Chapters

An argument begins when someone makes a claim — a statement offered as true — and gives support: reasons, evidence, or examples meant to make the claim believable. Before judging any argument, first find the claim, then find what is holding it up.

Deductive reasoning uses rules, definitions, or necessary relationships to reach a conclusion. If the premises are true and the argument is valid, the conclusion must be true — there is no maybe about it.

Inductive reasoning moves from observed cases to a broader pattern or a likely next case. Good induction makes its conclusion probable, never certain — more cases, more varied cases, and no known exceptions all strengthen it.

Abductive reasoning starts from a surprising or puzzling observation and infers the explanation that best accounts for it. The conclusion is a hypothesis — the strongest candidate among rivals — held provisionally until better evidence arrives.

Analogical reasoning concludes that two things probably share a further property because they are alike in relevant, known ways. Its strength depends on whether the similarities actually matter to the conclusion — and on whether any difference breaks the comparison.

Causal reasoning is about what makes what happen: identifying causes, tracing effects, and proposing mechanisms. Good causal claims rest on more than sequence or coincidence — they point to a process that connects cause to effect, and they survive checks against rival causes.

Statistical reasoning draws conclusions from rates, proportions, samples, and probabilities — moving from how often something happens to how likely it is here. Its strength lives in the details: how big the sample is, how it was gathered, and whether the rate actually applies to the case at hand.

Bayesian reasoning updates how confident you should be in a belief as evidence arrives: start from how likely it was beforehand, then shift in proportion to how strongly the new evidence favors it over the alternatives. Neither the starting rate nor the new clue settles it alone — the skill is combining them.

Authority reasoning accepts a claim because a credible source vouches for it. Most of what anyone knows arrives this way, and it is perfectly good reasoning — as strong as the source's expertise in the relevant domain, their track record and incentives, and the agreement of other qualified sources.

Expected-value reasoning chooses an action by listing the possible outcomes, assigning each a probability, and multiplying that probability by the outcome's magnitude — then comparing the weighted totals across options. The center of gravity is always a decision under uncertainty: not 'what will happen?' but 'given the odds and the stakes, what should I do?'

Precedent reasoning applies the treatment established in a past decision to a sufficiently similar current case. Three moves are required: identify what was decided before, establish that the current situation is relevantly similar, and extend the prior treatment. Its force comes from continuity and settled practice — not merely from structural resemblance between the two cases.

Practical reasoning moves from a goal and beliefs about means to a decision to act: I want G; doing A is the best available way to get G; so I should do A. Judging it means checking all three joints — is the goal worth having, does the means really deliver it, and at what cost compared to the alternatives.

Moral reasoning argues about what ought to be done, deriving its verdict from values, duties, rights, or principles rather than from facts alone. Facts enter as inputs — what happened, who is affected — but no pile of facts yields an 'ought' until a value joins them.

Comparative reasoning ranks options against each other along shared, named criteria — better, worse, or equal — and shows its work by saying which criteria count and how the options fare on each. Its conclusion is a ranking, not yet a decision.

Conductive reasoning weighs several independent reasons — often on both sides — to reach a verdict that no single reason could carry alone. Each reason counts on its own; remove one and the others still count; the conclusion rests on the balance.

Definitional reasoning settles a question by making a word's meaning precise: what the term requires, and whether the case at hand meets those requirements. Many disputes that look like arguments about the world are arguments about a word — and dissolve the moment the word is pinned.

Conceptual reasoning works out what follows from the structure of an idea itself — what promising entails, what forgiveness excludes, what a gift must be free of. Where definitional reasoning pins a word's usage, conceptual reasoning unfolds the idea behind the word, and its conclusions hold in any vocabulary.

Classification reasoning places a case into a category and draws conclusions from what is known of the category's members: identify the marks, sort the case, inherit the traits. Its strength hangs on two hinges — whether the case truly belongs, and whether the inherited trait truly runs through the category.

Part–whole reasoning moves between properties of parts and properties of the whole — from sound timbers to a sound hull, or from a heavy cargo to heavy crates. Some properties survive the trip in one direction or both; some do not survive it at all. The reasoning is only as good as the check that this property, this trip, travels.

Dialectical reasoning refines a position by taking objections seriously: state the view, meet the strongest counterargument fairly, and defend or amend in response. Its product is not victory but a better position — one that carries its objections' scars as credentials.

A thought experiment constructs a hypothetical scenario — a deliberate 'what if' — to test whether a principle holds, reveal its implications, or expose its limits. The scenario makes no claim about the real world: you build it, reason inside it consistently, and read what the outcome tells you about the principle under test. The hypothetical is the instrument; the principle is the subject.

Reductio ad absurdum proves a claim by assuming its opposite and reasoning from that assumption until a contradiction or manifest absurdity follows — which shows the assumption must be false, and therefore the original claim true. Three moves: assume NOT-P, reason from it consistently, arrive at something impossible; since NOT-P cannot hold, P must.