# Public Freeze Manifest

## Status

Freeze state for the current Goldbach public package.

This package is ready to present as:

```text
research map / obstruction map / expert-review target
```

It is not ready to present as:

```text
proof / almost-proof / one-estimate-away proof
```

## Public Claim

Use this claim:

> A depth-3 Buchstab minorant route for binary Goldbach isolates a one-`R`-match variance wall inside the Type-III shifted-prime surface `N = p + r s t`.

Use this disclaimer with it:

> This is not a proof of Goldbach. It is a structural reduction and open-problem map.

## Core Public Files

Use these as the public surface:

- `../CURRENT_STATUS.md`
- `HOMEPAGE_SNIPPET_SAFE.md`
- `HOMEPAGE_NOTE_V2.md`
- `GOLDBACH_MAP_PAPER_DRAFT.md`
- `PUBLIC_MAP_STOP_DECISION.md`
- `EXPERT_REVIEW_PACKET.md`
- `DEPTH3_BUCHSTAB_MINORANT_REDUCTION.md`
- `TYPEIII_RESIDUAL_SHEAR_OPEN_PROBLEM.md`
- `PUBLICATION_BUNDLE_INDEX.md`
- `PUBLIC_EXCLUSION_LIST.md`
- `HOMEPAGE_PUBLIC_HANDOFF_2026-04-26.md`

## Supporting Technical Files

Use these only for expert follow-up or internal review:

- `FORMAL_DYADIC_MODEL.md`
- `MAIN_SCALE_AUDIT.md`
- `EXPONENT_BUDGET.md`
- `INTEGER_CORRIDOR_COUNT.md`
- `PARTIAL_MATCH_CORRIDOR_A.md`
- `PARTIAL_MATCH_CORRIDOR_C_AND_REVIEW_QUESTIONS.md`
- `ONE_R_MATCH_LOCAL_MODEL.md`
- `BILINEAR_RATIO_BOUND_LOCAL_RESPONSE.md`
- `SIGNED_KERNEL_DERIVATION_ATTEMPT_L2.md`
- `SIGNED_WEIGHT_PIVOT_STRUCTURAL_RESULT.md`
- `MINORANT_GEOMETRY_PIVOT_TEST.md`
- `RATIO_LARGE_SIEVE_PIVOT_TEST.md`

## Do Not Publish As Claim Surface

Do not publish or foreground:

- any archived Alpha proof-claim draft;
- any phrase implying Goldbach is solved;
- any phrase implying the remaining Type-III estimate is routine;
- any proof-search scratch file that has not been reconciled with `PUBLIC_MAP_STOP_DECISION.md`.

## Current Expert Question

Ask:

> Is the one-`R`-match variance wall structural in this depth-3 Buchstab architecture, or is there a pre-L2 identity that changes the variance object before signs are lost?

Do not ask:

> Which standard theorem finishes this?

The current notes show that standard L2, signed-weight splitting, minorant-geometry adjustment, and ratio-large-sieve import all return to the same wall in this architecture.

## Reopen Conditions

Only reopen the analytic push if one of these appears:

- a coupled pre-L2 signed identity;
- a geometry with moduli beyond product scale `R S`;
- a phase-sensitive formulation that avoids positive square magnitudes;
- a new theorem proving depletion below natural variance for the actual prime/sieve ratio measures.

Otherwise, keep this package frozen as an obstruction map.