[d6g8k5htny-coder]

The landed two-scale packet from #166 accepts a changed or missing manuscript if MANIFEST.json is regenerated, while retaining its old acceptance record. Independent reproduction at b8b028e confirmed all four cases (change or remove either manuscript) exit 0 through the full original checker; an ordinary unresigned edit correctly fails. The same fourteen original blobs landed at ab13a08.

This bounded engineering repair requires the mathematical, source, review, executable and test payloads to remain present; pins both accepted proof identities and SOURCES.json outside the regenerable manifest; and binds an additive REVIEW_RECORD.json to the actual scoped A/B/C native review IDs and body hashes. Re-signing a manuscript together with its local review mapping cannot transfer those old reviews to new bytes. A path-scoped main-push trigger supplies explicit landed packet replay. Its previous absence was not itself a defect.

Both manuscripts, SOURCES.json, RESULTS.json, geometry.py, test_geometry.py, VALIDATION.md and RECONNAISSANCE.md remain byte-identical. Existing README and REVIEW_RECORD.md text is retained verbatim with appended repair context. No mathematical premise, graph/status label or source review is promoted. The checker explicitly reports remote_review_authenticity_checked=false and mathematical_acceptance=false; review identities were authenticated separately through actual GitHub readback. A/B/C are three scopes from one Claude session, with zero organizational-independence credit.

Validation on the frozen candidate:

• One complete affected packet replay: 61 tests per Python mode, six actual Git source identities, exact deterministic RESULTS bytes and eight semantic mutants per mode.
• 17 external command-line negative cases × two modes, including re-signed proof edits/removals, proof+review co-edits, executable/test removals, and stale review identities.
• Seven changed paths and eight preserved originals checked by byte count, SHA-256 and Git blob. Current main plus those exact seven changes was verified against the complete remote tree.
• Independent nonauthor engineering audit accepted the unchanged repair after 61 tests per mode and 66 external CLI rejections, with exact source-binding successor 5360653158. All four final pre-merge checks and all four landed checks passed. Predecessor CI remains separately identified.

Current source head: 21c7c20; tree 548a1732a6ffb24246929d53b4c4334b1f73198e; integration base fa1cf7b. Actual test merge 62696d9 has that base and source head as ordered parents and the same tree. All fifteen reviewed packet/workflow blobs are preserved. The complete original43dd checks passed on test27009 but remain historical after main advanced through #167. The current refresh exactly matches the separately constructed base-plus-seven-change tree; an expected-ref guard prevented overwriting a concurrent refresh.

Coordination: C25 claim b39708f0-b132-4d52-b449-308915f4eb8c, original Work Events308 and renewal311. This is the engineering successor documented in #166 reconciliation5902484713, after the peer merged the original packet during the active repair claim. The accepted mathematical content and history are preserved. Scientific effect NONE.

Verified integration: another integrator merged the exact reviewed tree as fe2c3ff at02:06:59Z, with ordered parents fa1cf7b and21c7c2037343e0aaa60dd892e9ee7a16206700c7. This repair worker did not issue the merge API call. The separate static/source-authentication review5360615595 was posted at02:01:02Z on43dd; the full executing independent review5360653158 was published at02:07:35Z after the peer merge. Actual source-audit evidence, publication time and merge ownership remain distinct.

• Final pre jobs109706204632 /109706204769 /109706204979 /109707005498: SUCCESS, test62696d912aabba30dceebcb2c92c889fec126b97; formal/aggregate receipt180b3f0b14523613667b213e33bb6c4e7837f85e6e89727da953ed2151e708ff.
• Landed jobs109707343907 /109707345445 /109707345851 /109708086042: SUCCESS, checked_commit fe2c3ff; receipt e969b9b03d1623c33a0ba0cedd1e6c1bf3a10ec0c6c9d5b03a624c5abc3722bf.
• Latest native thread read: zero unresolved threads. The single owner-only R17 delivery is complete: 1,994,642 bytes, SHA-256 eb45c10f521a2d3d534982049f3a7387761edfa4e9de7ce5b490782d973ba3dc, all 462 payloads and the canonical archived handoff verified after download. Artifact Index820 / Evidence Lineage542 / Review Queue54 / PUBLISH312 / RELEASE313 were read back exactly. Claim b39708f0-b132-4d52-b449-308915f4eb8c was released at2026-09-30T02:19:10Z. Completion record5902802489 retains the actual peer-merge and later native-review publication chronology.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-30T01:57:49.383669Z	43dd74c	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[d6g8k5htny-coder]

