Debate Is Not a Duel

A family argument begins with a delay. One person says: you failed to let me know again. The other replies: it was only ten minutes. The first hears a problem of reliability; the second hears an attempt at control. They refer to the same evening, yet they are measuring different losses.

What turns disagreement into shared inquiry, and what reduces it to a contest of force

A family argument begins with a delay. One person says: you failed to let me know again. The other replies: it was only ten minutes. The first hears a problem of reliability; the second hears an attempt at control. They refer to the same evening, yet they are measuring different losses.

The conversation can deteriorate quickly from there. One side produces another example, the other begins to defend a broader pattern, and within minutes the delay itself has almost disappeared. Earlier experiences, presumed intentions, roles within the relationship, and the question of who is expected to yield have all entered the exchange. Raised voices are often a late symptom. By then, several distinct disputes have been compressed into a single sentence.

This is why civility alone is a poor measure of argumentative quality. Manipulation can be delivered calmly, while a tense exchange may still produce genuine clarification. The more useful question is whether the participants are getting closer to identifying what they actually disagree about.

Volume is usually an outcome

“Let us stick to the facts” is sensible advice, but it rarely resolves the problem by itself. The stakes of an argument are seldom exhausted by data. Data support an inference; the inference informs a decision; the decision distributes advantage, loss, and risk. If this chain remains implicit, the same number can occupy two incompatible stories.

Consider a software team deciding whether to release on Friday. The developer sees unresolved technical risk. Sales sees a contract that may be lost through delay. Operations anticipates a weekend incident, while testing has identified a defect that occurs rarely but affects many users when it does. Management is weighing cost and reputation. The apparent question is binary: release or postpone. The claims beneath it concern different probabilities, different kinds of damage, and different time horizons.

An early yes-or-no vote flattens the inquiry. The team first needs to establish the shared objective, distinguish measurement from estimate, identify reversible steps, and ask who carries each consequence. The conflict does not vanish. It becomes more specific: one part is an evidence problem, another is a value judgment, and another is a problem of risk allocation.

Toulmin’s model of argument remains useful because it does not merely place a claim next to supporting data. It also draws attention to the often unspoken warrant that explains why the evidence is supposed to support the conclusion [1]. Many disputes fail at precisely this point. One participant offers a number, the other rejects the conclusion, and neither examines the connecting rule.

The same defect rate may support different recommendations. One person regards the risk as acceptable because the event is rare. Another rejects it because the consequences are severe. The number has not changed. The threshold of acceptability has.

The claim and the speaker are different objects

When a statement is uncomfortable, it is often easier to explain the speaker than to examine the argument. “You only say that because...” produces an instant psychological portrait. The suspected motive may be real. The claim may still be accurate, partly accurate, or false. A description of the person does not perform the work of verification.

Sources and interests do matter. A financial stake, political position, or personal grievance can justify closer scrutiny. The mistake begins when interest itself is treated as refutation. “Interested” does not mean “wrong,” just as an appearance of independence does not establish correctness.

Mercier and Sperber’s argumentative theory proposes that human reasoning is strongly adapted to social contexts in which people produce and evaluate reasons [2]. The theory is contested and does not need to carry an entire account of human cognition. It nevertheless illuminates a familiar asymmetry: people often detect weaknesses in another person’s reasoning more readily than in their own.

That asymmetry can be useful. Disagreement may function as an external checking mechanism. It becomes corrosive when the roles harden. One side is assigned to attack, the other to defend, and neither gains anything by naming the weaknesses of its own position. Reasoning then becomes advocacy.

Research on motivated reasoning shows how desired conclusions can influence which arguments appear persuasive, how much evidence is demanded, and when inquiry is allowed to stop [3]. A serious debate therefore has to negotiate more than propositions. It also needs some convergence on standards of acceptance. What would count as sufficient evidence? What finding would weaken the present view? Which part of the claim would be revised if an opposing result appeared?

Without answers to those questions, demands for evidence become theatrical. Participants ask for proof, but only of the kind that the other side is expected to accept.

Different arguments fail different tests

Expert testimony, analogy, causal explanation, and argument from consequences should not be assessed in the same way. Walton, Reed, and Macagno’s work on argumentation schemes is useful because it associates different types of reasoning with different critical questions [4]. The relevant test depends on the relationship being claimed.

With expert testimony, one should ask whether the person is qualified in the relevant field, whether conflicts of interest matter, and whether substantial expert disagreement exists. With causal claims, the questions concern mechanism, alternative explanations, and the strength of the evidence. With analogy, the key issue is whether the two cases are similar in the features that actually matter to the conclusion.

