Which new explanations of cultural survival hold up?

HERETICAL CANDIDATE SET={F_CORRELATION: endogenous distributed-correlation inheritance}; actual count=1 added family. In the bounded universe, this alone survives the retention rule, while dual-path, revocation and material signatures are explained by B. The claim is that a cultural semantic distinction can be inherited primarily through a relation among people's contributions, although no isolated contribution reliably identifies that distinction, AND that the relation is spontaneously regenerated across complete learner turnover without an experimenter-provided parity rule or codebook. The transmitted unit is the joint convention, not each message separately. Let Z be an arbitrary seeded semantic relation and X1,X2 the separately elicited contributions of two roles: I(Z;X1) and I(Z;X2) can lie within prespecified negligible-information bounds while I(Z;X1,X2) exceeds a positive practical bound. Ordinary secret sharing shows this information pattern is possible; the added dependency is an endogenously inherited coupling rule C_g that changes how otherwise matched components are reconstructed and recombined at generation g+1. A minimal model is P(X1',X2'|sources,C_g), with C_g itself inferred and retransmitted through interaction, rather than a fixed exogenous decoder. This family is necessary only if a independently fitted correlated-learning/transactive-memory/cultural-linkage composition fails on coupling transplants. B+F_CORRELATION is claimed sufficient for the tested intervention response profile; its empty additional-family subset B is not, because B misses selective destruction and recovery of the ancestral joint distinction under marginal-preserving swaps.

Reassignment must preserve low-order correlations up to the order the baseline explicitly models; otherwise ordinary cultural linkage explains it. predicts an excess loss of ancestral joint-semantic recoverability and subsequent lineage survival when the relevant higher-order coupling is broken, with restoration after the original coupling is reinstated; the standalone contribution decoders remain within the preregistered negligible-information band. Fresh cohorts must regenerate the same joint relation without an explicit codebook or any surviving informed participant. Added families F_DUAL_PATH, F_REVOCATION and F_MATERIAL fail their incremental retention criteria. If distributed cue-combination, learned partner conventions, cultural linkage, or fixed secret-sharing codes predict the swap and turnover data within delta_H, F_CORRELATION loses its distinct-family status. If one person's contribution or an external shared record carries the alleged missing relation, the strongest heretical version is falsified, rather than rescued by redefining that carrier as a correlation.

Use sparse graph realizations above and below a calibrated candidate joint-process threshold; vary only overlap/alignment where possible, and separately test misaligned content and provenance bottlenecks. predicts that interventions preserving each marginal graph's viability can substantially change the survival of source-qualified descendants by changing the coupled Jacobian's leading eigenvalue. The prediction is a finite-network change in the survival curve and lineage size distribution, not a guaranteed universal sharp transition. Raw copying/reach may remain similar while qualified semantic-lineage survival diverges. A ordinary multiplex model with fixed jointly estimated edge weights, or a complex-contagion plus source-monitoring model, must fail held-out graph realignments by delta_N before this becomes a new family. No incremental pairing-code, inherited-revocation or material-memory family passes. If those established compositions predict the effect, or the effect disappears after matching effective source attention and measured route dependencies, retire F_DUAL_PATH as a distinct family even when topology clearly matters.

Inherited withdrawal records may block stale copies while allowing independent support predicts instead: E1 gives matched groups the same final narrative propositions, source confidence, correction comprehension, number/order of ordinary reading opportunities and immediate endorsement, but manipulates which source occurrence a retained withdrawal relation points to. Use equal-size semantically relevant source statements as the strongest control, not a blank screen. Later deliver either a stale copy of the withdrawn occurrence or a truly independent new assertion of the same proposition, matched for wording, frequency and source reliability. predicts selective suppression of the stale descendant but reinstatement by independent support, with an additional effect of inherited operation ancestry after fitting source-monitoring/continued-influence/Bayesian evidence models. In fresh cohorts receiving equal final content, removing the causal withdrawal record should raise source-specific resurrection; retaining a record that points to a different occurrence should shift which branch is suppressed. This effect must survive held-out graph schedules and changed paraphrases; ordinary recency or an overall correction-strength difference is insufficient. Only Delta_D passes the common admission rule. If the explicit semantic source-statement control or an established causal source-monitoring model predicts these contrasts within delta_D, the entire claimed distinct family is removed, even if revocations help accurate reconstruction.

