Ordinary transformations may explain retelling chains without an extra recursive state
Human–model retelling patterns may follow from composed ordinary transformations, even when semantic loss is real and order matters. Reject this account if a reproducible intervention discrepancy exceeds uncertainty and the meaningful margin after adequate state enrichment and model checks.
Stage of verification
- Hypothesis published2026-10-05
- Indirect evidenceAssessed at 4 of 10
- Direct testAwaited
Map of the hypothesis
Hover over an icon or tap it to see its name.
Lens
Kind of knowledge gap
Target map
Every target of every published hypothesis, each with the actions a hypothesis can propose on it. The targets and the actions of this hypothesis are drawn solid.
Scale or classification
Cultural transmission mechanism classification
Classification of cultural transmission mechanisms into causally distinct families
Where this hypothesis actsHuman–AI retelling chains evaluated on independent seeds and held-out contexts
Hypotheses on this target 9
Telling states apart9
Direct measurement
Indicator replacement
What is proposed
Telling states apart
Determine whether recursive mechanisms are distinguishable from compositional nulls
With whatInstrument or assay
HowFreeze calibrated one-step and reconstruction nulls; test held-out predictions and resource-matched ancestry-access and source-regeneration interventions
Possible result
Possible reclassification of distinct recursion as ordinary composed transformations, retaining observed attractors
From the recordUnder predictive equivalence on independent seeds and genuinely held-out contexts, remove the distinct recursive family's novelty and priority, while retaining the observed attractor phenomenon.
All targets of the lab
Every target read from the published hypotheses, each kind around its pictogram. A larger mark means more hypotheses act on that target. Point at a mark and the actions proposed on it branch out of it.
Solid and named: the targets of this hypothesis
Explore in depth
The logic
The train of thought that ends in this hypothesis. Each stage is the reason the next exists. The master question narrows to a goal, the goal to an unknown nobody has closed, the unknown to the hypothesis proposed here. Every step below says what it rests on and what carries it.
Stories can lose details or repeatedly settle into similar versions as people and text-generating software pass them back and forth. The unexpected move is to propose that these patterns need no additional hidden condition belonging to the story's history: independently measured changes at each handoff might explain the whole sequence. This is a proposal generated by the pipeline, not a result anybody has measured in the supplied material. Its claim concerns the need for an additional cause, not the reality of lost meaning or recurring versions.
- The current text and supplied circumstances shape what a person or software system keeps, changes or omits at one handoff.
- That output becomes the next handoff's input, so ordinary preferences for some content over other content accumulate across the chain.
- Changing the order of human and software handoffs can change the outcome even without an additional hidden historical cause.
- Reconstruction from available information, renewed access to the original source and ordinary learning by a returning person also change later versions through separately measured routes.
- Repeated application of these ordinary changes is proposed to produce semantic losses, meaning losses in what the story says, and attractors, meaning recurring patterns toward which versions tend to converge.
- Measured inputs and ordinary learning are proposed to account for those patterns without a further causally necessary condition attached to the chain's history.
A note passed through two editors can end up repeatedly losing the same details because each editor follows ordinary editing habits. Reversing the editors can change the final note even if neither follows a special rule for the note's ancestry.
Where the picture breaks: People can remember earlier edits, pursue changing goals and learn, while software output can vary even with the same input. The proposal therefore depends on measuring relevant circumstances and ordinary learning; the picture does not establish that human–software retelling is fully explained by fixed editing habits.
- Master questionstep 01 of 04
Cultural information changes as people copy it, reinterpret it, compete for attention and keep it alive. The research goal is to identify roughly five genuinely new, testable explanations of those processes and rank useful first experiments, including experiments involving recommendation systems and text-generating software. It requires separate measures of how many people encounter an item, copying accuracy, changes in meaning, adoption and persistence, together with competing explanations, meaningful controls and clear ways each proposal could fail.
Rests on: The stated goal defines memetics as the study of cultural transmission, transformation, competition and persistence. It explicitly requests a research agenda grounded in evidence, with established explanations distinguished from new conjectures and affordable initial tests distinguished from broader validation.
Stated in the chain - Goal pillarstep 02 of 04
The competing families of explanations and their first experiments are to be ranked according to what the available evidence can support.
Rests on: The master question explicitly requests a ranking based on novelty, explanatory power, ability to distinguish competing explanations, feasibility and expected reduction in uncertainty.
