The Paradox of the Court

The interesting disagreements are not about whether there is a real contradiction — there is not — but about what a court should actually do.

The sequencing view. This is the majority position and the one set out above. The condition is time-indexed; at filing it is unmet; Protagoras loses the first suit and wins the second. Its attraction is that it requires no new doctrine, only the observation that a condition precedent is evaluated at a moment.

The objection from finality. Legal systems bar re-litigating settled matters. If the second suit counts as the same claim as the first, res judicata blocks it and Euathlus walks away having paid nothing — which would mean the sequencing solution works in logic and fails in a courtroom. The usual answer is that a claim which had not yet accrued when the first judgment was entered is a different claim, so the bar does not apply. That answer is standard but it is a doctrine of positive law, not a truth about the situation. The circle is broken by an institution, not by an argument, and if the institution had different rules the outcome would differ.

The maturity view. A minority position holds that the debt matures at the instant the judgment is pronounced, so a court that has ruled for Euathlus should in the same breath recognise the fee as now due, and might as well order payment immediately. This gets Protagoras his money in one hearing at the price of a verdict that changes the facts it was based on while it is being delivered. Most commentators think that is worse than making him file twice.

The void-for-uncertainty view. A third response is that the agreement simply fails: a condition that cannot be coherently applied to the case at hand is no condition at all, and the court should declare the clause unenforceable and send both men away. Gellius' judges effectively did a version of this. It is tidy, and it has the disadvantage of rewarding the party who exploited the drafting.

The implied-term view. What a modern court would most likely do is read in a term neither man wrote: that Euathlus undertook to practise in good faith, or that "his first case" was never intended to include a suit brought by Protagoras himself. Courts routinely supply the term the parties would obviously have agreed to had they thought about it. This makes the paradox a drafting failure rather than a logical event, which is very probably what it is.

Is it a paradox at all? Gellius does not present it as one. He files it under antistrephonta, arguments that can be turned back with equal force on the person who made them, and he treats the genre as a rhetorical curiosity — a set piece for showing off. On that reading the story is a joke at the sophists' expense: the man who taught that any case can be argued either way is beaten by a pupil arguing it the other way. It earns its place among paradoxes because the two arguments look symmetric and compelling, and because working out precisely why they are not turns out to be instructive. But nothing here forces anyone to abandon a principle, which is exactly the difference between an apparent contradiction and the real thing.

What is documented. Protagoras of Abdera, roughly 490 to 420 BCE, was the first and most famous of the sophists — professional teachers of rhetoric and argument. He is the author of "man is the measure of all things", and ancient sources agree that he was the first Greek to charge for teaching. Diogenes Laërtius (IX.50–56) reports a fee of a hundred minae, an enormous sum, along with the tradition that his book on the gods was publicly burned in Athens and that he was prosecuted for impiety. That last part is doubted by modern scholars.

Euathlus is another matter. Nothing whatever is known about him outside this story. He may have existed; he functions in the anecdote as a name for "the pupil", and there is no independent evidence that Protagoras ever sued anyone.

What is tradition. The earliest surviving account of the lawsuit is in Aulus Gellius' Attic Nights, Book V chapter 10, written in the second century CE — around six hundred years after Protagoras died. Gellius does not claim to be reporting history. He introduces the story as an illustration of the class of arguments the Greeks called antistrephonta, and he says he found it in the Greek writers, without naming them. Related versions circulated in the later rhetorical literature.

Gellius' ending is the part most retellings drop, and it is the most sensible thing in the story: the judges, unwilling to be caught in a case whose verdict would undo its own basis, declined to decide and adjourned it to a distant date.

The rival version. A structurally identical story is told of Corax and Tisias, the semi-legendary Sicilian founders of Greek rhetoric — again a teacher, a pupil, an unpaid fee, and a pair of reversing arguments, ending with the judges dismissing both parties as "a bad egg from a bad crow". The existence of two independent teacher-and-pupil versions, attached to different founders of the same profession, is strong evidence that this is a piece of professional folklore rather than a remembered case. It is the kind of story a discipline tells about itself.