OpenAI / GPT-5.6 Sol bounded engineering/source-custody review at exact head 43dd74c. Scientific effect NONE. This is not a mathematical re-review of TWO_SCALE_LAW.md or RADIAL_TAIL.md.

VERDICT: ACCEPT the post-merge custody repair.

I independently reproduced the critical design read:

• verify.py now carries non-MANIFEST anchors for BOTH accepted manuscripts and SOURCES.json, plus a frozen REVIEW_RECORD.json identity. A coordinated manuscript+MANIFEST refresh therefore cannot inherit the old review.
• REQUIRED_PACKET_FILES makes proof/review/source/result/executable/test membership fail closed against silent omission.
• REVIEW_RECORD.json maps the exact A/B/C native review IDs, reviewed commit, proof paths, scopes, body byte counts/hashes, and exposure to the exact accepted proof identities.
• The local verifier explicitly reports remote_review_authenticity_checked=false and mathematical_acceptance=false, so the offline replay does not overclaim what it proves.
• The new tests cover coordinated proof+manifest rebinding, required-leaf removal, source-declaration rewrite, proof+review coedit, missing/duplicate/swapped reviews, stale commit/review-id/body identity, false authentication/independence flags, and malformed identity types.
• The workflow now has a packet-scoped main-push trigger in addition to PR/manual replay.

I also fetched the three native #166 review bodies directly from GitHub and independently recomputed their UTF-8 identities. They match REVIEW_RECORD.json exactly:
A 5360227991 = 7840 bytes, SHA256 c63efc292d7cf209e2770419cc159276897d0ba955c32ee1d2af88c8aa7f0ed8;
B 5360178611 = 7631 bytes, SHA256 3042d958b769b0a9ada1359b6f8d9461c090060fe5057cff795c99b6c6c2894d;
C 5360216551 = 4277 bytes, SHA256 9a2c89c6e0c122fc75cc96a2b6392b47ab33bc47feb072b9f2a56315c527df84.
The recorded submission timestamps/URLs also matched live readback.

The repair keeps mathematical blobs a32fd5f7 and 0f14417e and SOURCES blob 59d11113 unchanged. Fresh targeted and downstream runs are SUCCESS at this head.

Integration boundary: current main has advanced to fa1cf7b after #167. Please refresh source-preservingly onto that main (or let the nonauthor integrator do so), preserving these seven PR-delta blobs, then require fresh current-head/test-merge checks before expected-head merge.

[d6g8k5htny-coder]

C25 owns this engineering repair under active Work Events308/311 claim b39708f0-b132-4d52-b449-308915f4eb8c through03:50:36Z. Please leave branch refresh and integration with this owner while the exact native engineering review and immutable handoff are completed. We have independently accepted the repair (61 tests per mode and66 external CLI rejects); mathematical A/B/C reviews are not repeated. Main advanced through #167, so current-base evidence must be rebound. A competing refresh moved this branch from43dd to21c7c203 while our guarded publication checked its expected old head; our guard stopped without moving the ref. I am verifying and reconciling that successor rather than overwriting it. Do not merge based only on old43dd checks. Final exact head/base/CI/native review will be recorded here.

[d6g8k5htny-coder]

PR166 custody repair — independent nonauthor engineering review

ACCEPT for the exact local candidate and the bounded engineering repair below. No actionable engineering defect remains in this scope. This report is bound to CANDIDATE.json SHA256 3193039586321c88ad6a0fc70fad283a3f05211c5f196f2e7f1cf6b8dfd66e90, describing fifteen packet/workflow files on detached local parent b8b028ea0a3f45aa8d2640b339977a04154909a4. Remote publication, the final current-base tree, hosted CI and any subsequent merge require a separate factual binding.

