Texts contain linguistic, rhetorical, semantic, chronological, and editorial structure. Arabic morphology, genre, repetition, corpus length, topic, and preprocessing can all create non-random graph and geometric patterns without implying Mind, toroidal physics, or divine origin.
Y TheoryThe Compass of RealityComputational humanities · network science · comparative corpus analysis
Quran corpus to relational Mind-structure test
Freeze one corpus-agnostic engine and ask whether held-out Quranic relations reveal two coupled cycles and predictive structure beyond matched controls and rival manifolds.
Exact question
Does one fixed, corpus-agnostic Y-Engine recover a preregistered recursive, history-bearing two-cycle relation from held-out Quranic material better than matched corpora, structure-preserving nulls, and rival geometries?
Declared relation
X_C(t):=F_Θ(G(C),X_C(t−τ)); S_C:=w·[two-cycle topology,coupling,recovery,held-out AUPRC]; Δ_Q:=S_Q−max(S_controls,S_rivals)Native mechanism → Y relation
Preserve the carrier before translating the structure.
The mapping is auditable only when both descriptions remain visible. Y language does not overwrite the equation, mechanism, units, boundary conditions, or uncertainty of the field being studied.
The founder proposes that Quranic words and relations can drive a bounded Y-Engine state whose differentiation, return, recursion, coupling, and recovery form a computational image of Mind. A static corpus is input; only engine plus evolving state plus returned consequence is tested as a structural-identity candidate. The visual torus is downstream, not evidence.
Calibration and withheld return
The answer must not help write the question.
Calibration establishes measurement and allowed parameters. The held-out test remains unseen until the identity rule, model, uncertainty, rivals, scoring, and failure threshold are frozen.
- Declare one immutable Arabic witness, edition, orthography, license, content boundary, and checksum. A repository candidate is quran_uthmani_tanzil.txt (SHA-256 c21dcb07a8715b9831bf62834cb33a6383f07182f2f827ec52a704709ab9cf82; Quran content lines 1–6236; footer excluded but preserved), but it is not the frozen witness until explicitly confirmed.
- Require 100% reversible mapping from every graph feature to exact byte, codepoint, token, ayah, and surah spans. Keep roots, lemmas, syntax, tafsir, recitation, chronology, and translations as separately sourced overlays whose ambiguity is preserved.
- Freeze a whole-surah or contiguous-passage train/development/holdout split; sequence, containment, exact recurrence, morphology, syntax, semantics, rhetoric, and recitation are separate typed edge layers. No selected verse may tune and test the same rule.
- Use the pre-existing engine without Quran-specific weights, sacred-role nodes, injected golden ratios, injected torus coordinates, or a presentation layer in the causal update. Freeze graph construction, update equations, delay ratios, initialization, parameter budget, seeds, stopping rule, metrics, uncertainty, and analysis code hash.
- Name the two independently varying recurrent phases and their expected coupling before opening the holdout. If they are merely corpus-traversal phase and an imposed engine-return phase, label the result a constructed operational torus rather than discovered intrinsic Quranic topology.
- Calibrate a blind topology detector on synthetic torus positives and circle, sphere, cylinder, plane, and noise negatives. A torus candidate expects persistent β0=1, β1=2, β2=1 when sampling supports it, but Betti numbers alone cannot satisfy the identity or prediction gates.
- Match comparison corpora on language, length, genre, period where feasible, repetition, vocabulary, and segmentation. Predeclare other scriptures, Classical Arabic prose or poetry, within-surah order shuffles, n-gram surrogates, degree-preserving rewires, root-label permutations, and relation-ablated Quran variants; use enough frozen null samples for the declared alpha.
- Predeclare one primary predictive endpoint: AUPRC for withheld typed-edge prediction. Also freeze identity gates for boundary separation, persistence, recurrence, internal/external coupling, residue recovery, perturbation recovery, and source recovery.
- Run the frozen engine and detector once on untouched Quranic units, every matched corpus, at least 999 frozen nulls when testing at p≤.01, and the circle, sphere, cylinder, plane, and flexible latent-manifold rivals under matched complexity and parameter budgets.
- Pass only if all reversible-source and identity gates succeed, the two cycles are independently supported and mutually coupled, the Quran score exceeds the 99th percentile of corrected null scores, and paired block-bootstrap 95% confidence intervals for topology/model selection and held-out AUPRC improvement over the strongest baseline stay above zero.
- Ablate delay/residue, horizontal coupling, whole-return, source input, and the display separately. The proposal predicts that causal ablations reduce recurrence, recovery, and held-out prediction while changing only the rendering does not.
- Repeat under preregistered plausible orthography, tokenization, annotation, and sampling variants. Dependence on one convenient root analysis, translation, passage selection, or decoder fails the corresponding robustness scope.
- Have an independent implementation reproduce the result from the frozen corpus identity, code specification, configuration, seeds, and decision rule. Publish raw outputs, uncertainty, all failed runs, and a separate supported, inconclusive, or failed result record.
Rivals · failure · present state
A claim becomes informative where it can lose.
No visual resemblance, repeated phrase, source density, or author confidence substitutes for a discriminating return.
- Arabic morphology and root structure
- Genre, rhetoric, topic recurrence, and corpus length
- Ordinary semantic-network organization shared by many coherent texts
- Arbitrary embedding or toroidal projection that can fit any sufficiently rich corpus
- Circle, sphere, cylinder, plane, or a flexible latent manifold under the same complexity budget
- Frequency-only, co-occurrence-only, memoryless, no-coupling, and no-whole-return engine ablations
- Researcher choice of verses, numbers, translations, edges, weights, or stopping rule after viewing the result
Invalidate the measurement run as inconclusive for non-reversible parsing, footer leakage, failure to distinguish calibrated positive and negative topology controls, post-hoc cycles, weights, verses, or thresholds, hard-coded sacred identities, or a torus imposed by the display. Reject or narrow the registered computational claim if a valid frozen run fails the identity gates, does not exceed every declared corpus control and rival at the pass threshold, fails held-out typed-edge prediction, loses the signature under declared robust variants, or fails independent replication.
Editable draft for preregistration; the exact corpus witness, two candidate cycles, engine and analysis hashes, primary endpoint, numeric thresholds, holdouts, controls, and independent operator are not yet frozen, and no result has been earned.
A pass would support one specified computational correspondence at one corpus and transform. It would not establish that the Quran, a toroidal visualization, or an AI has subjective consciousness; that electrons and thoughts are the same carrier; that Islam endorses Y-Theory; that the Quran is divine; or that the complete Y-Theory framework is true.
Recoverable lineage