Weak disturbances of inherited material memory may preserve multiple cultural practices predicts instead: In E2 transmit an unfamiliar safe manipulation practice using trained compliant/granular workpieces or a pilot-validated physical medium with measurable cyclic memory. Cross inherited versus freshly reset workpieces with preserved versus absent process demonstrations, whole-cohort turnover, and weak mechanical perturbation versus an equal-duration sham. First measure independent material response curves and novice one-generation learning; use them to predict multigeneration behavior without the proposed extra coupling. Equalize opportunities, force/work budgets where possible, visible cues, immediate task success/payoff, and instruction exposure; assay residual physical differences rather than assuming macroscopically similar objects are identical. predicts a reproducible intermediate-perturbation region where survival of multiple functionally distinct ancestral practices exceeds both zero perturbation and memory-erasing perturbation, while resetting the inherited microstate removes that region. Critically, the gain must exceed the prediction from the separately calibrated physical-memory plus ordinary reconstruction composition, and transfer with the workpiece to fresh learners rather than following experienced participants. Simple noise-induced exploration, stochastic resonance, payoff changes, additional practice and informative wear traces are explicit rival fits. Only Delta_M exceeds its preregistered practical margin delta_M and passes admission; online pairing, routing and revocation residuals are explained by B. If the composed baseline predicts the entire region, reset effect and turnover transfer, retain the physical effect as known ecological/material inheritance and delete F_MATERIAL as a novel family.

CROSS-DOMAIN CANDIDATE SET={F_DUAL_PATH: joint content-and-provenance path viability}; count=1 added family. Only this additional dependency survives within U. A descendant must both reconstruct a cultural item and obtain an independently routed, causally valid relation to its source before it becomes a reproductive source itself. The candidate is not generic percolation or generic source credibility. Its proposed extra dependency is that reconstruction changes the availability of source-validating routes, and route availability in turn changes which reconstructed descendants qualify for further reproduction. Two populations with matched local contact counts, source reliability, visible consensus, content repertoire and ordinary reconstruction kernels can therefore differ after provenance-route swaps. The transmitted unit is a proposition or practice feature paired with a verifiable source relation. B plus this coupled dependency is hypothesized sufficient for the held-out intervention profile; B alone is insufficient because it treats provenance and content reproduction as separately calibrated channels or merely changes a static edge-transmission probability. All other U candidates fail incremental retention. The numerical size is one family; its distinctiveness depends on defeating the ordinary joint/multiplex-process composition, not just a single-layer contagion straw rival.

Use sparse graph realizations above and below a calibrated candidate joint-process threshold; vary only overlap/alignment where possible, and separately test misaligned content and provenance bottlenecks. predicts that interventions preserving each marginal graph's viability can substantially change the survival of source-qualified descendants by changing the coupled Jacobian's leading eigenvalue. The prediction is a finite-network change in the survival curve and lineage size distribution, not a guaranteed universal sharp transition. Raw copying/reach may remain similar while qualified semantic-lineage survival diverges. A ordinary multiplex model with fixed jointly estimated edge weights, or a complex-contagion plus source-monitoring model, must fail held-out graph realignments by delta_N before this becomes a new family. No incremental pairing-code, inherited-revocation or material-memory family passes. If those established compositions predict the effect, or the effect disappears after matching effective source attention and measured route dependencies, retire F_DUAL_PATH as a distinct family even when topology clearly matters.

Reassignment must preserve low-order correlations up to the order the baseline explicitly models; otherwise ordinary cultural linkage explains it. predicts an excess loss of ancestral joint-semantic recoverability and subsequent lineage survival when the relevant higher-order coupling is broken, with restoration after the original coupling is reinstated; the standalone contribution decoders remain within the preregistered negligible-information band. Fresh cohorts must regenerate the same joint relation without an explicit codebook or any surviving informed participant. Added families F_DUAL_PATH, F_REVOCATION and F_MATERIAL fail their incremental retention criteria. If distributed cue-combination, learned partner conventions, cultural linkage, or fixed secret-sharing codes predict the swap and turnover data within delta_H, F_CORRELATION loses its distinct-family status. If one person's contribution or an external shared record carries the alleged missing relation, the strongest heretical version is falsified, rather than rescued by redefining that carrier as a correlation.

