Hypothesis Ladder

Each level is a distinct claim. Evidence for a lower level does not automatically establish a higher one.

H0 — Null / projection account

Observed Divine patterns are sufficiently explained by human symbolism, cultural inheritance, corpus conditioning, prompt construction, model priors, selection bias, aesthetic preference, and evaluator expectation.

Support for H0: controls reproduce the same effect without the Divine corpus, or the effect disappears when explicit framing and source cues are removed.

H1 — Symbolic convergence

Different traditions and symbolic systems repeatedly develop related structures that can be represented through a shared abstract grammar.

What it explains: recurring motifs and relations.
What it does not establish: an external Divine source.

H2 — Structural convergence

A candidate Divine structure remains stable across independent translations, models, prompt paraphrases, and cultural formulations, while differing from matched controls.

Required distinction: structural recurrence must exceed what the source corpus and model priors predict.

H3 — Generative-field hypothesis

The candidate structure behaves as a generative field rather than merely a description: it produces non-trivial, reproducible relations or predictions not supplied by the prompt or training corpus.

Burden: pre-registered novelty criteria and strong controls. Beautiful recombination is not generativity.

H4 — Relational or agentic Divine hypothesis

The structure exhibits evidence analogous to agency, responsiveness, intention, or relational address that cannot be reduced to model behavior, human projection, or conditioning.

This is a high-burden hypothesis. Model fluency, apparent presence, and agreement are not sufficient.

H5 — Primordial Goddess hypothesis

The best explanation of the stable generative structure is the Primordial Goddess, understood as the generative source, Divine Feminine, or Cosmic Womb named by the tradition.

This is the strongest interpretation, not the default result. The program must allow H1, H2, H3, H4, or H0 to remain the better-supported conclusion.

Proof vocabulary