The Architecture of Arguments

The claim is your destination—where you want the audience to end up. A strong claim is specific, falsifiable, and clearly stated. A weak claim is vague, unfalsifiable, or buried in qualifications.

Evidence is the raw material of argumentation—the stuff you point to as support for your claim. Good evidence is specific, verifiable, and relevant. Bad evidence is vague, unverifiable, or tangential.

Types of evidence include: statistics and data, expert testimony and studies, historical examples, analogies to similar situations, personal experience or anecdotes, and logical principles or established facts. Not all types carry equal weight—peer-reviewed studies generally trump anecdotes, but a compelling story can move an audience that statistics leave cold.

The warrant is the most overlooked—and most important—part of any argument. It's the logical bridge between your evidence and your claim. Without it, you're just stating facts and conclusions without explaining how they connect.

Impact is what gives your argument weight. Without impact, even a perfectly constructed argument feels trivial. “So what?” is the question impact answers.

Statistics seem objective—just numbers, right? But statistics are chosen, framed, and presented by people with agendas. Learning to spot the difference between good and bad statistical evidence is a crucial skill.

The bad version is missing everything that matters: Who conducted the study? When? With whom? How was the question asked? Can it be verified? The good version answers these questions, making the evidence actually useful.

Use this checklist to quickly evaluate any piece of evidence:

1. 1Recency: Is this evidence current, or has the situation changed?
2. 2Accuracy: Is the source reliable? Can it be verified?
3. 3Validity: Does this evidence actually measure what it claims to measure?
4. 4Enough: Is the sample size adequate? Is this cherry-picked?
5. 5Neutral: Does the source have conflicts of interest or obvious bias?

Most warrants are implicit—people don't state them because they seem obvious. But what seems obvious to one person may be questionable to another.

1. 1State the evidence and the claim clearly
2. 2Ask: "What has to be true for this evidence to support this claim?"
3. 3Make the assumption explicit
4. 4Evaluate whether that assumption is actually valid

Impacts come in two flavors: concrete and abstract. Both matter, but they work differently.

Concrete impacts are tangible outcomes: lives saved, money gained or lost, time wasted, jobs created. They're easy to understand and often easy to quantify. “This policy will save 50,000 lives annually” is a concrete impact.

Abstract impacts involve values and principles: justice, freedom, equality, dignity, autonomy. They're harder to quantify but often matter more to people. “This policy violates the fundamental right to privacy” is an abstract impact.

Abstract impacts are powerful but slippery. “This harms justice” sounds important but feels vague. The solution: connect abstract values to concrete experiences.

The second version names the abstract value (democracy) but immediately grounds it in what it means for a real person. It transforms “undermines democratic principles” into “you lose your vote.” The impact suddenly feels personal.

Most debates involve competing impacts on both sides. Your job isn't just to show that your side has impact—it's to show that your impact outweighs theirs. Three dimensions matter:

1. 1Magnitude: How big is the impact? More lives, more money, more freedom at stake = larger magnitude.
2. 2Probability: How likely is the impact? A 90% chance of a small harm may matter more than a 1% chance of a catastrophe.
3. 3Timeframe: When does the impact occur? Immediate harms often feel more urgent than distant future risks.

Arguments without clear impact often sound like this: “Studies show X is true.” Okay... so what? The audience needs to know why X being true matters to them, to their values, to the world they care about. Without that connection, even true claims feel like trivia.

Before you make any argument, ask yourself: “If I convince them, what changes?” If you can't articulate meaningful stakes, your argument isn't ready.

“X causes Y.” The most common warrants in policy debates. Vulnerable to: correlation vs. causation (Does X really cause Y?), reverse causation (Maybe Y causes X), common cause (Maybe Z causes both), multiple causes (X is one factor among many), and broken mechanism (How exactly does X lead to Y?).

“X is like Y, so what's true of Y is true of X.” Analogies can be powerful but they're always imperfect. Vulnerable to: false equivalence (Are X and Y really similar in relevant ways?), key differences (What breaks the analogy?), and overextension (Does the analogy hold for this specific aspect?).

“Expert/authority says X.” Appeals to expertise can be legitimate—or fallacious. Vulnerable to: credential questions (Is this person actually an expert here?), bias (Conflicts of interest?), consensus (Do other experts agree?), and context (Quoted accurately?).

“The data shows X.” Data-driven arguments seem objective but are full of interpretive choices. Vulnerable to: cherry-picking (Representative sample?), methodology (How was data collected?), interpretation (Other ways to read this?), and significance (Meaningful difference or noise?).

“X is right/wrong because of principle Y.” Unavoidable in important debates. Vulnerable to: competing values (What when values conflict?), consistency (Does this apply equally elsewhere?), foundation (Why accept this principle?), and application (Does this really apply here?).