The lying criterion (Eco) — the complete line
The criterion: semiotics studies everything that can be used to lie. The half of the sentence that carries the argument: "If something cannot be used to tell a lie, conversely it cannot be used to tell the truth: it cannot in fact be used 'to tell' at all."
1 · It is a capacity test on the medium, not a verdict on acts
It asks: does this system permit an emitter to produce the expression in the absence of — or contrary to — the state of affairs? Three things are tested at once: a code exists, an emitter is free over it, telling (assertion or any other force) is possible. Whether a given emission is a lie, an honest error, or fiction is a different, pragmatic question (lying-vs-false-statement).
2 · The constitutive argument: detachability IS aboutness
For X to REPRESENT Y — rather than be caused by Y or be part of Y — X must be able to occur without Y. A "sign" perfectly coupled to its object (fires only when the fire burns) is not read but measured; it is the object's extension, not its representative. The relation "about" is born exactly where the coupling breaks. So lie-capacity is not a defect signs happen to have; it is the constitution of signhood. Philosophy of mind hit the same wall from its own side (Dretske and the misrepresentation literature): natural information cannot misrepresent — tree rings do not lie about the tree's age — and the mark of genuine representation is the capacity to misrepresent while functioning correctly. The lying criterion is the semiotic edition of the intentionality criterion.
3 · The symptom boundary, and the malingering edge
A fever cannot lie — not because it is honest but because nobody emits it: a causal trace (pure index), no producer, no convention. It can mislead (fever without infection), but misleading without an emitter is not lying. Edge case that sharpens the line: the moment you FAKE a fever's appearance, that appearance stops being a symptom and becomes your sign. The line runs through "is there a subject free to produce it", not through kinds of things.
4 · Why "lie" and not "error" or "false"
A lie requires more than possible falsity: an emitter with at-will command of the code — the capacity to deploy the sign in the object's absence (linguistics calls it displacement; it separates human language from the bee dance). And here sits the brightest coin of the whole line: the capacity to lie is the same engine as the capacity to imagine, plan, hypothesize, write fiction, design what does not yet exist. All of them produce expressions of what the world does not contain. The lie is merely the morally flagged corner of the detachment engine — and Eco names the field by its scandal because the scandal is the cleanest test: whether a system supports fiction can be argued; whether it supports a lie is checkable.
5 · The failed reformulation: "falsifiable implication"
Restating the criterion as "carries implications that can be false" keeps the semantic leg (content decoupling) and drops the other two, and misclassifies in both directions:
- Excludes non-assertoric signs: commands, questions, names, and ATMOSPHERE cannot be false (no assertion, no truth value) — yet all can deceive: insincere questions, loaded presuppositions, feigned orders, the assertion-free brand mood. Deception is pragmatic and rides every force; falsity is semantic and rides only assertion. A falsity criterion would expel exactly the sub-assertoric layer where (per confabulation-in-marketing) the money lives.
- Includes emitterless falsity: a broken thermometer emits false readings but nobody can USE it to lie — until you rig it, at which point it has become your sign. The word "used" carries the agent, the code, and the at-will deployment.
Terminology mine flagged for the controlled dictionary: "falsifiable" now circulates with two incompatible senses — Popper's (empirically refutable, a virtue of theories) and falsity-capable (a property of contents). Eco's criterion is neither.
6 · The spectrum: lying freedom x verification strength
All sign systems permit lies; they differ in how long a lie survives — the strength of the verification machinery policing the system:
| system | can lie? | the machine that catches it | lie half-life |
|---|---|---|---|
| symptom (pure index) | no (no emitter) | — | — |
| math notation | yes | deduction, built in (Peirce's stage two lives inside the system) | minimal |
| empirical claims | yes | experiment, external and slow | medium |
| brand symbols | yes | almost nothing | maximal |
Math is not the exception but the extreme case: 2+2=5 is writable (fully sign); what cannot be done is making it true — object-level rigidity, not sign-level. Marketing occupies the unpoliced end; a whole industry stands where lies go unverified. Civilization, on this axis, reads as a stack of sign systems each carrying its own immune system; interpretation itself exists only because signs can mislead — otherwise reading would be table lookup, and abduction would be unemployed.
The line in one sentence: lying capacity = detachment capacity = representation itself — the lie is the invoice meaning issues for its own existence.
Kin: abduction · lying-vs-false-statement · confabulation-in-marketing. Eco's remaining machinery (Model Reader, encyclopedia, limits of interpretation): the-eco-line.