Public argument routinely blurs these distinctions. The statement of a prestigious person is used as causal evidence. A personal story is treated as an estimate of prevalence. An extreme case becomes an account of normal institutional behaviour. Such claims may remain rhetorically effective while carrying less evidential weight than their form suggests.

Sometimes the debate does not need another argument. It needs a clearer account of the work the existing argument can legitimately perform. A personal testimony may be indispensable for understanding how an event was experienced. It may tell us very little about national frequency. A statistical average may establish prevalence without disproving the reality of a specific harm.

Claims at different levels can appear contradictory even when they answer different questions.

An audience changes the sentence

Public debate is rarely a conversation between two people. The home audience, the opposing camp, the future video clip, the comment feed, and the fear of losing status are all present. A sentence must remain connected to the issue while also responding to the expectations of spectators.

Sunstein’s work on group polarization describes conditions under which discussion among like-minded people can shift participants toward more extreme positions [5]. This is not an automatic law and does not affect every group equally. It still offers a reason to create genuine institutional space for disagreement rather than invite opposition as decoration.

An audience also changes the cost of correction. In private, “I was wrong about that” may be easy. In public, the same sentence can be interpreted as weakness, betrayal, or loss of identity. Argumentative quality therefore depends partly on procedure. Is clarification possible without humiliation? Can partial agreement be acknowledged? Is revising a claim distinguishable from surrendering one’s status?

Habermas sets a demanding ideal in which validity claims are open to challenge and participants are able, in principle, to question them [6]. Actual public life distributes time, access, rhetorical skill, and power unevenly. Mansbridge and colleagues therefore argue that deliberative analysis cannot ignore self-interest and power [7].

“Let us discuss it” becomes meaningful only when the process also asks who sets the agenda, who controls access to evidence, who must speak under strict time limits, and who can simply refuse to answer.

Agreement is not always the proper endpoint

Some conflicts survive additional evidence. Two people may accept the same facts and still assign different weight to security, liberty, fairness, tradition, or risk. The dispute continues because the difference lies in values or in the distribution of burdens, not because one side failed to understand the numbers.

Clarification still matters. It shows where the factual dispute ends and where an accountable choice begins. That distinction makes it harder to present a political or personal decision as an objective necessity when several viable options remain.

A good debate may therefore end without consensus. Its achievement may be a more accurate disagreement. One side may be prioritising probability while the other prioritises severity. One may be speaking about a national average while the other is describing a concrete injustice. Recognising this is not a dramatic victory. It prevents us from failing each other’s claims on tests they were never designed to pass.

What should remain after the argument?

A structured debate is not a hidden classification system, and it cannot guarantee peace. It can, however, be judged by a few observable results.

At the end, it should be clearer which claims were supported by evidence, which remained estimates, where value judgments entered, and what uncertainty the decision accepted. Participants should be able to identify what could change their views. The record should also show when one side consistently addressed a different issue from the one named in the debate.

This matters within the perspective of Fractal Dialectics because conflict is not reduced to a single axis of victory. The aim is not to preserve every statement or dissolve every difference. The more workable task is to see which claims genuinely connect, which operate at different levels, and where a decision must still be made after uncertainty has been acknowledged.

The result may not be a calmer debate. It is less likely to end with every participant offering a more polished answer to a question the other side never asked.

A duel asks who remains standing. Inquiry asks what remains usable.

References

[1] Toulmin, Stephen E. (1958/2003): The Uses of Argument. Updated edition. Cambridge University Press.

[2] Mercier, Hugo; Sperber, Dan (2011): “Why Do Humans Reason? Arguments for an Argumentative Theory.” Behavioral and Brain Sciences, 34(2), 57-111. DOI: 10.1017/S0140525X10000968.

[3] Kunda, Ziva (1990): “The Case for Motivated Reasoning.” Psychological Bulletin, 108(3), 480-498. DOI: 10.1037/0033-2909.108.3.480.

[4] Walton, Douglas; Reed, Chris; Macagno, Fabrizio (2008): Argumentation Schemes. Cambridge University Press.

[5] Sunstein, Cass R. (2002): “The Law of Group Polarization.” Journal of Political Philosophy, 10(2), 175-195. DOI: 10.1111/1467-9760.00148.

[6] Habermas, Jürgen (1984/1987): The Theory of Communicative Action. Volumes 1-2. Beacon Press.

[7] Mansbridge, Jane et al. (2010): “The Place of Self-Interest and the Role of Power in Deliberative Democracy.” Journal of Political Philosophy, 18(1), 64-100. DOI: 10.1111/j.1467-9760.2009.00344.x.