Stated in the chain - Gap questionstep 03 of 04
Repeated human–software retelling may require an extra memory of the chain's history, or its changes in meaning may be predictable by combining ordinary one-handoff transformations and independent reconstructions given equal resources. Predictions must face new situations that were not used to fit the explanations.
Rests on: The ranking goal requires comparisons that distinguish a new causal explanation from established alternatives. The master question explicitly includes human–software cultural transmission and requires checking whether an apparently new mechanism is already explained under another name. This stage turns those stated requirements into a candidate comparison, rather than reporting that the comparison has already earned first place.
Stated in the chain - Hypothesisstep 04 of 04
Ordinary transformations are proposed to explain repeated retelling without an additional chain-level state, meaning a causally necessary condition carried by the chain's history beyond the explanations already measured. Human and software handoffs would first be measured separately, including effects of the current text and stated circumstances; combining those measurements would then predict later versions. Separate measurements would also cover reconstruction from the available information, returning to the source and ordinary learning when the same person participates again. The order of human and software handoffs is allowed to matter. Applying the same measured handoff rules repeatedly assumes that those rules remain stable within the chosen grouping of texts and stated circumstances. The explanations must be fixed before the evaluation chains are examined, and the proposal fails if a reproducible effect of an intervention exceeds their uncertainty and a previously chosen meaningful margin after their adequacy has been checked.
Rests on: The gap question explicitly supplies the alternative that independently measured, resource-matched reconstruction and combined one-handoff changes can predict new retelling chains. The hypothesis develops that alternative into a bounded proposal with a prediction and a failure condition; the preceding question does not establish that the prediction will succeed.
Stated in the chain
What is carried, and what is not. No screened sources were supplied, so none of the six mechanism links has screened literature support in this record; their basis here is the stated competing explanation and the proposed measurements. Nothing supplied establishes the sequence end to end, although the final hypothesis follows an alternative already named in the gap question.
How a result here could mislead · 3
- Failure to find an advantage for a history-dependent explanation could be mistaken for evidence that the advantage is meaningfully absent. Uncertain one-handoff estimates, uncertain judgments about meaning and too few independent chains can all hide a real discrepancy, while producing many software outputs can create a misleading appearance of a large sample. What closes it: The smallest meaningful improvement must be fixed before evaluation, and uncertainty in the handoff measurements and judgments about meaning must be carried into the range of predicted outcomes. The design must estimate variation between chains and starting stories, then simulate its ability to detect meaningful departures. Independent chains and starting stories, rather than generated-output count alone, determine the amount of independent evidence. If calibration, the preliminary estimation of the ordinary explanations and their uncertainty, leaves insufficient precision to establish predictive equivalence, meaning performance close enough to fall within the chosen margin, the result must remain inconclusive.
- An ordinary explanation that is too crude can make missing wording, unequal resources or returning-person learning look like a new historical mechanism. Conversely, an explanation repeatedly expanded after seeing the results could absorb every failure and make the proposed absence of an extra mechanism impossible to disprove. What closes it: Human-only chains, software-only chains and alternating chains in both starting orders must face evaluation on independent starting stories and genuinely new circumstances. The allowed family of ordinary explanations, input features, evaluation measures and uncertainty must be fixed before the evaluation chains are seen. Calibration must capture relevant wording and current circumstances, measure ordinary learning independently and explicitly match supplied information and resources. Any permitted enrichment of the measured inputs must have stated limits and face fresh evaluation data; checking absolute predictive adequacy, meaning whether the explanation predicts observations well at all, is required alongside comparing rivals. Repeated refitting to the same failed evaluation chains cannot count as an independent success.
- Different later retellings after the same written checkpoint could be credited to an extra historical cause even though participants received different evidence, effort or access to earlier versions. A single total accuracy measure could also conceal the rivals' distinct patterns: deliberate resistance to correction, loss of particular causal relations, failure to reinstate earlier actions or selective removal of details that constrain later reinterpretation. What closes it: The checkpoint must match the current text, supplied circumstances and resources, while interventions distinguish each claimed extra cause from the independently measured ordinary routes. Access to earlier versions and reconstruction from the source must be manipulated with explicit resource matching. Evaluation must separately track propositions, meaning statements that can be checked against the source; causal-role reversals, meaning swaps in who did what to whom; missing exceptions; and retention of privately supplied source statements. The supplied record names four rivals but does not provide full matched intervention designs for each. Consequently, success on a general retelling task alone would not exclude every rival in the particular circumstances where it predicts an effect.
