Strip the wonder-tale of its names, its kings and its firebirds, and a fixed skeleton is left: 31 functions enacted by 7 roles, and the functions always fall in the same order — any may be omitted, none may be swapped. V. Ya. Propp turned narrative into a grammar, and the grammar is runnable: a tale is a subset of the 31 indices, and it is valid iff its functions ascend. The blue team builds the grammar; the red team scrambles it.
source V. Ya. Propp, Morphology of the Folktale (Leningrad: Academia, 1928; Eng. trans. L. Scott, 1958/1968) — Russian formalism, the structural grammar of the wonder-tale. Room: THE NOVEL. Rendered as structure, not quoted.
Petrograd/Leningrad, the 1920s: the Russian Formalists (Shklovsky, Jakobson, the OPOJAZ circle) ask not what a tale means but how it is made. Propp takes ~100 Afanasyev wonder-tales and does something new: he ignores content (who does it, with what) and records only function — the act as a unit of the plot.
His claim, 1928: the dramatis personae change endlessly, but their functions are constant, few, and few in kind. Form over content — the structuralist turn, a generation before Levi-Strauss put a name on it.
The functions distribute across exactly seven spheres of action. A role is not a person: one character can fill several roles, and several characters can share one.
Propp’s central structural claim: the functions always occur in the same sequence. A tale never runs the wedding before the villainy, the return before the departure. What varies is which functions appear — omission is free, reordering is forbidden.
That reduces “is this a well-formed tale?” to one computable test: are the chosen function-indices strictly increasing? The centre engine is exactly that test.
The blue team re-runs the law on load: a known in-order tale must be accepted, a known out-of-order tale must be rejected, and the counts must hold (31 functions, 7 roles). If red drops the ordering law (6), this badge is where it shows.
Build a tale by choosing functions. Toggle them on; they slot in ascending order by default (a lawful tale). Presets load known tales; shuffle scrambles the order to forge an unlawful one.
The checker walks the sequence once and asks a single question at each step: is this function’s index greater than the one before it? One descent anywhere and the tale is rejected.
| step | f | function | prev < f ? |
|---|
Rule: valid ⇔ strictly increasing. Omissions (gaps) are fine; a single inversion is fatal.
LINEAGE. This ordered skeleton is the seed of a century: Propp → Levi-Strauss’ mythemes → Campbell’s hero’s journey → the screenwriting beat sheet. A valid tale above is, structurally, the shape all four describe.
The engine is exact on its domain and silent off it. The famous over-claim lives in that silence.
“All stories are literally the same.” Cut. Propp’s grammar is a wonder-tale grammar — a strong result over one genre, then over-extended by pop-culture into “every story ever.” The engine only certifies wonder-tale form.
“There are exactly 7 plots / 1 monomyth.” Cut. That is Campbell/Booker downstream, not Propp. Propp counts 31 functions and 7 roles — not 7 plots.
“Order does not matter, only the ingredients.” Cut. Order is the whole claim. Same functions, wrong order → not a tale. The checker proves it live.
The red move: drop the ordering law. Make the checker accept any order — ingredients only. Now a scrambled non-tale like [11,8,31] passes as “valid,” and the witness (9), which demands that exact sequence be rejected, turns red.
With the law dropped, “valid tale?” degenerates into “are these real functions?” — order stops meaning anything and every scramble passes. The attack is real, and the witness catches it.