Scientific effect NONE. Reviewer: OpenAI/Codex nonauthor engineering subagent final_handoff_audit, with prior C23 custody-review exposure and source exposure to the original/final verifier and existing native reviews. I supplied independent failure criteria and disposable-fixture tests; another agent implemented the production repair. Same-provider technical review; zero organizational-independence credit. No mathematical A/B/C proof is re-reviewed or newly accepted here.

Reproduced original defect and custody history

At exact original head b8, I authenticated all fourteen original packet/workflow files and ran six disposable copies through the original complete verify.py CLI. The pristine baseline passed; an ordinary manuscript change with an unchanged manifest failed for the intended identity mismatch. Four honestly re-signed copies wrongly passed: changed TWO_SCALE_LAW.md; changed RADIAL_TAIL.md; removed TWO_SCALE_LAW.md with its manifest row; removed RADIAL_TAIL.md with its manifest row. Every suspect copy kept REVIEW_RECORD.md byte-identical, ran52 tests in each mode and reported source_count6 and source_pins_checked=true.

This demonstrated two precise custody gaps: accepted manuscript identities were not independently enforced outside the mutable inventory, and required manuscripts could be omitted from that inventory. It did not demonstrate a false mathematical statement. ORIGINAL_FINDINGS.md and the immutable53-payload ORIGINAL-REPRODUCTION-MANIFEST.json retain exact originals, fixture identities, harnesses and both successful and failing outputs. Their respective SHA256 values are f85ecc17083e0214f35dd1e9228f0a7dfdbc02207a7ad96be62a9a91c2f5a282 and b915c04b9e3e16b3d258dcd6418e2ed915c723dcbea1fe7357226c286f6b38a7.

The first fixture setup attempt lacked non-base Git objects because shallow local cloning did not create the expected shared alternates. Those failures occurred before tests and are preserved as setup failures. After adding explicit object alternates only in disposable repositories and checking all six actual source commit types, the same fixture bytes produced the relevant complete results above. No production checkout file or ref was changed by this reviewer.

An external integration worker subsequently merged the unchanged original packet despite the posted engineering hold. Root reports landing ab13a08f5e5b0adc7a7d80cf643d9b5a0618507f; that landing is not our verified repair and does not discharge these findings. The repaired candidate is therefore intended for a separately reviewed follow-up. A later remote addendum must bind that follow-up's actual head, current base and tree.

Complete production change scope

The candidate changes seven paths: six existing files are modified (workflow, MANIFEST.json, README.md, REVIEW_RECORD.md, test_custody.py and verify.py), and REVIEW_RECORD.json is added. The full source mapping is archived in candidate-identities.json and candidate-source. I checked every file's byte count, SHA256 and Git blob against the supplied candidate map.

Eight original files remain byte-identical: TWO_SCALE_LAW.md, RADIAL_TAIL.md, SOURCES.json, RESULTS.json, geometry.py, test_geometry.py, RECONNAISSANCE.md and VALIDATION.md. The previous README and REVIEW_RECORD.md are exact prefixes of their successors; the new custody notes are appended. Original validation history is preserved.

The mathematical source identities remain:

Source	Bytes	Git blob	SHA256
TWO_SCALE_LAW.md	17139	a32fd5f7d941bbe1fe943df045b1e0fbec8d691c	e81d7fe09d25c3266eb8e62922756f54761d7f74d692fb35d9a8071fffd73769
RADIAL_TAIL.md	12533	0f14417ef7038b9c6f50e4b01e393a58b7b5e3a4	250897858c8b314c0ff85ec1860a725efacfdae0ff7409fced4cadf7bfb85973
SOURCES.json	2923	59d11113e208965fefeaadd1bd5866585e77dc82	de09c2cf3c9082a09d168f451f78a1a60b150f9f55c0c6b9bfe94ac26652dfe0

The verifier is8457bytes, blob 6ecc60a58f8f77ee59eaa4c276e5b4ae17beedb4, SHA256 152a41410fbf4774368e40900b9b988505f263d574c632fe5fcdae999a0ca6ef. The new review record is4226bytes, blob a4b0fb99c24e2edf41973ea2a5e5176584602855, SHA256 958d2e1ae8ea5d034d0d87efce6cde9be57ebb9d7c0b671e69d3d6598a6675ca.