Inherited withdrawal records may block stale copies while allowing independent support predicts instead: E1 gives matched groups the same final narrative propositions, source confidence, correction comprehension, number/order of ordinary reading opportunities and immediate endorsement, but manipulates which source occurrence a retained withdrawal relation points to. Use equal-size semantically relevant source statements as the strongest control, not a blank screen. Later deliver either a stale copy of the withdrawn occurrence or a truly independent new assertion of the same proposition, matched for wording, frequency and source reliability. predicts selective suppression of the stale descendant but reinstatement by independent support, with an additional effect of inherited operation ancestry after fitting source-monitoring/continued-influence/Bayesian evidence models. In fresh cohorts receiving equal final content, removing the causal withdrawal record should raise source-specific resurrection; retaining a record that points to a different occurrence should shift which branch is suppressed. This effect must survive held-out graph schedules and changed paraphrases; ordinary recency or an overall correction-strength difference is insufficient. Only Delta_D passes the common admission rule. If the explicit semantic source-statement control or an established causal source-monitoring model predicts these contrasts within delta_D, the entire claimed distinct family is removed, even if revocations help accurate reconstruction.

Weak disturbances of inherited material memory may preserve multiple cultural practices predicts instead: In E2 transmit an unfamiliar safe manipulation practice using trained compliant/granular workpieces or a pilot-validated physical medium with measurable cyclic memory. Cross inherited versus freshly reset workpieces with preserved versus absent process demonstrations, whole-cohort turnover, and weak mechanical perturbation versus an equal-duration sham. First measure independent material response curves and novice one-generation learning; use them to predict multigeneration behavior without the proposed extra coupling. Equalize opportunities, force/work budgets where possible, visible cues, immediate task success/payoff, and instruction exposure; assay residual physical differences rather than assuming macroscopically similar objects are identical. predicts a reproducible intermediate-perturbation region where survival of multiple functionally distinct ancestral practices exceeds both zero perturbation and memory-erasing perturbation, while resetting the inherited microstate removes that region. Critically, the gain must exceed the prediction from the separately calibrated physical-memory plus ordinary reconstruction composition, and transfer with the workpiece to fresh learners rather than following experienced participants. Simple noise-induced exploration, stochastic resonance, payoff changes, additional practice and informative wear traces are explicit rival fits. Only Delta_M exceeds its preregistered practical margin delta_M and passes admission; online pairing, routing and revocation residuals are explained by B. If the composed baseline predicts the entire region, reset effect and turnover transfer, retain the physical effect as known ecological/material inheritance and delete F_MATERIAL as a novel family.

SCOUT FROM DISTRIBUTED DATABASES. Candidate set={F_REVOCATION: inherited causal revocation state}; count=1 additional family, with F_CORRELATION, F_DUAL_PATH and F_MATERIAL excluded as incremental novelties in U. The cultural unit is a proposition/action feature together with a source-scoped operation saying that a specified earlier occurrence was withdrawn. Absence alone is not withdrawal, and withdrawal of one source does not invalidate independent support. The proposed extra causal dependency is that descendant production inherits a partial order of add/withdraw operations, which determines whether a later repeated statement reinstates a feature, beyond the producer's current endorsement, remembered correction, familiarity, source reliability and final visible content. Formally let A_g be occurrence-tagged assertions and R_g a set of specifically revoked occurrence tags; merging unions A and R, and the surviving occurrences are A minus R. A later new occurrence with a new tag need not be deleted. This observed-remove rule is an established distributed-data design, not itself a new law. The conjecture is that people spontaneously transmit and use a compressed analogue of this operation state across generations, and that its content-conditioned decay/reconstruction changes cultural persistence beyond an ordinary source-monitoring and correction model supplied with the same evidence. B+F_REVOCATION is sufficient only if causal operation ancestry predicts source-specific resurrection after matched swaps; the smaller set B lacks that dependence. The experiment must not hard-code the rule into the interface and then call correct software execution a human discovery.

Use equal-size semantically relevant source statements as the strongest control, not a blank screen. Later deliver either a stale copy of the withdrawn occurrence or a truly independent new assertion of the same proposition, matched for wording, frequency and source reliability. predicts selective suppression of the stale descendant but reinstatement by independent support, with an additional effect of inherited operation ancestry after fitting source-monitoring/continued-influence/Bayesian evidence models. In fresh cohorts receiving equal final content, removing the causal withdrawal record should raise source-specific resurrection; retaining a record that points to a different occurrence should shift which branch is suppressed. This effect must survive held-out graph schedules and changed paraphrases; ordinary recency or an overall correction-strength difference is insufficient. Only Delta_D passes the common admission rule. If the explicit semantic source-statement control or an established causal source-monitoring model predicts these contrasts within delta_D, the entire claimed distinct family is removed, even if revocations help accurate reconstruction.