What would make this wrong. The claim of no necessary extra historical cause would fail within the tested scope if an intervention produced a reproducible discrepancy larger than both predictive uncertainty and the previously chosen meaningful margin after current text, supplied circumstances, resources and ordinary learning were adequately measured or matched, and the ordinary explanations had passed appropriate adequacy checks. For example, changing an earlier history of exercised revision rights could alter later source-faithful production at an otherwise matched checkpoint in a way the fixed ordinary explanations could not predict. Such a failure would reject their claimed sufficiency; it would not by itself establish which of the four proposed historical explanations is correct.
What it would change. If fixed ordinary explanations accurately predict new retelling chains and the candidate historical additions provide no meaningfully better predictions under adequately sensitive tests, the distinct recursive family, meaning the family claiming an additional cause arising from repeated retelling, would lose novelty and priority within this comparison. Recurring versions and losses of meaning would remain phenomena to explain, but the wider research agenda would have a reason to direct costly expansion toward other mechanisms. This first test uses short fictional texts, privately assigned source statements, one fixed software-model version and independent human chains; its priority depends on obtaining useful one-handoff measurements and is not a global ranking of the master project's families. Even success would leave other languages, populations, software models and cultural practices unestablished, as well as the wider outcomes of audience reach, adoption and persistence.
The gap this hypothesis explains
Two live hypotheses pull in opposite directions here, and the field has not chosen between them.
Do human–machine retellings need a new explanation, or can existing accounts predict how meanings change in unfamiliar settings?
Original wording · exactly as the pipeline generated it
Do human–AI retelling chains require a distinct recursive mechanism, or can resource-matched independent reconstruction and composed one-step channels predict their semantic trajectories in held-out contexts?
What this question is asking
The question concerns how a story’s meaning changes when people and artificial intelligence systems repeatedly retell versions produced earlier in a chain. It asks whether those changes require an additional recursive mechanism: an effect of repeated feedback that existing accounts of individual retellings cannot explain. The alternatives are independent reconstruction, where each retelling is rebuilt separately from specified source material, and composed one-step channels, where predictions for individual retellings are linked together to predict a whole chain; the comparison holds available resources comparable and concerns new chains and settings excluded from developing the predictions. The accompanying gap description claims that existing work already shows limited effects of cultural attractors and content biases, but treats the need for an additional recursive mechanism as unestablished; no screened sources are supplied to verify that account.
- Artificial intelligence; human–machine or human–AI retelling chain
- Artificial intelligence (AI) here means a computer system that generates or rewrites language. A human–machine retelling chain is a sequence in which people and such systems retell material derived from earlier versions; the supplied input does not specify their order or arrangement.
- Recursive mechanism
- A proposed process in which the consequences of earlier exchanges feed back into how later retellings are produced. In this question, a distinct recursive mechanism must add something beyond the influence already represented by linking ordinary retelling steps; the supplied material does not specify that extra dependence.
- Independent reconstruction
- An alternative account in which a retelling is rebuilt separately from specified source material instead of being explained by an additional process spanning the chain. Exactly what each reconstruction receives and what it is independent of are not specified in the supplied input.
- One-step channel; composed one-step channels
- A one-step channel is an account of how one input version can become an output version in a single retelling. Composing channels means linking those accounts, using possible outputs from one step as inputs to the next, to predict changes across a chain.
- Resource-matched; resource control
- These terms mean keeping relevant available resources comparable between the accounts or processes being compared, or accounting for differences in those resources. Such resources could include effort or access to information, but the supplied material does not identify which are controlled.
- Semantic trajectory; meaning change
- Semantic means concerning meaning. A semantic trajectory is the sequence of changes in what a story conveys over successive retellings; it can include several dimensions rather than one single score, and no particular measure is specified here.
- Held-out context
- A setting excluded from developing or adjusting an account and then used to assess its predictions. The question asks whether predictions remain useful beyond the settings used to construct them, but does not specify what differs between settings.
- Independent chains
- Separate sequences of retellings used to assess whether a prediction extends beyond the particular sequence from which it was developed. They are distinct from independent reconstruction, which names one of the competing accounts of how retellings are produced.
