The reasoning vocabulary,
defined.

An ad hominem fallacy rejects an argument by attacking the person making it rather than the substance of what they said. The attack may be entirely accurate (the speaker may indeed be hypocritical, unlikeable, or conflicted) but none of that tells you whether their argument is sound. The argument has to be engaged on its own merits or left alone.

An appeal to authority claims that something is true because a notable or credentialed person said so — without engaging the substance of their reasoning, or checking whether the claim is inside the scope of their expertise. Expertise is evidence; it is not a substitute for evidence. The relevant question is always what the authority is saying and how they arrived at it.

An appeal to emotion persuades by evoking fear, pity, pride, or outrage in place of evidence or reasoning. Emotions are a legitimate part of how humans make decisions; the fallacy is using them to substitute for an argument, not to accompany one. The question to ask is: if I strip the emotional language out, is there still an argument underneath?

An appeal to popularity concludes that a claim is true because many people believe it, or that a thing is good because many people use it. Consensus can be useful evidence — it often reflects accumulated reasoning, but it is not itself an argument. Most people believed the earth was flat once; the belief was widely held and wrong.

The argument from ignorance treats absence of evidence against a claim as evidence for it — or vice versa. "No one has proven it false" is not an argument that it is true. The absence of disproof is a gap in knowledge, not a positive reason to believe. The burden of proof sits with the person making the claim, not with the listener.

To beg the question is to assume the conclusion within the premise. The argument appears to make a case, but it only restates its starting claim in new words. (In English this phrase is often misused to mean "raises the question"; the original technical meaning is this narrower, logical one.)

The burden of proof is the obligation of the person making a claim to produce evidence for it. It does not sit with the listener to disprove it. Shifting this burden, demanding that the audience or the opponent show why a claim is false rather than demonstrating why it is true, is one of the most common rhetorical moves in public argument, and closely related to the argument from ignorance.

Cherry-picking is selectively presenting only the data, examples, or quotes that support a conclusion, while omitting those that do not. It is often technically accurate (the chosen evidence is real) but the picture it paints is misleading. The test is not "is this true?" but "would the story change if the missing evidence were added?"

Confirmation bias is the cognitive tendency to notice, seek out, and remember information that confirms existing beliefs, while discounting or ignoring information that contradicts them. Unlike a fallacy, which is a structural problem in an argument, this is a structural problem in the reasoner. It operates silently; by the time you notice it, your priors have already filtered the evidence.

A false dichotomy frames a complex situation as a binary choice when more options exist. It is often deployed to force a listener into the speaker's preferred side — accept this, or accept the unacceptable alternative. The real world usually contains a third, fourth, or tenth option.

False equivalence treats two things as morally or factually comparable when they differ in important ways — scale, intent, severity, or kind. It is the close cousin of whataboutism: where whataboutism deflects, false equivalence flattens. Both make the reasonable move of comparison do work it cannot bear.

The Gish gallop is a debate tactic in which a speaker overwhelms an opponent with a rapid barrage of claims, each of which would take time to refute individually. The sheer volume creates the impression of a strong argument — no single claim gets pinned down before the next arrives. Named after the creationist debater Duane Gish, who was known for the technique.

A hasty generalization draws a broad conclusion from a sample too small, too biased, or too unrepresentative to support it. It is the everyday form of the more technical statistical error: n=1 becomes n=everyone. Anecdotes can illustrate an argument; they cannot prove a generalisation.

A loaded question presupposes something contested — so that answering it at all concedes the premise. The classic example is "have you stopped beating your wife?": a yes admits past beating, a no admits ongoing beating, and there is no clean answer that does not accept the presupposition. The correct response is to reject the frame, not the options.

A motte-and-bailey is a rhetorical two-step. The speaker advances a bold or provocative claim (the "bailey"), and when challenged, retreats to a milder, more defensible version (the "motte") — then, once the challenger backs off, quietly returns to the bailey. The image comes from medieval fortification: the castle keep on high ground (motte) was indefensible in peacetime, so life happened down in the lower yard (bailey) until attackers arrived.

A non sequitur is a conclusion that does not follow from the premises offered. The Latin translates directly as "it does not follow." Unlike more specialised fallacies, this is the generic name for any argument where the bridge between evidence and conclusion is broken — even when both the evidence and the conclusion are individually true.

The no true Scotsman fallacy is an appeal to purity. A counterexample to a generalisation is dismissed not by engaging the counterexample, but by redefining the category to exclude it. The generalisation then becomes unfalsifiable, and therefore unhelpful.

Phronesis is the Aristotelian virtue of practical wisdom — the capacity to evaluate information, recognise your own biases and assumptions, and judge what to do in particular situations. It is distinct from episteme (theoretical knowledge) and techne (craft skill). Aristotle treated it as the virtue that holds the others together. It is also the namesake of this project.

The post hoc fallacy assumes that because B followed A, A must have caused B. Sequence is not causation. The fallacy is seductive because many real causal relationships do look like this — but many do not, and the resemblance is the trap.

A red herring is an irrelevant topic introduced to divert attention from the actual argument under discussion. The name comes from the practice of dragging a strongly-scented smoked fish across a trail to throw hunting dogs off scent. In argument, the move is to change the subject in a way that looks like engagement.

A slippery slope argument claims, without evidence of the intermediate steps, that a modest first action will inevitably lead to an extreme outcome. A slippery slope is only fallacious when the causal chain is asserted rather than shown. Sometimes slopes really are slippery; the point is to distinguish demonstrated chains from asserted ones.

Steelmanning is the practice of reconstructing and engaging with the strongest possible version of an opposing argument, rather than the weakest. It is the opposite of strawmanning. The discipline is: state your opponent's view so well that they would recognise it, then respond to that. It is slow, costly, and the single best antidote to most of the fallacies on this page.

A strawman argument misrepresents an opponent's position in a weaker or more extreme form, then refutes the distortion rather than the real view. The resulting victory is over something no one actually said. The antidote is steelmanning: demand that your account of the other side is one they would recognise.

Tu quoque ("you too") deflects a criticism by pointing out that the accuser is guilty of the same thing. Even where the accusation of hypocrisy is accurate, it does not address whether the original criticism is valid. A hypocrite can still be right. The political variant is whataboutism.

Whataboutism is a specific form of tu quoque used in political discourse. It deflects a criticism by raising an unrelated grievance against the accuser's side. Unlike standard tu quoque, whataboutism does not even require the deflection to be hypocritical — just to be another complaint, loud enough to change the subject.