Reassignment must preserve low-order correlations up to the order the baseline explicitly models; otherwise ordinary cultural linkage explains it. predicts an excess loss of ancestral joint-semantic recoverability and subsequent lineage survival when the relevant higher-order coupling is broken, with restoration after the original coupling is reinstated; the standalone contribution decoders remain within the preregistered negligible-information band. Fresh cohorts must regenerate the same joint relation without an explicit codebook or any surviving informed participant. Added families F_DUAL_PATH, F_REVOCATION and F_MATERIAL fail their incremental retention criteria. If distributed cue-combination, learned partner conventions, cultural linkage, or fixed secret-sharing codes predict the swap and turnover data within delta_H, F_CORRELATION loses its distinct-family status. If one person's contribution or an external shared record carries the alleged missing relation, the strongest heretical version is falsified, rather than rescued by redefining that carrier as a correlation.

Coupled content and source-validation routes may determine cultural-lineage survival predicts instead: Cross content graph and provenance graph routing while matching the marginal graphs' degree sequences, total opportunities, source-independence counts and contact timing. Use sparse graph realizations above and below a calibrated candidate joint-process threshold; vary only overlap/alignment where possible, and separately test misaligned content and provenance bottlenecks. predicts that interventions preserving each marginal graph's viability can substantially change the survival of source-qualified descendants by changing the coupled Jacobian's leading eigenvalue. The prediction is a finite-network change in the survival curve and lineage size distribution, not a guaranteed universal sharp transition. Raw copying/reach may remain similar while qualified semantic-lineage survival diverges. A ordinary multiplex model with fixed jointly estimated edge weights, or a complex-contagion plus source-monitoring model, must fail held-out graph realignments by delta_N before this becomes a new family. No incremental pairing-code, inherited-revocation or material-memory family passes. If those established compositions predict the effect, or the effect disappears after matching effective source attention and measured route dependencies, retire F_DUAL_PATH as a distinct family even when topology clearly matters.

Weak disturbances of inherited material memory may preserve multiple cultural practices predicts instead: In E2 transmit an unfamiliar safe manipulation practice using trained compliant/granular workpieces or a pilot-validated physical medium with measurable cyclic memory. Cross inherited versus freshly reset workpieces with preserved versus absent process demonstrations, whole-cohort turnover, and weak mechanical perturbation versus an equal-duration sham. First measure independent material response curves and novice one-generation learning; use them to predict multigeneration behavior without the proposed extra coupling. Equalize opportunities, force/work budgets where possible, visible cues, immediate task success/payoff, and instruction exposure; assay residual physical differences rather than assuming macroscopically similar objects are identical. predicts a reproducible intermediate-perturbation region where survival of multiple functionally distinct ancestral practices exceeds both zero perturbation and memory-erasing perturbation, while resetting the inherited microstate removes that region. Critically, the gain must exceed the prediction from the separately calibrated physical-memory plus ordinary reconstruction composition, and transfer with the workpiece to fresh learners rather than following experienced participants. Simple noise-induced exploration, stochastic resonance, payoff changes, additional practice and informative wear traces are explicit rival fits. Only Delta_M exceeds its preregistered practical margin delta_M and passes admission; online pairing, routing and revocation residuals are explained by B. If the composed baseline predicts the entire region, reset effect and turnover transfer, retain the physical effect as known ecological/material inheritance and delete F_MATERIAL as a novel family.

SCOUT FROM DRIVEN DISORDERED MATERIALS. Candidate set={F_MATERIAL: noise-maintained practice repertoire through inherited material microstate}; count=1 additional family. Within U only this additional dependency survives. The unit is a reproducible practice function together with an inherited working object/workspace; action sequence and visible finished form are alternative, separately scored boundaries. Repeated enactment changes the object's latent mechanical configuration. That configuration changes the next novice's reconstruction/search kernel, and their enactment changes the next configuration. The candidate extra dependence is not merely 'artifacts matter': subthreshold perturbations of that inherited configuration can maintain several ancestral practice possibilities that deterministic repeated use progressively erases, despite continued demonstrations of all possibilities. A model has x_(g+1)=G(x_g,a_g,eta_g) and a_(g+1) drawn from K(.|demonstration,x_(g+1),context), where x is measured material microstate, a is performed action, eta is externally randomized mechanical perturbation, G is the material update and K the novice's action-generation kernel. The source-by-microstate-by-perturbation interaction must predict cultural descendants beyond a calibrated material-memory model composed with ordinary affordance learning and niche construction. The one-family set is sufficient if this extra dependence predicts held-out repertoire losses and recoveries; B alone is not. The other three proposed families do not pass incremental admission. No claim is made that internet text without an inherited material substrate obeys this mechanism.