- Cultural attractor
- A form of cultural material toward which repeated transformations are proposed to tend, such as a recurring way of telling a story. The term names a tendency across transformations rather than a claim that every story reaches one fixed endpoint; the supplied description asserts relevant effects without supplying their evidence.
- Content bias
- A tendency for features of the material itself to affect what is remembered, retold, or changed. This names a class of possible tendencies, not one demonstrated effect with a fixed size in all settings.
- Bounded transformation effect
- A reported change in transmitted material established only within particular conditions or measurements. Here it is the gap description’s characterization of earlier work, not a finding that can be verified from supplied sources.
- Predictive advantage
- Better agreement between an account’s predictions and what is subsequently observed than a competing account achieves. The question requires an advantage that matters for explaining meaning changes, but supplies no criterion for how much improvement qualifies.
- Causal mechanism
- A process that produces an outcome through specified intermediate steps. Correctly predicting an outcome does not by itself establish which process produced it, because different processes can sometimes yield similar observations.
- Node; pipeline
- In the supplied gap description, a node is an item or stage within the research pipeline, the sequence of steps that generated the proposed question. A statement attributed to a node is not itself a supplied literature finding.
The gap description states that attractor and content-bias work establishes bounded transformation effects and that resource-control work supplies alternatives, while no node establishes the necessity of an added recursive mechanism.
The description assumes that earlier work has documented limited changes in cultural material caused by tendencies to converge on certain forms or to preserve some kinds of content more readily than others. It also assumes that accounting for differences in available effort and information supplies competing explanations, without having established a need for an extra effect of repeated feedback. If supported, this would locate the unresolved issue in the extra explanatory value of the proposed mechanism rather than in whether stories ever change during retelling.
The supplied screened_sources list is empty. The gap description reports what an earlier pipeline considers established, but provides no source text or source identifiers with which to check the reported transformation effects, resource comparisons, or coverage of prior explanations. It also does not establish that relevant searches were sufficiently broad; the absence of supplied evidence neither supports nor refutes these assertions.
The same question asked without the part nothing read establishes:
- Can accounts of separate retellings predict meaning changes in new human–machine storytelling chains when available resources are comparable?
- Does an account that adds dependence on earlier exchanges predict meaning changes in unfamiliar human–machine storytelling settings better than accounts built from individual retellings?
- Existing accounts predict the changes If independently rebuilt retellings or linked predictions for individual retellings account for meaning changes in new chains and settings under comparable resources, the observed trajectories would not require the added recursive explanation within that scope. Those predictions would explain the changes without establishing that every internal process in people or machines had been identified.
- An added recursive account is needed If the existing accounts fail and an added account of dependence on earlier exchanges reliably predicts the otherwise unexplained meaning changes, the added account would have predictive value for the settings assessed. That advantage would support retaining the extra dependence in the explanation, although predictive success alone would not prove that the proposed causal process is uniquely responsible.
- The answer depends on the setting If existing accounts succeed in some settings while an added recursive account predicts better in others, the extra explanation would have a limited range of use. Treating either result as universal would then produce mistaken expectations about meaning changes outside the settings where it holds.
A retelling changes the version available to the next storyteller, so changes introduced at one step can affect what happens later. Existing accounts of separate retellings might already predict this accumulation, even when the final story differs greatly from the starting version. Treating every accumulated change as evidence of a new mechanism could therefore assign explanatory value to something the existing accounts already cover. Conversely, if an additional dependence on earlier exchanges changes later meaning beyond those accounts, leaving it out could make predictions fail when the chain or setting changes.
Attractor and content-bias nodes establish bounded transformation effects; resource-control nodes expose alternatives, but no node establishes necessity of an added recursive mechanism.
Before prioritizing recursive human–AI theory, establish a meaningful semantic predictive advantage over calibrated alternatives on independent chains and held-out contexts.
The proposed novelty and priority can collapse if established channels predict the same trajectories; independent mechanistic falsification must precede investment in broader validation.
The mechanism it proposes
The engine's own statement of the hypothesis, in full.
