Building and Evaluating Normative Arguments About Laws and Institutions

What You Are Doing When You Make a Normative Argument

A normative argument about laws and institutions answers a practical question: what should we do together, through rules and organizations, and why? In this capstone chapter you will build a complete argument that moves from a clarified concept (such as justice, freedom, or legitimacy) to principles, then to institutional design, and then through rounds of objections and revisions. The aim is not to “win,” but to produce a proposal that is justified, transparent about trade-offs, and resilient under disagreement.

The basic architecture (use this as a checklist)

• Concept: define the value at stake (e.g., justice) and what counts as a violation of it.
• Principles: state general normative claims that guide evaluation (e.g., “laws must treat persons as equals”).
• Institutional link: show how principles imply constraints on laws, policies, and enforcement.
• Objections: anticipate the strongest challenges (moral, empirical, and institutional).
• Revisions: adjust the concept, principles, or institutional design to handle objections without abandoning the core commitment.
• Trade-offs: state what you are sacrificing, why it is acceptable, and what safeguards reduce the cost.

Step 1 — Select a Topic That Forces Institutional Choices

Choose a topic where the disagreement is not only about personal morality but about collective rules. Good capstone topics have (a) clear institutional levers, (b) predictable objections, and (c) measurable risks.

Topic selection filter

Practical step-by-step: write your “policy question” in one sentence

1. Start with an action verb: adopt, prohibit, subsidize, mandate, decentralize, constitutionalize.
2. Name the institution: parliament, agency, court, school system, central bank, police oversight board.
3. Specify the mechanism: tax, voucher, licensing, audit, independent review, automatic enrollment.
4. Add one constraint: budget cap, sunset clause, rights-protecting limit, or review requirement.

Template: “The [institution] should [action] by [mechanism], subject to [constraint], because [core value].”

Step 2 — Clarify the Concept You Are Using (Without Hiding the Hard Parts)

Concepts like justice, freedom, equality, and legitimacy are not self-interpreting. Your first task is to make your concept action-guiding: it must tell you what counts as an improvement or a violation in the real world.

How to define a concept for institutional argument

• Core meaning: the minimal idea you refuse to give up.
• Scope: which domains it applies to (criminal law, labor markets, education, migration, digital platforms).
• Metric/indicator: what you will look at to judge whether the concept is realized.
• Priority: whether it can be overridden, and by what.

Example: defining “justice” for a policy argument

Core meaning: Justice is the fair structuring of social rules so that persons can live as secure equals under those rules.

Continue in our app.

• Listen to the audio with the screen off.
• Earn a certificate upon completion.
• Over 5000 courses for you to explore!

Or continue reading below...

Download the app

Scope: Applies to laws and public institutions that allocate burdens and benefits (taxation, policing, welfare, education, housing).

Indicators: predictable legal protection, non-arbitrary treatment, access to basic opportunities, and distribution of burdens that can be justified to those who bear them.

Priority: Justice can permit trade-offs with efficiency, but not at the cost of treating some as mere instruments or placing them under avoidable domination.

Practical step-by-step: concept test questions

• Borderline cases: What cases make your concept ambiguous? Write two and decide.
• Counterexample check: Can your definition label an obviously unjust institution as “just”? If yes, refine.
• Action-guidance check: Could two opposite policies both satisfy your definition? If yes, add specificity (scope, indicators, or priority rules).

Step 3 — State Principles That Can Actually Do Work

Principles are the bridge between your concept and institutional design. They should be general enough to apply beyond one case, but specific enough to constrain choices.

Three kinds of principles you can combine

• Constraint principles: set limits (e.g., “no policy may impose avoidable severe burdens without justification”).
• Priority principles: rank values when they conflict (e.g., “basic security takes priority over marginal gains in convenience”).
• Procedural principles: specify decision rules (e.g., “those affected must have meaningful avenues to contest decisions”).

Practical step-by-step: write principles in “if–then” form

Turn vague claims into conditionals that guide design.

Principle 1 (Equal Standing): If a rule predictably burdens a subgroup, then it must include a justification and mitigation that those burdened could reasonably accept as free and equal persons.  Principle 2 (Least Intrusive Means): If the state pursues a public aim, then it should choose the effective option that imposes the least avoidable burden on basic interests.  Principle 3 (Contestability): If an institution exercises discretionary power, then it must be subject to transparent reasons, review, and accessible appeal.

Notice each principle contains: a trigger condition, a requirement, and an implicit rationale tied to your concept.