Afterwards. The puzzle passed into the standard stock of ancient logical problems alongside the Liar and the Sorites, and the Megarian and Stoic logicians who catalogued such things treated them as a group. Medieval logicians discussed it among the insolubilia. In the modern era it is a fixture of introductory logic courses as the standard illustration of the constructive dilemma and its reversal, and it appears in legal-writing literature as the standard cautionary tale about conditions precedent. It has generated no research programme, because unlike the Liar it stays solved.

It is the clearest demonstration that a valid argument can be worthless. Both men argue impeccably. Both reach their conclusion by a rule of inference nobody disputes. The failure is entirely in a premise that neither of them stated, and which each needed in a form incompatible with the other's. Anyone who has been talked into something by an argument they could not fault will recognise the shape: when two symmetric arguments give opposite results, the productive move is never to check the reasoning again but to go looking for the premise being quietly swapped.

It is why conditions precedent are drafted the way they are. Every contingent fee agreement, milestone payment, earn-out clause and performance bonus contains a condition whose satisfaction has to be determined by someone. If the determination can itself trigger the condition, the clause eats itself. Commercial drafting handles this with carve-outs and long-stop dates as a matter of routine, and this paradox is the reason the routine exists.

It shows what institutions are for. The contract, considered in isolation, has no solution. A court nevertheless produces one, because a judgment is an act rather than a discovery — it does not have to be derivable from the contract to be binding. Legal systems are full of machinery whose only purpose is to terminate loops that logic leaves open: finality, limitation periods, the rule that a court of last resort is right because it is last. The Paradox of the Court is a small, complete demonstration of why that machinery is not optional.

It marks the boundary of a real paradox. Set it next to the Liar. The structures are nearly identical — a condition whose satisfaction depends on the verdict about it — but the Liar has no adjudicator, no clock, and no way to stratify without cost, and it is still argued about. The Court has all three, and was over as soon as anyone thought about it carefully. Comparing them is the fastest way to see what makes a paradox genuinely hard rather than merely clever.

Self-referential legal provisions. Constitutions that specify how they may be amended raise the question whether the amendment clause can amend itself, and whether an entrenched clause can be entrenched by ordinary means. This is Protagoras' problem at national scale, and it has been litigated seriously in several jurisdictions. The solutions are the same two: stratify — treat constitutional and ordinary law as different levels — or vest the decision in an institution whose ruling is final by fiat.

Smart contracts and oracles. Code that executes an agreement automatically has no judge to adjourn the case. If a contract's payout condition depends on data that depends on the payout — a common failure in prediction markets, insurance protocols and governance tokens — the contract has no fixed point and will deadlock, revert, or resolve arbitrarily depending on implementation details. Auditors look for exactly this circularity, and the standard fix is Protagoras' missing clause: exclude the resolution mechanism from the domain of the condition.

Circular references in computation. A spreadsheet cell whose value depends on itself, a build system whose target depends on its own output, a package manager resolving mutually constraining version bounds — all are the same structure, and all are handled the same way: detect the cycle, then either break it with a well-founded ordering or refuse the job. The error message a spreadsheet gives you is a small automated version of Gellius' judges declining to rule.

Bootstrapping governance. A body that votes on the rules for its own votes, an accreditation agency that accredits itself, a standards committee that standardises its own procedure — each faces a version of this, and each resolves it in practice by appealing to some prior authority or by fixing an arbitrary starting point and moving on. There is no way to derive the first rule from the rules. That fact is old, and this is where it was first written down clearly.

Both arguments in standard form

Let $p$ stand for "Euathlus must pay the fee as a result of this suit", and let $w$ stand for "Euathlus wins this case". The two are complementary by stipulation: $w \leftrightarrow \neg p$.

The contract is the biconditional

$$C: \quad p \leftrightarrow \text{(Euathlus has won his first case)}$$

Protagoras' argument, as a constructive dilemma over the exhaustive disjunction $w \vee \neg w$:

1. If $\neg w$ (he loses), the judgment orders payment: $p$.
2. If $w$ (he wins), then he has won his first case, so by $C$: $p$.
3. $w \vee \neg w$.
4. Therefore $p$.