PHENOMENON-DOESN'T-EXIST: the apparently distinct recursive human–AI mechanism is an epiphenomenon of composing ordinary content-biased transformations, resource-matched reconstruction and the observation process. The empirical attractors and semantic losses may be completely real. What does not exist in the tested scope is a causally necessary extra chain-level state. Calibrate H(y|x,c) and A(y|x,c) on randomized one-step transformations, allowing sufficiently rich current-artifact features x and declared context c. For row-vector distributions under a stationary discretization, alternating chains predict p0(HA)^k or p0(AH)^k; H and A need not commute. Independently calibrate q(y|theme, task, supplied information, budget), source regeneration, and ordinary repeated-individual learning/reconstruction where a human returns. Coarse coding, unequal budgets and unmeasured lexical cues can otherwise look like recursion. These are a prespecified family of compositional nulls, not a post hoc universal model allowed to absorb every discrepancy.
Testing and possible results
The prediction that would tell it apart
A hypothesis that predicts what its rivals predict is not worth running an experiment over. This is the observation on which this one differs.
Freeze nulls and their uncertainty before seeing evaluation chains. In human-only, model-only, HA and AH chains, held-out proposition transitions, causal-role reversals, exception loss and private-source retention fall inside the propagated predictive envelope, and any candidate extension improves proper predictive scores or prespecified discrepancies by less than the smallest meaningful margin. Order effects are allowed. At an identical-current-artifact checkpoint, matched context, resources and independently measured ordinary learning explain later differences; no additional lineage-right, local-fatigue, motor-history or joint-contract state earns predictive value. Intervene on ancestry access and source regeneration with explicit resource matching: context-conditioned kernels predict the changes without chain-specific refitting. Under predictive equivalence on independent seeds and genuinely held-out contexts, remove the distinct recursive family's novelty and priority, while retaining the observed attractor phenomenon. This IH loses if a reproducible intervention-specific discrepancy exceeds uncertainty and the meaningful margin after competent state enrichment and absolute model checks; that loss does not automatically identify which extension is right.
What testing it would take
The engine's own read on whether this is testable with methods that already exist.
This is the affordable first-stage gate for this L3: short fictional texts, private randomized source propositions, one version-fixed model and independent human chains. It can rule out costly family expansion before broader sampling. Estimate chain/seed effects and kernel/annotation uncertainty, then simulate design power against meaningful deviations; generated-output count is not the independent sample size. If model/annotation calibration consumes too much of the available budget, report the test as underpowered for equivalence instead of announcing no recursion. Multilingual, new-population, new-model and cultural-practice validation is a separate stage. The first experiment's priority is conditional on the ability to estimate useful kernels; it is not a global ranking of all families in the master project.
Other explanations
Every other hypothesis the engine wrote for the same gap, and the observation that would separate the two.
Freeze nulls and their uncertainty before seeing evaluation chains. In human-only, model-only, HA and AH chains, held-out proposition transitions, causal-role reversals, exception loss and private-source retention fall inside the propagated predictive envelope, and any candidate extension improves proper predictive scores or prespecified discrepancies by less than the smallest meaningful margin. Order effects are allowed. At an identical-current-artifact checkpoint, matched context, resources and independently measured ordinary learning explain later differences; no additional lineage-right, local-fatigue, motor-history or joint-contract state earns predictive value. Intervene on ancestry access and source regeneration with explicit resource matching: context-conditioned kernels predict the changes without chain-specific refitting. Under predictive equivalence on independent seeds and genuinely held-out contexts, remove the distinct recursive family's novelty and priority, while retaining the observed attractor phenomenon. This IH loses if a reproducible intervention-specific discrepancy exceeds uncertainty and the meaningful margin after competent state enrichment and absolute model checks; that loss does not automatically identify which extension is right.
- Rival 01 of 04What would separate them
Overridden revision rights may make accurate model corrections provoke deliberate errors predicts: Cross correction accuracy with revision authority. In development sessions, establish either participant final approval or neutral editorial approval using matched stories and identical accepted text. Subsequently provide identical verified source-correct model repairs while experimentally retaining or overriding the previously exercised approval right. Include a yoked observer with the same texts, actions, timing and accuracy evidence but no ownership of that lineage; model-versus-human source labels are independently counterbalanced. At an identical current-artifact checkpoint, the rights hypothesis predicts more intentional correct-to-incorrect or correct-to-incompatible transitions after accurate override than after accurate authorized repair, despite equivalent private source-question accuracy. This negative correction-dose slope should transfer with assignment of the lineage's revision right and disappear when the right is prospectively relinquished; an unrelated right on another lineage should not suffice. Binding fatigue predicts dependence on conflicting revision cycles, not legitimate versus illegitimate authority; embodied reinstatement predicts action matching; contract ambiguity predicts audit payoffs. Compare with independently calibrated reactance, algorithm-aversion, endowment, ordinary learning and belief-conditioned H kernels, not only an unconditioned H. If those established components predict the authority-by-history contrast within the meaningful margin on held-out lineages, retire the proposed distinct family. No residual interaction alone identifies a new norm mechanism.
