Methodology.

The ladder, in one image

Think of E0–E14 as zoom settings on a camera pointed at reality: E0 is physics itself, E5 is a single cell, E8 is a mind, E10 is an institution, E12 is a nation, E14 is all of humanity at once. Each rung is built from the one below it and does things the one below it cannot — cells can't hold grudges, minds can't declare war. Every film here is tagged by the zoom level where its story really happens, not where it's set: a heist about trust is an E9 story even in orbit, and Groundhog Day breaks physics at E0 but lives entirely at E8, inside one repeating mind. Read more — The Arena →

Why transposition

Some films relocate minds onto things that don't have them — animals, machines, toys, the dead. The relocation is never the point; it is an instrument, and it cuts two ways. Distance changes what can be said: a film about prejudice among humans arrives pre-litigated, but move the same dynamic onto predators and prey and the audience absorbs the structure before recognizing whose structure it is. And distance changes what can be imagined: a world of conscious machines or waking aliens is a laboratory for the emergent patterns a consciousness in such a reality would experience — dynamics that don't exist yet anywhere except in the thought experiment. The lens records four things per film: the flag (is consciousness transposed?), the substrate (what wears the mask), the story level (which rungs of the E0–E14 reality ladder the story actually fights on), and the abstracted plane (which aspects of reality the storyteller kept at full resolution, what was strategically forgotten, and what the compression distorts).

The second lens: parameter shifts

Some films are best understood as controlled experiments on reality: freeze one parameter at a non-actual value, hold everything else constant, and run the machine forward. Groundhog Day breaks time at E0 and spends its whole runtime at E8, inside one mind under infinite repetition. The lens records the shifted level (where on the ladder the dial sits), the parameter and its shift, the scope (universal, local, or personal), the traced levels (where the consequences land), what was held constant as control, and a rigor grade — did the world re-emerge lawfully from the shift, or bend for plot?

• Premise, not outcome. The experiment is the standing rule the world runs on, not the events that happen within it. A comet strike is an event; a world where sound means death is a rule — which is why A Quiet Place counts and a disaster film doesn't.
• Allegory is excluded. Talking animals standing in for human society belong to the transposition lens. A parameter shift must be a genuine counterfactual, not a metaphor wearing one's clothes.
• Supernatural furniture is excluded — unless a system is derived from it. Ghosts as atmosphere don't count; an afterlife whose citizens persist only while the living remember them does, because the film built a mechanism, not a mood.
• Diffuse worlds don't count. A setting that differs from ours in a hundred unexamined ways (a galaxy far, far away) freezes no single dial and traces nothing.

The interesting measurement is the gap between where the dial sits and where the consequences land — turn a parameter at E0, watch the ripples arrive at E10. Filter the atlas by Reality altered to see these films; the shifted level is marked in mint on each film's ladder.

How each film is read

Every film enters with its catalog metadata via TMDB and a full plot sourced from Wikipedia — a film without a substantive Wikipedia plot is not analyzed. The week Wikipedia gains the plot, the film joins the catalog. Each qualifying film then gets one reading per lens: a language model (currently openai/gpt-oss-120b) reads the plot against a fixed system prompt.

What the rules enforce

• Centrality. A robot sidekick, a talking pet, or one magical creature does not flag a film. The displaced universe must be the film's stage, not a prop on a human stage.
• Central does not mean protagonist. When the drama is about humanity's relationship with non-human minds — first contact, alien refugees, created minds demanding personhood — the transposed minds are the subject even when the hero is human.
• Transformation is not transposition. A human cursed into another shape is still a human mind. The enchanted objects around them, however, do count.
• All-nonhuman casts flag. A world populated entirely by anthropomorphized animals or machines is transposed even when its plot summary reads like a human drama.
• Manufactured minds are machines. Robots, AIs, androids, replicants, and sentient programs share one substrate, so the same entity never lands in two categories.
• Keywords are single words. "Fear economy" becomes fear; "wealth disparity" becomes inequality. One-word tags make every film aggregable against every other.

Status, and honesty about it

The atlas currently holds … analyzed films of a …-film catalog, growing weekly as new releases are analyzed by an automated pipeline. The transposition rules were calibrated on a fifty-film pilot with hand-chosen expected answers — including deliberate traps — and reached 48/50 on the flag after three tuning rounds. The parameter-shift rules went through four pilot rounds and two production stress tests before scaling.

Readings are machine-made with selective human review, and the negatives get audited too: an adversarial second pass — a prompt whose only job is to refute the classification — swept 629 films labeled "reality unaltered", flagged 169, and every flag was arbitrated by the calibrated standard prompt. Final score: 19 genuine corrections, about a 3% miss rate, all re-run through the rules and live in the atlas. The audit is cheap and repeats after every large batch.

Boundary films exist and are kept, not hidden: a cursed-human lead over an enchanted-object cast, or a human mind in a machine body, can legitimately be read either way, and model variance at that boundary is real. Treat any single film's reading as a considered first pass, and the aggregate patterns — which survive individual errors — as the substance of the tool.

Corrections

If a film's reading is wrong, say so — suggest a correction and it goes straight to the maintainers. Corrections are re-run through the same rules rather than hand-edited, so a fix becomes a better rule, not a silent exception.

Missing a film entirely? Request it here — the same form takes longer notes on any analysis you disagree with.

Reuse

The full dataset is published as a Kaggle dataset — id, title, year, substrate, story levels, keywords, plane, and credits per film, plus the parameter-shift fields (flag, shifted levels, parameters, traced levels, rigor). Film metadata comes from TMDB; this product uses the TMDB API but is not endorsed or certified by TMDB. The readings themselves are © The Emergence Machine — cite the tool when you build on them.

← Back to Movie Substrate