Cross inherited versus freshly reset workpieces with preserved versus absent process demonstrations, whole-cohort turnover, and weak mechanical perturbation versus an equal-duration sham. First measure independent material response curves and novice one-generation learning; use them to predict multigeneration behavior without the proposed extra coupling. Equalize opportunities, force/work budgets where possible, visible cues, immediate task success/payoff, and instruction exposure; assay residual physical differences rather than assuming macroscopically similar objects are identical. predicts a reproducible intermediate-perturbation region where survival of multiple functionally distinct ancestral practices exceeds both zero perturbation and memory-erasing perturbation, while resetting the inherited microstate removes that region. Critically, the gain must exceed the prediction from the separately calibrated physical-memory plus ordinary reconstruction composition, and transfer with the workpiece to fresh learners rather than following experienced participants. Simple noise-induced exploration, stochastic resonance, payoff changes, additional practice and informative wear traces are explicit rival fits. Only Delta_M exceeds its preregistered practical margin delta_M and passes admission; online pairing, routing and revocation residuals are explained by B. If the composed baseline predicts the entire region, reset effect and turnover transfer, retain the physical effect as known ecological/material inheritance and delete F_MATERIAL as a novel family.

Reassignment must preserve low-order correlations up to the order the baseline explicitly models; otherwise ordinary cultural linkage explains it. predicts an excess loss of ancestral joint-semantic recoverability and subsequent lineage survival when the relevant higher-order coupling is broken, with restoration after the original coupling is reinstated; the standalone contribution decoders remain within the preregistered negligible-information band. Fresh cohorts must regenerate the same joint relation without an explicit codebook or any surviving informed participant. Added families F_DUAL_PATH, F_REVOCATION and F_MATERIAL fail their incremental retention criteria. If distributed cue-combination, learned partner conventions, cultural linkage, or fixed secret-sharing codes predict the swap and turnover data within delta_H, F_CORRELATION loses its distinct-family status. If one person's contribution or an external shared record carries the alleged missing relation, the strongest heretical version is falsified, rather than rescued by redefining that carrier as a correlation.

Coupled content and source-validation routes may determine cultural-lineage survival predicts instead: Cross content graph and provenance graph routing while matching the marginal graphs' degree sequences, total opportunities, source-independence counts and contact timing. Use sparse graph realizations above and below a calibrated candidate joint-process threshold; vary only overlap/alignment where possible, and separately test misaligned content and provenance bottlenecks. predicts that interventions preserving each marginal graph's viability can substantially change the survival of source-qualified descendants by changing the coupled Jacobian's leading eigenvalue. The prediction is a finite-network change in the survival curve and lineage size distribution, not a guaranteed universal sharp transition. Raw copying/reach may remain similar while qualified semantic-lineage survival diverges. A ordinary multiplex model with fixed jointly estimated edge weights, or a complex-contagion plus source-monitoring model, must fail held-out graph realignments by delta_N before this becomes a new family. No incremental pairing-code, inherited-revocation or material-memory family passes. If those established compositions predict the effect, or the effect disappears after matching effective source attention and measured route dependencies, retire F_DUAL_PATH as a distinct family even when topology clearly matters.

Inherited withdrawal records may block stale copies while allowing independent support predicts instead: E1 gives matched groups the same final narrative propositions, source confidence, correction comprehension, number/order of ordinary reading opportunities and immediate endorsement, but manipulates which source occurrence a retained withdrawal relation points to. Use equal-size semantically relevant source statements as the strongest control, not a blank screen. Later deliver either a stale copy of the withdrawn occurrence or a truly independent new assertion of the same proposition, matched for wording, frequency and source reliability. predicts selective suppression of the stale descendant but reinstatement by independent support, with an additional effect of inherited operation ancestry after fitting source-monitoring/continued-influence/Bayesian evidence models. In fresh cohorts receiving equal final content, removing the causal withdrawal record should raise source-specific resurrection; retaining a record that points to a different occurrence should shift which branch is suppressed. This effect must survive held-out graph schedules and changed paraphrases; ordinary recency or an overall correction-strength difference is insufficient. Only Delta_D passes the common admission rule. If the explicit semantic source-statement control or an established causal source-monitoring model predicts these contrasts within delta_D, the entire claimed distinct family is removed, even if revocations help accurate reconstruction.