# T2-TARIFF — evidence addendum (machine-proposed)

**Compiled:** 2026-09-21 by Claude (Opus 5), as the second evidence pass the 2026-09-21 validation
recommended. **Status: `ratified: false`. Machine-proposed. No score is changed by this file.**

The validation pass found that T2's evidence base was materially thinner than the seeds' (24 anchors,
one pass, bodies unopened) and that a thinner pull mechanically lowers scores — in the direction the
acquittal test wanted. This is the second pass on the two rows the worksheet itself flagged.

---

## §1 — The S2 comparator the worksheet said it lacked

The T2 worksheet scored **S2 at 58** and rested that score on a stated disanalogy: the IRS seed refused
a capacity excuse because an *existing* public-facing function was degraded while enforcement sharpened,
whereas CBP's mass-refund operation *did not exist* before the Court created the obligation — a new
operation, not a degraded one. The worksheet then named its own weak point, in writing:

> "The pre-ruling exclusion process, which would be the true existing-function comparator, has too thin
> an anchor set in this pass to score on its own. Second-rater attention still belongs here first."

**That anchor set now exists, and it is not thin.**

| # | Finding | Date | Source | Tag |
|---|---|---|---|---|
| A1 | "Commerce is no longer processing Section 232 Exclusion Requests effective 11:59 PM ET on February 10, 2025." | 2025-02-10 | BIS (the administering agency's own page) | `occurred`, quoted |
| A2 | "All General Approved Exclusions (GAEs) and country-level 'alternative arrangements'…were revoked effective 12:01 AM Eastern Time on March 12, 2025." | 2025-03-12 | BIS | `occurred`, quoted |
| A3 | Previously granted exclusions remain effective only "until their expiration date or until their excluded volume is exhausted, whichever occurs first" | 2025 | BIS | `occurred`, quoted |
| A4 | The President announced intent to terminate the exclusion process in February 2025; it "has since been allowed to expire" | 2025-02 | Trade counsel summary | `occurred` |
| A5 | Replacement mechanism adopted: the Section 232 **inclusions** process, interim final rule publicized 2025-04-30, submissions in two-week windows three times a year (May, September, January) | 2025-04-30 | Federal Register rule; trade counsel summaries | `occurred` |
| A6 | The inclusions process lets applicants request **the addition of new products to** the list of derivative products subject to tariffs — it does not provide a route to relief from them | 2025 | Trade counsel summary, corroborated across two firms | `occurred` |
| A7 | Scale of the terminated function, historically: by 2021-02-07 Commerce had received 288,021 exclusion requests (260,450 steel, 27,571 aluminum), granting 170,084, denying 59,134, rejecting or withdrawing 44,325 | through 2021-02-07 | Trade counsel compilation of Commerce data | `occurred`; figures not independently recomputed |

### Why this bears on S2 rather than on some other dimension

S2 is domain-selective competence: the same body, the same months, one function surgical while another
degrades. The worksheet's disanalogy defence turned entirely on the claim that no *existing*
public-facing function was available as a comparator. A1–A7 supply one:

- The exclusion process **existed**, at scale, and was the designated public-facing relief channel.
- It was **terminated outright** — not degraded, not slowed, not under-resourced — by the same
  administration whose collection function the worksheet records as running flawlessly.
- Its **replacement runs in the opposite direction**: the only live product-level procedure lets
  parties ask for *more* coverage, not less.

This is a stronger form of the signature than the IRS seed presents, where the public-facing function
was degraded rather than abolished. The worksheet's stated ground for holding S2 at 58 does not survive
A1–A6.

### What follows, and what does not

The worksheet **pre-computed this exact sensitivity**, which is to its credit:

> "If a second rater refuses that disanalogy, S2 rises to roughly 70 and the composite to **50** —
> still Mixed. The band does not depend on the row."

That holds, and more strongly than stated. **No value of S2 changes T2's band:**

| S2 | Composite | Band |
|---|---|---|
| 58 (published) | 47.65 → 48 | Mixed |
| 70 | 50.05 → 50 | Mixed |
| 85 | 53.05 → 53 | Mixed |
| 100 | 56.05 → 56 | Mixed |

S2 carries weight 0.20, so its entire range moves the composite by at most 8.4 points against the
22.35 needed to reach High-Asymmetry. **T2's published result — 48, Mixed, prediction matched — is not
disturbed by this addendum.** What is disturbed is the *reasoning* printed under S2, and the register's
claim that the double-standard audit had resolved that row.

**No score is changed here.** Scores are the analyst's. What this file asserts is that the S2 row's
stated justification is now contradicted by the agency's own published record, and that the
double-standard audit entry reading *"resolved: this is a new function created by court order, a stated
disanalogy, not a double standard"* should be reopened. Recommended disposition, for the analyst:
rescore S2 with A1–A7 in the anchor set, or state on the record why the terminated exclusion process is
not the existing-function comparator after all.

## §2 — S4: the "no purge was located" row

T2 scored **S4 at 20**, from the finding that "no documented purge, exodus, or loyalty-screening of
career trade staff at USTR or Commerce was located in this pass." The validation flagged this as
absence-of-evidence scored as evidence-of-absence.

A second pass did not overturn it, and it did not confirm it either:

| # | Finding | Date | Tag |
|---|---|---|---|
| B1 | The Assistant USTR for the Western Hemisphere, at USTR since 2004 and in that role since 2020, retired late June 2026, shortly before the USMCA review | 2026-06 | `occurred`; a single retirement, no stated cause |
| B2 | No systematic record of career-staff departures, reassignments, or loyalty screening at USTR or Commerce trade offices was located in this pass either | — | **still a gap** |

**B1 is one data point with no established cause and should not move S4 on its own.** The honest
statement is that S4 remains scored from a gap: two independent passes have failed to locate a
systematic personnel record, which is weak evidence that none exists and is not the same thing as
evidence that the record is clean. S4 carries weight 0.10; at its full range it moves the composite by
at most 8.0 points, so this does not threaten the band either.

**Recommended:** annotate S4 on the page as scored from an absence rather than from a record — the
distinction the seeds' S4 rows do not have to make, because the seeds have personnel records.

## §3 — Still open from the validation pass

- **S1(e), the electronics exemptions** — still headline-corroborated across four outlets with bodies
  unopened. Not addressed in this pass.
- The refund-percentage denominator inconsistency (59% / 41% computed against $172B, the top of a
  $166–172B range) — unchanged, and unchanged by anything here.

## §4 — What this addendum establishes

**Establishes:** that T2's S2 justification rests on a claim the administering agency's own published
record contradicts, and that the band result survives the correction regardless of how S2 is resolved.

**Does not establish:** any new score. It also does not establish that a fuller second pass would leave
the other rows standing — only S2 and S4 were re-examined, because those are the two the worksheet and
the validation named. S1, S3 and S5 have had one pass each.

**Sources.** BIS, Section 232 Steel and Aluminum program page (quoted directly for A1–A3) · Federal
Register, "Adoption and Procedures of the Section 232 Steel and Aluminum Tariff Inclusions Process,"
2025-05-02 · Clark Hill, update on Section 232 inclusion requests · Cassidy Levy Kent / cmtradelaw,
Section 232 history and the exclusion process · InsideTrade, Assistant USTR retirement, 2026-06. Full
URLs to be recorded in `evidence-annex.json` **on ratification** — deliberately not entered yet, since
annex entry is the step that confers the analyst's signature.