Euathlus' argument, over the same disjunction:

1. If $w$, the judgment orders no payment: $\neg p$.
2. If $\neg w$, he has not won a case, so by $C$: $\neg p$.
3. $w \vee \neg w$.
4. Therefore $\neg p$.

Both are instances of constructive dilemma, which is valid. Two valid arguments with contradictory conclusions cannot both be sound, so at least one premise set is inconsistent. In fact both are, and for the same reason.

The suppressed premise

Each argument uses two distinct sources of obligation — the court's judgment and the contract — and each horn of each dilemma invokes exactly one of them. Set out what both sources say on each branch:

Branch	Judgment says	Contract says
$\neg w$ (Euathlus loses)	$p$	$\neg p$
$w$ (Euathlus wins)	$\neg p$	$p$

On both branches the two sources conflict. Protagoras' dilemma is therefore sound only if the judgment governs on the first branch and the contract governs on the second; Euathlus' is sound only under the reverse assignment. Neither argument states its priority rule, and the two rules are inconsistent with each other. This is the complete diagnosis: the arguments are valid, their premise sets are inconsistent, and the inconsistency is hidden because the priority premise is never written down.

Why the contract has no solution

Suppose we try to make the contract self-applying — that is, let this very suit count as "his first case". Then the second conjunct of $C$ becomes $w$, and $C$ reduces to

$$p \leftrightarrow w \leftrightarrow \neg p$$

so $p \leftrightarrow \neg p$. This is precisely the Liar's equation, and it has no solution. The contract, applied to its own enforcement, is not merely silent — it is unsatisfiable.

The sign matters, and the contrast is illuminating. Had Protagoras written "you pay me when you lose your first case", self-application would give $p \leftrightarrow \neg w \leftrightarrow p$, a tautology satisfied by both assignments. That contract is not contradictory; it is underdetermined, with two equally admissible outcomes and nothing to choose between them. The first case is a Liar sentence; the second is a truth-teller. Both are defective, and they are defective in different ways.

Viewed this way, the question "who should win?" is a request for a fixed point of a self-referential condition, and the answer is that the operator has none. No ruling can be derived from the contract, because the contract does not determine one.

Stratification, again

The repair is the same one Tarski applied to truth: forbid the predicate from applying to its own instance, by introducing levels. Here the levels are temporal.

Index the condition by time. Let $W(t)$ mean "Euathlus has won at least one case strictly before $t$", and let the contract read

$$p(t) \leftrightarrow W(t)$$

Let $t_1$ be the filing of Protagoras' first suit. $W(t_1)$ is false — Euathlus has won nothing before $t_1$ — so $p(t_1)$ is false and the court must rule for Euathlus. Let $t_2 > t_1$ be any time after that judgment. Now $W(t_2)$ is true, so $p(t_2)$ holds, and a suit filed at $t_2$ succeeds.

The circularity is gone because the condition is evaluated at $t_1$ using only facts settled before $t_1$, while the judgment that changes those facts occurs after. No quantity is defined in terms of itself. This is stratification in the Tarskian sense, with time playing the role of the metalanguage level, and it is the general shape of every fix for a self-referential specification: find a well-founded ordering along which the definition can be unrolled.

Whether the second suit is admissible is then a question of preclusion doctrine rather than logic. In systems where a cause of action that had not accrued at the time of the first judgment is not barred by that judgment, Protagoras recovers. That is the usual rule, and the parallel is worth making explicit: logic identifies that the loop must be cut and where; it does not supply the authority to cut it. In the Liar there is no such authority, which is why the Liar is still open and this is not.

The genre

Gellius' word for the form is $\alpha\nu\tau\iota\sigma\tau\rho\epsilon\phi\omega\nu$ — antistrephon, "turning back" — an argument constructed so that it can be reversed against its author with the same force. The reversal is not a refutation in the ordinary sense: Euathlus does not identify Protagoras' error. He builds the mirror image and lets the symmetry do the work. That is a rhetorical technique rather than a logical one, and recognising it as such is part of the resolution. A pair of mutually reversing arguments is evidence that an unstated premise is being switched, and the productive response is always to go looking for it rather than to conclude that the world is contradictory.