Step 4 — Connect Principles to Institutions (The “Design Translation”)

Many arguments fail at the translation step: they state attractive principles but never show what those principles require in the architecture of law. Treat institutions as systems with inputs (rules), processes (decision procedures), and outputs (enforcement and outcomes).

A translation map: from principle to institutional feature

Practical step-by-step: build an “institutional specification”

1. Name the institution (agency, court, board, ministry, municipal authority).
2. Define its powers: what it may do, what it may not do.
3. Define decision procedures: criteria, evidence standards, timelines.
4. Add accountability: reporting, review, appeal, sanctions for misuse.
5. Plan implementation: budget, staffing, rollout, evaluation metrics.

Output format: write 8–12 bullet points as if drafting a policy memo or legislative summary.

Step 5 — Run Objection/Response Rounds (Make the Argument Earn Its Confidence)

Objections are not interruptions; they are part of justification. A robust normative argument anticipates where reasonable people will resist and shows either (a) why the objection fails, or (b) how the proposal can be revised to accommodate it.

Four families of objections to prepare

• Conceptual objection: “Your concept is misdefined or incomplete.”
• Principle objection: “Your principle is too strong, too weak, or inconsistent with another commitment.”
• Empirical/feasibility objection: “Your plan assumes facts that are false or uncertain.”
• Institutional/strategic objection: “Even if ideal, it will be captured, abused, or produce perverse incentives.”

Practical step-by-step: the objection worksheet

Round structure (do at least two rounds)

1. Round 1: respond with minimal changes (clarify definitions, add a safeguard).
2. Round 2: if the objection persists, consider deeper revision (re-rank priorities, narrow scope, add independent oversight, or accept a trade-off explicitly).

Step 6 — Evaluate Your Argument Using Four Standards

Evaluation is not only about whether your conclusion feels attractive. Use explicit standards so that readers can see why your proposal deserves acceptance even under disagreement.

1) Coherence (internal consistency and inferential strength)

• Consistency: Do your principles conflict? If so, do you have a priority rule?
• Non-ad hoc reasoning: Are exceptions principled or just convenient?
• Valid link: Does the institutional design actually follow from the principles, or is it a leap?

Quick test: Can you summarize your argument as a chain of 5–7 numbered claims where each step is supported by the previous one?

2) Feasibility assumptions (real-world constraints without surrendering normativity)

• Administrative feasibility: Can the institution implement the rule with available capacity?
• Compliance incentives: Will people predictably evade, resist, or adapt in ways that undermine the goal?
• Information limits: Does your policy require knowledge officials cannot reliably obtain?
• Transition costs: What happens during rollout, not just after equilibrium?

Quick test: List three assumptions your proposal needs; for each, state a fallback plan if it fails.

3) Respect for persons (how the policy treats individuals as agents)

• Non-instrumentalization: Are some people used merely as means for others’ benefit?
• Agency and voice: Do affected persons have ways to understand, contest, and influence decisions?
• Proportional burdens: Are burdens justified to those who bear them, especially when burdens are concentrated?

Quick test: Write a one-paragraph justification addressed to the most burdened group, using reasons they could recognize as persons with equal standing.

4) Handling disagreement (stability under plural values)

• Public justification: Can you offer reasons that do not depend on a single comprehensive worldview?
• Reasonable pluralism: Does your design allow dissenters to live with the policy without humiliation or exclusion?
• Institutional humility: Are there review mechanisms and sunset clauses that acknowledge fallibility?

Quick test: Identify one reasonable objection you cannot fully answer; then show how your institution contains the damage (review, opt-outs where appropriate, compensation, or democratic revision).

Capstone Sequence: Build a Full Argument End-to-End (Worked Template)

Use the following sequence as a fill-in structure for your own topic. The example below uses a policy area (algorithmic decision systems in public benefits administration) because it forces you to connect values to institutional safeguards.

A) Topic selection

Policy question: A public benefits agency should be allowed to use algorithmic tools to flag fraud and prioritize case review, but only under a legal framework that prevents arbitrary denial, ensures contestability, and protects equal standing.

B) Concept clarification

Concept: Justice (for this argument) requires that access to essential public benefits be governed by rules that are non-arbitrary, explainable to recipients, and reliably protect people from wrongful deprivation.

What counts as a violation: A person loses or is denied benefits through a process they cannot understand, cannot effectively challenge, or that predictably misclassifies certain groups without correction.

