Evidence Synthesis
Turn a corpus into a traceable conclusion. Treat a conclusion as provisional
unless multiple independent primary sources support it under conditions that
match the question.
The bundled references are modern, reusable reconstructions grounded in the
cited works. Do not present their exact workflows as historical claims that the
named figures stated verbatim.
Select the reasoning references
Read every reference needed for the source shape. Combine references when the
question crosses rows.
| Reference |
Use it for |
| Cochrane |
systematic search, study appraisal, heterogeneity, publication bias |
| Toulmin |
claim, evidence, warrant, qualifier and rebuttal structure |
| Feynman |
rederivation, plain-language integrity checks, cargo-cult detection |
| Darwin |
long-horizon accumulation, difficulty books, strongest contrary case first |
| Laplace |
explicit prior-to-posterior belief updates and sensitivity analysis |
| Gadamer |
interpretation, pre-understanding and charitable reading |
| Strauss |
open and focused coding of qualitative material |
| Geertz |
behavior in context, emic/etic separation and thick description |
Workflow
- Bind the question. State the population or context, intervention or
phenomenon, comparison, outcome, time horizon and decision at stake. Mark
missing elements instead of silently filling them.
- Build a source ledger. Record each primary source, its design, setting,
date, measured outcome, relevant passage or equation, limitations and
independence from the other sources. Trace secondary claims back to the
original source before relying on them.
- Verify comparability. Check that definitions, units, baselines,
populations, versions and experimental conditions match. Keep incompatible
evidence in separate strata.
- Make the inference visible. For every material claim, write its evidence,
warrant, backing, qualifier and strongest rebuttal. Label observations,
author interpretations and your own inferences separately.
- Seek disconfirmation. Search for null results, contrary cases, adverse
outcomes, missing data, publication bias, alternative interpretations and
dependency between nominally separate sources.
- Synthesize at the supported strength. Report agreement and disagreement,
explain heterogeneity and downgrade the conclusion when evidence is indirect,
dependent, incomplete or condition-mismatched. Do not invent priors,
thresholds, effect sizes or missing measurements.
- Audit the draft. Re-derive key steps, explain them in plain language,
verify every citation against the source and list blind spots that remain.
Output contract
Return these sections:
- Question and scope
- Source ledger with primary-source links or precise citations
- Evidence matrix separating observations, interpretations and inferences
- Claim map with evidence, warrant, qualifier and rebuttal
- Synthesis stating what the corpus supports and under which conditions
- Strongest counter-evidence
- Uncertainty and blind spots
- Decision implications and the evidence that would change them
- Refusals or unresolved items
Use calibrated language. Prefer “the inspected evidence supports” to a universal
claim. Say “I don't know” when the available record cannot answer the question.
Refuse or stop when
- The requested conclusion is fixed in advance and disconfirming evidence is
out of scope.
- A material claim cannot be traced to a primary source.
- The source text, data, equation or experimental conditions cannot be checked.
- Nominally multiple sources are not independent and no independent
corroboration exists.
- Quantitative synthesis would require invented values or unjustified model
choices.
- Privacy, consent or safety constraints make the requested evidence use
improper.
When stopping, name the missing evidence and the smallest verification step
that could unblock a defensible answer.
1---2name: evidence-synthesis-23description: Synthesize multiple papers, benchmarks, interviews, reviews, or other sources into a defensible and qualified conclusion. Use for conflicting evidence, literature synthesis, claim auditing, argument mapping, qualitative coding, belief updates, or any request that must expose counter-evidence, uncertainty, and source limitations.4---56# Evidence Synthesis78Turn a corpus into a traceable conclusion. Treat a conclusion as provisional9unless multiple independent primary sources support it under conditions that10match the question.1112The bundled references are modern, reusable reconstructions grounded in the13cited works. Do not present their exact workflows as historical claims that the14named figures stated verbatim.1516## Select the reasoning references1718Read every reference needed for the source shape. Combine references when the19question crosses rows.2021| Reference | Use it for |22|---|---|23| [Cochrane](references/cochrane.md) | systematic search, study appraisal, heterogeneity, publication bias |24| [Toulmin](references/toulmin.md) | claim, evidence, warrant, qualifier and rebuttal structure |25| [Feynman](references/feynman.md) | rederivation, plain-language integrity checks, cargo-cult detection |26| [Darwin](references/darwin.md) | long-horizon accumulation, difficulty books, strongest contrary case first |27| [Laplace](references/laplace.md) | explicit prior-to-posterior belief updates and sensitivity analysis |28| [Gadamer](references/gadamer.md) | interpretation, pre-understanding and charitable reading |29| [Strauss](references/strauss.md) | open and focused coding of qualitative material |30| [Geertz](references/geertz.md) | behavior in context, emic/etic separation and thick description |3132## Workflow33341. **Bind the question.** State the population or context, intervention or35 phenomenon, comparison, outcome, time horizon and decision at stake. Mark36 missing elements instead of silently filling them.372. **Build a source ledger.** Record each primary source, its design, setting,38 date, measured outcome, relevant passage or equation, limitations and39 independence from the other sources. Trace secondary claims back to the40 original source before relying on them.413. **Verify comparability.** Check that definitions, units, baselines,42 populations, versions and experimental conditions match. Keep incompatible43 evidence in separate strata.444. **Make the inference visible.** For every material claim, write its evidence,45 warrant, backing, qualifier and strongest rebuttal. Label observations,46 author interpretations and your own inferences separately.475. **Seek disconfirmation.** Search for null results, contrary cases, adverse48 outcomes, missing data, publication bias, alternative interpretations and49 dependency between nominally separate sources.506. **Synthesize at the supported strength.** Report agreement and disagreement,51 explain heterogeneity and downgrade the conclusion when evidence is indirect,52 dependent, incomplete or condition-mismatched. Do not invent priors,53 thresholds, effect sizes or missing measurements.547. **Audit the draft.** Re-derive key steps, explain them in plain language,55 verify every citation against the source and list blind spots that remain.5657## Output contract5859Return these sections:60611. **Question and scope**622. **Source ledger** with primary-source links or precise citations633. **Evidence matrix** separating observations, interpretations and inferences644. **Claim map** with evidence, warrant, qualifier and rebuttal655. **Synthesis** stating what the corpus supports and under which conditions666. **Strongest counter-evidence**677. **Uncertainty and blind spots**688. **Decision implications** and the evidence that would change them699. **Refusals or unresolved items**7071Use calibrated language. Prefer “the inspected evidence supports” to a universal72claim. Say “I don't know” when the available record cannot answer the question.7374## Refuse or stop when7576- The requested conclusion is fixed in advance and disconfirming evidence is77 out of scope.78- A material claim cannot be traced to a primary source.79- The source text, data, equation or experimental conditions cannot be checked.80- Nominally multiple sources are not independent and no independent81 corroboration exists.82- Quantitative synthesis would require invented values or unjustified model83 choices.84- Privacy, consent or safety constraints make the requested evidence use85 improper.8687When stopping, name the missing evidence and the smallest verification step88that could unblock a defensible answer.
cdeust