- Rival 02 of 04What would separate them
Conflicting revisions may erode connected story memories even after the text is repaired predicts: Use graph-matched fictional narratives with experimentally known causal links. Randomize whether an equal number of incompatible intermediate corrections repeatedly touches one connected neighborhood or dispersed unrelated links; restore the identical correct full text at a checkpoint and equate final exposure, total conflicting propositions, task time and output tokens. With the same human returning, the localized condition should show accelerating, spatially adjacent causal-role failures predicted by independently estimated a_t and revision-load amplitude, despite matched checkpoint text. Fresh humans should reset that excess; reinstating a gesture without repairing the affected bindings should not. Estimate load and binding accessibility in separate calibration participants to avoid the diagnostic test becoming retrieval practice. Compare the law against arbitrary flexible item-level learning/interference models, graph-conditioned one-step H and A, repeated-individual reconstruction, and exposure-position controls. A stable power-law relation fitted on one load/graph range must predict another without refitting its exponent. If endpoint text, ordinary interference and causal connectivity explain the trajectories; if growth is unrelated to connected damage; or if a_t merely redescribes the same scored errors, reject fatigue as a distinct mechanism. A good curve fit alone is insufficient.
- What would separate them
Reinstating learned gestures may preserve causal roles during human–model retelling predicts: At an identical-artifact checkpoint, cross semantically congruent role enactment at encoding with matched or swapped spatial enactment at later human production. Include no-enactment and equal-amplitude meaningless-movement controls, the same verbal generation and source-question practice, matched delays and workload, and fresh-human handoffs. Gesture instructions must not reveal any missing proposition; assign counterbalanced arbitrary locations to already supplied characters. This IH predicts an encoding-by-reinstatement interaction: congruent motor reinstatement selectively preserves the earlier causal roles, whereas swapping the learned locations increases role reversals even with the same current text. The interaction should persist after balancing ordinary verbal generation/retrieval practice and be absent for an unlearned movement mapping. Fatigue predicts localized cycle-dose deficits and fresh-person reset, but not this sign-changing mapping interaction; the rights and contract accounts predict their social manipulations instead. Calibrate an ordinary multimodal encoding-specificity model on nonrecursive tasks and replay controls. If it predicts the entire chain interaction, the motor explanation may be useful but the proposed new recursive family is eliminated. If matched motor perturbations have no meaningful role-specific effect despite a successful action-memory manipulation, reject this scout in favor of other models.
- What would separate them
Continuing evaluation with coarse checks may reward strategic omissions in retelling predicts: Cross continuation of the same evaluation relationship versus a one-shot handoff with coarse whole-story evaluation versus prespecified proposition-specific audit. Hold expected reward, current source, evidence access, output length/time, task wording and candidate policy fixed as far as feasible; report residual incentive differences. Explicitly distinguish descriptive accuracy from public approval. The contract account predicts more omission of auditable exceptions/attributions under continuing relationships with coarse verification, and a selective reversal under item-level audit. The expected gain from possible later reinterpretation should predict WHICH details disappear, even when those details are causally central, easy to recall and accurately answered in private. Remove future evaluation or assign liability to an independent editor while preserving authorship rights: strategic omissions should shrink; merely transferring revision ownership should not suffice. Compare with independently calibrated one-shot incentive, audience-design, risk-aversion, self-presentation and accountability kernels composed across rounds. Only a held-out continuation-by-verifiability effect beyond those components supports the proposed extra state. If the effects are fully predicted by ordinary task-conditioned editing, remove the recursive/new-family claim. No contract-specific selectivity, despite verified incentive comprehension, falsifies this mechanism in the task.
What stands behind it
Which of the figures above have a study behind them, which are the engine's own, and what it would take to refute the hypothesis. This audit never judges the idea.
Provenance audit: failed at enrich. Nothing below has been traced yet.