C) Principles

• Due-Protection Principle: If a decision threatens access to essential benefits, then the decision procedure must minimize wrongful deprivation and provide timely correction mechanisms.
• Equal-Standing Principle: If a tool produces systematically different error rates across groups, then its use is unjust unless corrected or replaced.
• Reason-Giving Principle: If the state acts against a person’s claim, then it must provide intelligible reasons sufficient for challenge.
• Accountability Principle: If an agency uses complex tools, then independent oversight must be able to audit performance and sanction misuse.

D) Institutional proposal (specification)

Institutional design: “Algorithmic Use and Benefits Protection Framework”

• Permitted use: Algorithms may be used for prioritization and fraud flagging, not for automatic denial or termination.
• Human decision requirement: Any adverse action requires a trained caseworker’s documented review.
• Notice and explanation: Recipients receive a plain-language notice stating the key factors that triggered review and what evidence can rebut it.
• Appeal timeline: A fast-track appeal process with a decision deadline; benefits continue during appeal unless there is clear evidence of intentional fraud.
• Bias and error audits: Quarterly audits for overall accuracy and group-differentiated error rates; mandatory remediation plans when thresholds are exceeded.
• Data governance: Limits on data sources; prohibition on using proxies that are strongly correlated with protected characteristics unless justified and monitored.
• Independent oversight board: Authority to inspect models, require changes, and publish public reports; includes technical experts and community representatives.
• Vendor constraints: Contract terms require audit access, documentation, and penalties for noncompliance; no “black box” exemptions.
• Sunset and renewal: Authorization expires unless renewed after evaluation against predefined metrics (wrongful denial rate, appeal success rate, disparities).
• Remedies: Compensation for wrongful deprivation and systemic corrective orders when patterns are found.

E) Objection/response rounds

Objection 1 (feasibility): “This is too costly and slows enforcement.”

Steelman: Oversight boards, audits, and continued benefits during appeal increase administrative costs and may allow fraud to persist longer, undermining public trust and program sustainability.

Response: The proposal treats wrongful deprivation as a high-cost error (human harm, legal liability, downstream social costs). It permits algorithmic prioritization to improve efficiency while forbidding automated adverse action. Cost concerns are addressed by (a) limiting audits to key metrics, (b) using sampling methods, and (c) targeting human review to high-risk cases.

Revision: Add a tiered review system: low-risk flags trigger information requests; high-risk flags trigger expedited human review, with stricter evidence thresholds for immediate suspension.

Objection 2 (institutional): “Oversight will be captured or become symbolic.”

Steelman: Oversight bodies can be underfunded, politically pressured, or dependent on the agency, producing reports without real change.

Response: Build independence into the institution: separate budget line, fixed terms, protected access rights, and enforcement powers (sanctions, mandatory remediation orders). Require public reporting to enable external scrutiny.

Revision: Add a legal right for recipients (or advocacy organizations) to trigger an external review when disparity thresholds are exceeded, creating a pathway that does not rely solely on internal will.

Objection 3 (conceptual): “Reason-giving is impossible with complex models.”

Steelman: Some models cannot provide meaningful explanations; simplified explanations may mislead recipients and undermine genuine contestation.

Response: The requirement is not to reveal proprietary code but to provide contestable reasons: the actionable factors and evidence categories that drove review. If a model cannot support this, it fails the reason-giving principle and should not be used for decisions that threaten essential benefits.

Revision: Restrict permissible models to those that can generate stable, understandable factor summaries and allow counterevidence; otherwise limit use to aggregate planning rather than case-level action.

F) Explicit trade-offs (state them, don’t hide them)

• Trade-off 1: Some fraud may persist longer due to continued benefits during appeal in exchange for reducing wrongful deprivation and preserving equal standing.
• Trade-off 2: Administrative costs rise due to audits and oversight in exchange for legitimacy, error correction, and prevention of systemic bias.
• Trade-off 3: Model choice is constrained (possibly reducing predictive accuracy) in exchange for contestability and reason-giving.

Capstone Deliverable: Your Final Institutional Proposal (One-Page Format)

When you finish, your output should look like a short policy brief that is visibly grounded in normative reasoning.

One-page structure you can copy

• Policy question (1 sentence)
• Concept definition (3–5 sentences): core meaning, scope, what counts as violation
• Principles (3–5 bullets): written in if–then form
• Institutional design (8–12 bullets): powers, procedures, accountability, implementation
• Objections and responses (2–3 pairs): steelman + response + any revision
• Evaluation table: coherence, feasibility assumptions, respect for persons, handling disagreement
• Trade-offs (3 bullets): what you give up and why it is acceptable

Evaluation table template (fill it in)