Multiple-Origin Evidence Weighting¶
Weighting model — instantiates Independent Convergence Recognition and Transfer Design
Assembles the heterogeneous evidence for a recurrence — independence, pressure match, sample diversity, negative cases, performance — and weights it into a single graded probability of genuine multiple origin.
The evidence for a convergence claim never arrives in one currency. There is a lineage map, a diffusion screen, a pressure comparison, a pile of negative cases, and some performance data — and someone has to combine them into a single judgment without pretending they are all equally strong. Multiple-Origin Evidence Weighting is that combination step. It gathers the heterogeneous evidence into one set, assigns each strand a weight reflecting how diagnostic and how solid it is, and integrates them into a graded estimate of the probability that the recurrence is genuine independent multiple-origin rather than copying, coincidence, or sampling artifact. Its defining property is that it computes the grade — it is the aggregation engine, the place where evidence of different kinds is made commensurable and combined, as opposed to the place where the result is recorded or the place where any single strand is produced.
Example¶
A strategy team observes that "freemium" pricing — free tier, paid upgrade — recurs across three unrelated software categories and wants to know how confident they should be that this is a real convergent answer to a shared pressure rather than fashion. The weighting model lays the strands side by side. Independence evidence (from the lineage work): strong — different founders, no shared investors. Pressure match: strong — all three face near-zero marginal distribution cost and high customer-acquisition friction, the pressure that makes "let them try free" pay. Sample diversity: moderate — three cases, but all in software, so the domains aren't very independent. Negative cases: a caution — several categories with the same pressures did not adopt freemium. Performance evidence: thin. Weighting these — heavy on the strong independence and pressure strands, discounted for low domain diversity and the live negative cases — yields, illustratively, a moderate grade: real convergence on a genuine pressure, but not the universal law an enthusiast might claim. The number is a weighted argument,[n1] not a measurement.
How it works¶
- Assemble the evidence set. Pull every strand — independence, pressure similarity, sample diversity, negative cases, performance — into one place, each tagged with its source and how solid it is.
- Weight by diagnosticity and strength. Give each strand a weight for how much it discriminates independent origin from the alternatives and how reliable it is; a strong-but-irrelevant strand earns little.
- Integrate to a graded estimate. Combine the weighted strands into an ordinal confidence, keeping evidence that points the other way (negative cases) as a live subtraction rather than a footnote.
- Expose the swing strand. Report which one strand, if it flipped, would most change the grade — so scrutiny and any further data collection target that strand.
Tuning parameters¶
- Weighting scheme — equal weights, expert-set weights, or a more formal likelihood-style combination. Formality buys defensibility; simplicity buys transparency and speed.
- Diversity penalty — how hard non-independent samples (many cases from one domain) are discounted. A steep penalty resists the illusion of many witnesses who are really one; a shallow one over-credits volume.
- Negative-case leverage — how much a disconfirming case pulls the grade down. High leverage is appropriately skeptical; too high lets a single outlier veto real signal.
- Grade granularity — a few ordinal bands versus a finer or numeric scale. Coarse bands match the true precision of the evidence; fine scales imply more certainty than the inputs support.
When it helps, and when it misleads¶
Its strength is that it forces a single honest reckoning across evidence of wildly different kinds and quality, and it makes the reasons for a confidence level inspectable — including which strand is doing the work and which contradicts it. It is what keeps a claim from riding on the one impressive strand while the weak ones ride along unexamined.
Its failure mode is false commensurability: converting soft, unequal evidence into one tidy grade invites treating that grade as a measurement, and hidden weight choices can quietly manufacture whatever confidence was desired. The classic misuse is running it as advocacy — weighting up the favorable strands after the conclusion is chosen. The guarding discipline is to fix the weighting scheme and the negative-case leverage before seeing how the grade lands, and to always publish the swing strand, so the grade is read as a contestable argument rather than a verdict from nowhere.
How it implements the components¶
convergence_evidence_set— it assembles the full heterogeneous body of evidence (independence, pressure, diversity, negatives, performance) into one weighted, source-tagged set.recurrence_confidence_grade— it computes the graded confidence by integrating the weighted strands, exposing the swing strand behind the number.
This model computes the grade but does not record or communicate it — the convergence_claim_record is the artifact built by Convergence Confidence Card, its nearest twin, which stamps and versions the grade this model produces. It also does not itself build the lineage_independence_map or score the pressure_similarity_profile; it consumes those strands from the siblings that produce them.
Related¶
- Instantiates: Independent Convergence Recognition and Transfer Design — supplies the graded confidence the archetype requires before transfer.
- Consumes: Lineage Independence Audit, Pressure Similarity Matrix, and Negative Convergence Case Search supply the strands it weights.
- Sibling mechanisms: Convergence Confidence Card · Pressure Similarity Matrix · Negative Convergence Case Search · Lineage Independence Audit · Hidden Diffusion Checklist · Homoplasy vs. Inheritance Review · Form–Function Decomposition · Analogy-to-Constraint Extraction Workshop · Cross-Domain Transfer Trial
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Multiple-Origin Evidence Weighting operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it assembles the heterogeneous evidence for a recurrence — independence, pressure match, sample diversity, negative cases, performance — and weights it into a single graded probability of genuine multiple origin.
Independent corroboration: The frozen evidence defines Multiple-Origin Evidence Weighting as 'Assembles the heterogeneous evidence for a recurrence — independence, pressure match, sample diversity, negative cases, performance — and weights it into a single graded probability of genuine multiple origin', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Weighting heterogeneous, partially dependent evidence into a graded probability is rooted in statistical weight-of-evidence and Bayesian reasoning.
Related originating lineages:
- Biology & Ecology — Convergent-evolution research supplies the historical problem of distinguishing independent recurrence from inheritance or diffusion.
- History & Historiography — Diffusion and lineage reconstruction supply evidence about whether recurrence reflects copying or independent origin.
- Philosophy — Confirmation theory contributes judgments about diagnosticity, negative cases, and underdetermination.
Review resolution: Both independent reviews agree on primary origin statistics_experimental_design; reconciliation resolves secondary fields (reported_ambiguity, alternate_origin_disagreement). Alternate origins retained (biology_ecology, philosophy, history_historiography) are the union of reviewer-supported formative lineages with explicit rationales, not a list of later application domains. Present-day breadth is represented separately as domain_reach=multi_domain; origin_mode=cross_disciplinary_synthesis records the historical relationship among lineages. Confidence is conservatively reconciled to medium, and encyclopedia_synthesis=true preserves either reviewer's finding that the encyclopedia generalized the mechanism.
Attribution caveat: The specific recurrence-evidence package is an encyclopedia synthesis over inference and convergence research. The exact convergence-weighting model is an encyclopedia synthesis across statistical, evolutionary, and historical evidence.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] Weight of evidence reasoning combines multiple independent indicators into an overall judgment by crediting each according to how strongly it discriminates the hypotheses — the informal cousin of a Bayes-factor combination. The model's grade is a weight-of-evidence summary, only as sound as the weights and independence assumptions behind it. ↩