P10-L08 · P10 · P10-M02
Audit another report for reproducibility and evidence boundaries
Prerequisites: P10-L07
Learning objectives
- Audit another report for reproducibility and evidence boundaries
- Review a report claim by claim and separate reproducibility defects from unresolved evidence.
Why this matters
Peer review should make a report more inspectable, not reward its most confident narrative. A polished dossier can still hide an unsupported identity bridge or a missing source.
Explanation
Review four layers independently: scope, saved records, transformations and conclusions. Re-run calculations from raw integers, check identifiers against the named network, and trace each material claim to a locator. A broken source link is a reproducibility defect even when the claim may be true. A fully reproducible graph can still be insufficient for human attribution. Neither defect is repaired by visual polish.
Write a finding for each contested sentence: supporting record, exact operation, evidence state, competing interpretation and requested correction. Prefer narrow amendments to blanket verdicts. If a receipt proves a transfer but no control bridge exists, preserve the transfer observation and remove the named-controller conclusion. Record irreducible gaps rather than asking the author to conceal them.
A Diagnostic Passport is a bounded snapshot context, not a certificate of investment suitability, fraud or identity. Review date and source observation date must remain separate. This assessment does not create a real Passport or any certification runtime. The reviewer’s role is to test whether another reader can reproduce what the dossier actually claims.
Key terms
Reproducibility defect: missing locator, bytes or method required to repeat a result.
Evidence defect: a conclusion exceeds what the available records support.
Review finding: precise correction with supporting rationale.
Historical example
ILLUSTRATIVE report R says: “A sent 40units to B; therefore A and B are one owner, and B is safe.” The saved fictional receipt supports40units but R omits its block from the source table. A graph caption equates shared counterparty with shared controller. The reviewer has no identity or safety assessment evidence.
Visual specifications
Peer-review evidence panel with columns sentence, source, transformation, boundary and correction. Retain the 40-unit observation; flag identity and safety as insufficiently evidenced, with no fraud score.
Caption: Fictional peer review audits claim boundaries rather than issuing a safety certificate.
SPECIFICATION_ONLY — the rendered visual has not been produced or independently reviewed.
What the evidence proves
The stipulated receipt supports a 40-unit transfer. The review can identify the missing block locator and two unsupported extensions without denying the underlying transfer.
What the evidence does not prove
It does not prove identity, benign intent, global safety or a genuine ZECOIN Passport. An editorial correction is not new chain evidence.
Common mistakes
Accepting the whole report because arithmetic is right; rejecting every observation because one conclusion is wrong; treating a Passport badge as a guarantee; replacing critique with accusations about the author.
Practical exercise
Produce three review findings for R. Separate a reproducibility correction from evidence-boundary corrections, then draft an acceptable revised conclusion and a remaining-evidence request.
Show worked correction
Finding1: add exact network/block/receipt locator and provenance to make the 40-unit observation reproducible. Finding2: delete common-owner inference unless a separate admissible control bridge is supplied; shared transaction relation is insufficient. Finding3: remove “safe,” because a transfer and a graph do not assess all risk. Revised conclusion: “The supplied fictional receipt records40units from A to B; identity, common control and safety are not established.” Request the missing locator first, then clearly scoped control evidence only if that claim remains necessary.
Checklist
- Re-run compatible-unit calculations.
- Trace every material conclusion to a locator.
- Separate missing reproducibility from overclaiming.
- Preserve valid narrow observations while correcting unsupported extensions.
Summary
Good peer review preserves supported observations and makes unsupported extensions and reproducibility gaps explicit.
Summary
- Correct arithmetic does not validate attribution.
- A missing locator and an excessive conclusion are different findings.
- Review does not manufacture evidence or certification.
Next lesson
Program complete; no new runtime, certification or financial action is implied.
Tools
Use the named ZECOIN tool only as an evidence-reading context. This lesson creates no tool output, account session or entitlement. Offline exercise; do not sign, deploy, approve or fund anything.
Sources & claim boundaries
- BTC: Bitcoin developer guide — transactions — Input references identify previous transaction outputs; scripts do not identify a human. Boundary: Mechanism documentation only; original illustrative values are stipulated, not observations. Historical records have separately frozen provenance and explicit collection gaps.
- API: Esplora public API specification — GET routes, integer satoshis and indexed confirmation/outspend fields. Boundary: Mechanism documentation only; original illustrative values are stipulated, not observations. Historical records have separately frozen provenance and explicit collection gaps.
Evidence classifications
ILLUSTRATIVE — entirely fictional offline inputs; no current observation.
Observation date
null — no real observation
Content version
1
Review date
null
Review status
needs_review
Visual specifications
P10-L08-V01
Audit another report for reproducibility and evidence boundaries
Fictional peer review audits claim boundaries rather than issuing a safety certificate.
Peer-review evidence panel with columns sentence, source, transformation, boundary and correction. Retain the 40-unit observation; flag identity and safety as insufficiently evidenced, with no fraud score.
Peer-review evidence panel with columns sentence, source, transformation, boundary and correction. Retain the 40-unit observation; flag identity and safety as insufficiently evidenced, with no fraud score.
Stack observations, assumptions, gaps and conclusion at 390 px; retain full IDs and a complete text equivalent. Any wide table scrolls locally.
Prose may follow RTL; IDs, quantities and time axes remain LTR. Preserve dependency direction.
P10-L08-ILLUSTRATIVE-v1
Sources & claim boundaries
BTC · PRIMARY_DOCUMENTATION
Bitcoin developer guide — transactions
- Supported claim
- Input references identify previous transaction outputs; scripts do not identify a human.
- Verification boundary
- Mechanism documentation only; original illustrative values are stipulated, not observations. Historical records have separately frozen provenance and explicit collection gaps.
- Checked at
- 2026-10-02
https://developer.bitcoin.org/devguide/transactions.html
API · PRIMARY_DOCUMENTATION
Esplora public API specification
- Supported claim
- GET routes, integer satoshis and indexed confirmation/outspend fields.
- Verification boundary
- Mechanism documentation only; original illustrative values are stipulated, not observations. Historical records have separately frozen provenance and explicit collection gaps.
- Checked at
- 2026-10-02
https://github.com/Blockstream/esplora/blob/master/API.md
Dataset provenance
id: P10-L08-ILLUSTRATIVE-v1
dataStatus: ILLUSTRATIVE
observedAt: null
source: Original offline scenario in realOrHistoricalExample
scope: Fictional stipulated labels and values. No wallet, deployment, transaction, token launch or production observation.