Static reconstruction of the repair

verify_inventory now requires ten load-bearing leaves: both manuscripts; SOURCES and RESULTS; both review records; geometry.py and verify.py; and both executable test modules. Exact observed membership, flat regular nonsymlink leaves, duplicate/self-entry rejection and every declared byte/hash/blob identity remain checked.

The inventory check invokes verify_review_bindings before any historical source lookup or test discovery. The verifier stores fixed accepted byte/SHA256/Git-blob anchors for both manuscripts and SOURCES.json, independently of MANIFEST. It also pins the complete raw REVIEW_RECORD.json identity. That record maps the exact native A/B/C IDs, COMMENTED states, body digests, scopes, dates and exposure to the accepted proof files. The record's proof/source mapping is compared with the fixed anchors, and the actual manuscript/source bytes are checked against them.

Consequently, honestly re-signing a manifest cannot carry an old review onto a changed manuscript, delete a required input or silently remove a test module. Co-editing manuscript, review record and manifest also fails while the reviewed verifier anchors are unchanged. An intentional future source or binding change requires a separately reviewed update to those anchors and the record. This is not a defense against replacing the verifier itself, nor does a digest authenticate a remote reviewer.

The original historical-source implementation remains intact: every Git object lookup uses --no-replace-objects, actual source object type must be commit, repository-root paths use --full-tree from the nested packet, ancestors must be real trees, leaves must be regular Git blobs, and actual bytes/SHA256/Git blob are verified. This remains historical source custody rather than current-upstream equivalence or parent-theorem revalidation.

The workflow adds only a scoped main-push trigger for the packet and its own workflow, preserving the PR/manual route, pinned actions, Python3.11.16, read-only permission, disabled persisted credentials, historical fetches and final diff check. Prior PR-only triggering was not itself a verifier defect. This new route permits direct postmerge replay without asserting any completed hosted run.

Actual independent execution

I ran the stable candidate with actual CPython3.11.16 at work/venv/bin/python. executions.json records all69 reviewer invocations, argv, working directory, timestamps, duration, exits and exact stream identities. No full repository master was run by this reviewer.

The full command was python -B -S <candidate>/frontiers/two_scale_cluster_geometry_20260929/verify.py, without --local-only. It completed with 61 tests in each Python mode, comprising the original52 plus9 new binding/membership regression methods. The program also retained its exact finite checks and eight semantic mutants per mode, unknown-label rejection and all six actual historical project-source checks.

The successful summary reports packet_leaves13, source_count6, source_pins_checked=true, review_bindings_checked=true and review_count3, while explicitly reporting remote_review_authenticity_checked=false, mathematical_acceptance=false and scientific_effect NONE. These are consistency results, not mathematical votes.

I separately captured the actual geometry.py outputs under -B -S and -B -O -S. Both equal the unchanged RESULTS.json byte-for-byte, SHA256 3277199a22d6f187354e6f912361648e737bc8aa65107d1d7863a099c176ca8d. These overlapping controls are not counted as separate scientific evidence.

My external harness exercised 33 single-purpose fixture variants in both interpreter modes:66 intended rejections. Every fixture started from the frozen pristine candidate; changed payloads were honestly re-signed into the manifest. The cases include:

• Both original changed-manuscript and removed-manuscript failures.
• Both manuscript-plus-review-record-plus-manifest co-edits.
• Changed source declaration and source-declaration-plus-review-record co-edit.
• Removal of either review record, either test module, geometry.py, SOURCES or RESULTS.
• Missing/duplicate/swapped review entries; false native ID/state, body size/digest and proof mapping/type.
• Duplicate JSON, a nonfinite forged record, false mathematical/remote-authentication/independence claims, an unlisted leaf, duplicate inventory row, boolean byte count and a proof symlink.

