Every judge accepts the same doctrine: liability holds exactly when there was a valid contract and it was breached. Vote on each question separately and the court can contradict that doctrine. Nobody has reasoned badly. The inconsistency belongs to the group.
Each judge must give a consistent account of the case. Their answer about liability follows from their answers about the contract and the breach. The paradox appears when the court takes a majority on each proposition rather than deciding as one body.
This is the discursive dilemma, first made vivid in jurisprudence by Kornhauser and Sager. It is not a trick of a three-person court: larger benches make the contradiction more likely before the rate levels off. Change the findings below and watch the routes through the court separate.
Kornhauser and Sager's example starts with a doctrine that looks harmless: a defendant is liable if and only if there was a valid contract and that contract was breached. Each judge can apply it without any ambiguity. Yet the court does not have to behave like a judge writ large. If a majority finds the contract valid, a majority finds a breach, and a majority finds no liability, the three majority answers cannot all be true together. The group has lost the relation its members preserve individually.
The striking feature is that no one is confused. The paradox is produced by the agenda: propositionwise voting treats the findings and the verdict as separate objects, even though the doctrine makes one depend on the others. Once the majority is allowed to assemble the court's answer one proposition at a time, it can assemble a view that no single judge would accept.
A court can be made of consistent people and still speak inconsistently.
One response is premise-based aggregation. The court votes on the two findings and derives the verdict from the doctrine. This preserves a coherent legal explanation, but it can reject the verdict most judges supported. A second response is conclusion-based aggregation. The court votes on liability and leaves the findings open. This preserves the majority verdict, but it cannot say which premise carried it. The third route reports every majority as it fell, preserving the vote on every question while accepting that the resulting set can contradict the doctrine.
None is simply the rational answer because each protects a different institutional value: reasons, verdicts, or direct responsiveness. List and Pettit turned that observation into a general impossibility result. Under plausible conditions, no aggregation rule can keep all of the attractive properties at once. The example therefore belongs beside debates about representation and expertise, not only in a logic class.
The simulation samples consistent judge profiles; it does not model evidence, advocacy, precedent, or the strategic behaviour of a real court. Its frequency is the frequency of logical configurations under the chosen distribution, not a prediction of litigation outcomes. Real judges may coordinate, give reasons publicly, or use a rule that avoids propositionwise voting altogether. The point is narrower and more useful: once a collective body must combine judgments, consistency is not inherited automatically from its members.
The classic bench makes the failure visible. The count makes it harder to dismiss as a three-judge trick. Change one finding, switch the connective, or enlarge the bench. The disagreement is not in the arithmetic; it is in deciding which thing a court is meant to preserve.