[ PROOF SURFACE / COMPUTATION ]

Genesis Determinism Surface

This surface emphasizes reproducibility receipts and keeps Genesis-DM3 license boundaries intact.

visual panel

genesis-compute-fixture

interactive module

Source substrate, cell receipts, parity notes, and open experiment gates remain in fixed order.

Replay depth substrate 74 receipt chain

related surfaces

source metadata

what it shows

**Bit-determinism on RM10 (Phase 0 BITDET cells)** — 4 cells × 6 cores × {10, 50, 200, 5000} iterations = **31,560 cross-checked `verify.json` hashes, all equal to source-canonical `97bd7d…`**. `unique_canonical_sha_count = 1` per cell. Mirror of dm3_runner's claim ξ.
**Cross-platform canonical match (M1 ↔ RM10)** — the same `snic_rust` pipeline produces byte-identical `solve_h2.json` SHA `62897b…` on both `aarch64-apple-darwin` (M1 host) and `aarch64-linux-android` (RM10). Hash-gate disposition for `verify.json` documented at [`harness/host/HASH_GATE_DISPOSITION.md`](harness/host/HASH_GATE_DISPOSITION.md): the M1-side `e8941414…` vs source-hardcoded `97bd7d…` diff is a serialization-layer trailing-newline artifact (`VERIFY_SUMMARY.json = 8ddb…` reproduces identically; all 7 internal gates pass; `solve_h2.json` matches exactly). Mirror of dm3_runner's claim τ at the canonical-pipeline level.
**Source-canonical match on RM10** — RM10 cross-compiled `snic_rust` reproduces source-hardcoded `CANONICAL_VERIFY_HASH = 97bd7d…` and `CANONICAL_SOLVE_HASH = 62897b…` byte-exact (no BENIGN diagnosis required on the device side; Phase 0 plan 00-06 verdict).
**All 7 internal verification gates pass** on every Phase 0 invocation: `gates_ok`, `dep_cert`, `gc_invariants`, `lift`, `stab`, `cad_sos`, `egraph`.

what it does not claim

Genesis is not a codec, productized service, unified platform, or portfolio-wide architecture.
Genesis does not prove a general computation substrate beyond the pre-registered v1.0 comparison surface.
The D6-vs-C3 comparison is not numerically closed; it has a structurally complete analytic disposition and v2.0 numerical work remains.
The flat sigma curve does not by itself decide substrate-easy versus pattern-degenerate explanations.

proof anchors

dossier proof files

reports/GENESIS_FINAL_REPORT_2026-05-01.md :: v1.0 final synthesis: four comparison verdicts, determinism scorecard, analytic Z₂ disposition, settled/deferred/falsification path
PRD_GENESIS_COMPARATIVE_v1_DRAFT_20260427.md :: Operator's source-of-truth PRD; four pre-registered comparisons; test program; deployment recipe; boundaries
project_contract.json :: 12-key formal contract: 6 claims, 7 observables, 9 acceptance tests, 7 forbidden proxies, 9 references, 6 deliverables; uncertainty markers
LANE_DISTINCTION.md :: Formal separation of Genesis (285v, D₆) from dm3_runner (380v, C₃, parked); Resolution clause for Phase A anchors
proofs/manifests/CURRENT_AUTHORITY_PACKET.md :: Current authority packet: substrate, hashes, 74 receipt counts, verdicts, open items
proofs/artifacts/sigma_curve_full.tsv :: Aggregated Phase 2 + 2.5 + Phase 3 prep + parity-sweep σ″ curve table; all 74 cells PASS

dossier metrics

Phase 0 verify hashes: 31,560 / 31,560 byte-identical; RM10 Android aarch64
Receipt cells: 74 / 74 PASS; in-repo proof surface
K2 cross-platform parity: S30 byte-identical; RM10 vs M1 host
Pre-registered comparisons: 3 settled empirically; 1 analytic/deferred; v1.0 comparison surface