hiding something. Most misunderstanding is that cooperative assumption meeting two different dictionaries — and most of the tedium this chapter’s precision would otherwise demand is spared by the maxims doing their silent work: we do not spell out every probability and factor because relevance and quantity are carrying them.

Humanity operates in thousands of languages whose structures resist translation, with one serving as the international standard regardless of the means by which it got the job. It works, the way inherited things work. But ideally for the most efficient ability to communicate, a new language founded in science — regular, learnable, built for precision rather than accredited by conquest — might be worth the inconvenient effort. The extension register holds here at full width: whether any designed language can outrun the network effects of the inherited ones is a factor question with a discouraging history, and the vote on it is cast one speaker at a time. The relationship being claimed is smaller and firmer: a civilization that runs on cooperation runs on communication, and its common tongue — whichever wins — is infrastructure, owed the same deliberate care as the roads. The toolkit, compactly. An argument is premises and a conclusion: the premises must be true and the conclusion must actually follow — and when the premises are probabilities, the conclusion stays a probability, however confident it feels. Do not lie or the whole exercise is ruined at the source. Prefer the dialectic to the lecture — exchange, counter, counter again — because minds change by participating, not by receiving. And know the five shapes an argument takes, each with its warning label: deduction (true premises, entailed conclusion — only as good as the premises); induction (probabilities combined — the conclusion is a probability, forever); abduction (eliminate the impossible, keep the remainder — but it is hard to know you listed every possibility); analogy (compare situations to borrow understanding — but every situation carries its own variables); and reductio ad absurdum (slide the position along its own scale — to other situations, other magnitudes, usually the extreme — and see whether the idea still works at each point; a principle that holds only at the one spot it was proposed for was never a principle, it was a preference wearing one. Legitimately, which means the tested extreme must genuinely follow from the position, not be strapped to it).

The conduct rules, which win more arguments than logic does. Stay on the subject; analyze what was said about the topic, not the person who said it — the opponent is separate from the debate unless the opponent is the debate’s subject. When emotion starts interfering, in them or in you, find common ground and find things to genuinely agree with; agreement is not concession, it is traction, and a defensive person cannot hear anything until the defense stands down. A good argument leaves no contradictions unexplained — hunt yours before they do. Admit it when you are wrong: it is the entire purpose of debating, and nothing purchases credibility faster. And remember the reciprocity underneath all of it — demonstrating that your mind can change is a precondition of theirs changing, because no one disarms in front of someone who never will.

One warning from the practical world, because good-faith rules can be farmed like unlimited forgiveness was. In any repeated negotiation, watch for the asymmetry where one side, valuing compromise, gives a little each round, and the other side simply asks again — a ratchet that converts one party’s virtue into the other’s gains, with the compromiser drifting toward the demander’s position and calling the drift moderation. The remedy is the trust engineering from Chapter Fifteen, applied to discourse: cooperate first, reciprocate always, and price non-reciprocity into the next round. Arguing to learn does not mean arguing to lose.

Carried forward: disagreement rebuilt as a shared instrument instead of a contest. The first question — winning, or ending up correct — decides whether an argument can produce anything; the steel-man, the three honest sources, and the five shapes locate what kind of dispute you are actually in; and the machine you argue with is governed, the ratchet priced. You now hold the tools to examine another mind and disagree with it without breaking anything. What remains of stage two is the boundary where words run out — the next chapter.