# SYNTHETIC revisit, completed (fictional, not law)

The second half of the amendment demo: the accepted candidate under a
fictional rate change, carried through to a new passing verdict. Nothing
here amends any act; the altered rule is labeled synthetic in its own file,
and the real package is untouched.

## The change

Class-22 rate 2.96 → 3.10 percent, one rule line in a copy of the accepted
candidate (`08-17-19-core.law`, TariffClass22). Every other shared file
starts byte-identical to the candidate.

## Detection

Re-running the candidate suite unchanged against the flipped rule:

```text
  FAIL [kz.corpus.employee_accident_insurance#authored] tests/core.lawtest / EAI-MINING-CLASS-22-PREMIUM
  FAIL [kz.corpus.employee_accident_insurance#authored] tests/tariff.lawtest / EAI-TARIFF-CLASS-22-PREMIUM
  FAIL [kz.corpus.employee_accident_insurance#authored] tests/premium-base.lawtest / EAI-PREMIUM-SUM-DISTINGUISHES
total: 37 checked, 34 passed, 3 failed, 0 not run; code 1
```

Exactly the three class-22 premium checks fail; the other thirty-four —
including the loss-30 payout control — survive. Observed with tool `law`
0.1.0 on a scratch copy; this directory holds the resolved state below.

## Resolution

EAI-D1 reopens with the rate. New expectations, derived — not copied
from the run: 0.0310 × 1,000,000 KZT insured sum = 31,000 KZT on the
same million, and 0.0310 × 2,000,000 KZT = 62,000 KZT on the doubled
sum. All three class-22 expectations move (29600 → 31000 twice,
59200 → 62000 once) citing the reopened record; EAI-D2 and EAI-D3 stand
untouched with green checks on both sides. Scope is unchanged: the same
five branches stay covered.

## New candidate

This directory is the post-amendment candidate. Pinned tool (`law` 0.1.0,
semantics law.core/0.2, published build). Run from the unpacked bundle
root (in the author repository the directory is
`docs/handbook/files/fixtures/amendment-revisited/`):

```sh
$ law engine check files/fixtures/amendment-revisited
check OK: files/fixtures/amendment-revisited
$ law test files/fixtures/amendment-revisited
total: 37 checked, 37 passed, 0 failed, 0 not run; code 0
```

Snapshot id over the fifteen files (same recompute as the candidate
README): `02469c81a757ccd8d7bce32cbdae750725f62e4d0f7518b551056abe1324da83`
(pre-R3 snapshot: `f3403a8e…8012b3b` over 14 files).

Verdict: accept-synthetic with recorded limits — the arithmetic is
verified, the amendment itself is fiction, and no real source was
re-pinned. A real amendment would additionally need fresh source bytes,
fresh pins, and a fresh semantic read before any verdict.