FICTIONAL EXAMPLE ONLY — annotated validator report
Example 1.0 | Validator 1.0 | Receipt template 1.0 | Methodology 1.0

All people, sources, dates, observations, outcomes and any populated review actions
are fictional. The input deliberately contains two wrong calculations and three
missing review fields. It assesses no canonical record and is excluded from every
real forecast aggregate. Its example.invalid URLs are placeholders, not evidence.

Files

assessment-receipt-validator-fictional-input-v1.json — intentionally flawed input.
assessment-receipt-validator-fictional-report-v1.json — exact CLI output, unannotated.
assessment-receipt-validator-fictional-report-v1-notes.txt — this explanation.

Reproduce

Save the input and the unchanged validate_assessment_receipt_v1.py in one directory.
Run with Python 3.10 or newer from that directory:

  python3 -B validate_assessment_receipt_v1.py assessment-receipt-validator-fictional-input-v1.json

The expected exit status is 1. The tool prints the report and does not write input
files, retrieve sources or contact the ledger. These notes are separate from the
report so its JSON remains the validator's actual output. The file field depends
on the path supplied; use the basename above to reproduce the downloaded report.

Reading field paths

$ is the receipt root. A dot enters an object; [0] means the first array item.
For example, $.calculations[1] identifies the second calculation.

ERROR 1 — $.calculations[0]

Message: Result does not reproduce from the supplied inputs

The fictional Brier calculation supplies p = 0.8 and y = 1 but reports 0.05.
The fixed arithmetic gives (0.8 - 1)^2 = 0.04. This is a rule violation, not a
missing value. In this fictional scenario the demonstration repair is 0.04.
For a real receipt, first check the probability, outcome and evidence; do not
change a result merely to make validation pass. This check does not verify them.

ERROR 2 — $.calculations[1]

Message: Result does not reproduce from the supplied inputs

The fictional event is declared met and supplies y = 1, but reports fulfillment
of 0.45. An indivisible resolved event has fulfillment exactly 1 if met or 0 if
missed. In this fictional scenario the demonstration repair is 1. The tool does
not turn a discretionary partial score into an event outcome or authenticate
the fictional met finding.

INCOMPLETE 1 — $.review.second_reviewer

Message: Required for a complete assessment

The second reviewer is null. A real assessment needs a distinct second reviewer
after review has occurred. Leave unknown information null; never invent a name.
This is missing information, not the arithmetic violation shown above.

INCOMPLETE 2 — $.review.reviewed_at

Message: Required for a complete assessment

The second-review date is null. Record its actual date or timezone-qualified
timestamp after review. Do not backdate it or substitute the assessor's date.

INCOMPLETE 3 — $.review.reproduced_from_receipt

Message: Required for a complete assessment

The reproduction flag is false. The required review has not been confirmed.
Set it to true only after the reviewer reproduces the result from the receipt.
Changing the flag alone does not perform review or prove who reviewed it.

Report flags

passed: false — this run fails because it has errors, even without --require-complete.
valid: false — two rule violations were detected.
complete_for_human_review: false — this is fictional and also has missing fields.
errors: the two calculation violations above.
incomplete_fields: the three missing review requirements above.
warnings: this is fictional; no source retrieval, factual verification, interpretation
  approval or publication authorization is performed.
ledger_writes: 0; network_requests: 0.

After the errors are corrected

Correcting only the two fictional calculations removes the errors. Ordinary
validation then passes with valid = true while the same three fields remain
incomplete. --require-complete still fails.

Even if all fictional review fields are populated, complete_for_human_review
remains false: receipt_kind is hypothetical_example. The demonstration is never
a real assessment and cannot pass --require-complete. Keep example_only status,
null canonical record_id and excluded aggregate treatment; do not relabel it to
make it pass.

Incomplete real drafts may pass ordinary validation. Real receipts declared
published or included fail if required fields remain incomplete. Neither mode
authenticates sources, verifies interpretation or grants permission to publish.

The validator can stop after an earlier structural error, so a report is not an
exhaustive list of every problem. Re-run after repairs and keep human evidence
review separate. This example changes no ledger record, score or eligibility.