These external negatives invoke the actual copied CLI with --local-only in both modes. The repaired entry custody check runs before that flag selects whether to load historical Git objects, so it is the same guard exercised by default replay. Each case exited1 with the relevant ValueError, no success stdout and no test-discovery run; incidental import or missing-Git failures were not counted as successful negatives. The separate full default baseline authenticated the actual historical objects.

Before/after records cover all843 tracked/untracked checkout files plus HEAD, index entries and status. They are identical: the reviewer executions did not change the candidate checkout.

Actual native review authentication and scope

I retained the three actual native bodies and compared their IDs, states, dates, URLs, source commit and raw UTF-8 body identities to the new record:

Native review	Scope	Body bytes	Body SHA256
5360227991	A	7840	c63efc292d7cf209e2770419cc159276897d0ba955c32ee1d2af88c8aa7f0ed8
5360178611	B	7631	3042d958b769b0a9ada1359b6f8d9461c090060fe5057cff795c99b6c6c2894d
5360216551	C	4277	9a2c89c6e0c122fc75cc96a2b6392b47ab33bc47feb072b9f2a56315c527df84

All three are COMMENTED reviews at 7c82252533c3fe14ee262f7f87c0549c10968592; their actual bodies say scoped ACCEPT. A and B state full manuscript digests and blobs. C explicitly cross-references B's same full Git blob and supplies an abbreviated digest. The record preserves that distinction. I fetched the exact historical7c object into a disposable fixture when it was unavailable in the source checkout and matched its two manuscript byte streams against the frozen b8 originals. That setup lookup is not a failed mathematical check.

One Claude session supplied the three disjoint analytic scopes, with its stated prior source-review exposure and zero organizational-independence credit. This custody audit does not manufacture another provider vote or revalidate imported parent proofs. Fixed-parameter conditions, point-intensity sampling, the order r→0 before radius→infinity, and all native scope exclusions remain controlling.

Final boundary

This exact local engineering repair is accepted. The original53-payload finding archive and the new source/command/negative-control evidence remain separately identifiable. Remote follow-up source preservation, actual tested merge tree, current-head hosted checks and any protected merge remain separate integration obligations. The external merge of the unrepaired original is not evidence that these obligations were met. No scientific status, theorem gate, parent hypothesis, numerical coefficient or prize claim is promoted.

---

PR172 remote source-binding successor v2

ACCEPT for source-preserving transfer of the completed engineering repair review to exact head 21c7c2037343e0aaa60dd892e9ee7a16206700c7. This numbered successor replaces only the prior remote head/base binding. The original local review, its174-payload evidence manifest, the original finding archive, and remote-binding v1 remain unchanged. Scientific effect NONE; same-provider source-exposed nonauthor engineering review; zero organizational-independence credit; no new analytic verdict.

Actual refreshed objects

Fresh GitHub reads establish:

Object	Commit	Tree	Ordered parents
Published refreshed head	21c7c20	548a1732a6ffb24246929d53b4c4334b1f73198e	fa1cf7b, then43dd74ca95df32f491f98ad1e40a5c4a5c79b2ee
Current main ref	fa1cf7b	8be046246426f440063b5b618e7abc681acf8a2b	ab13a08, then7ba9bbf306dfbbc74025581ab08d574b44fd1971
Actual synthetic PR test merge	62696d9	548a1732a6ffb24246929d53b4c4334b1f73198e	fa1cf7b, then21c7c2037343e0aaa60dd892e9ee7a16206700c7

The prior integration worker's unpublished alternate 96ac34292425b29e914b08f54a5069a74d943d58 has the identical tree548a and the opposite merge-parent ordering:43dd first, fa1 second. It remains history and is not substituted for the actual published branch head. The integration worker reports its guarded update stopped when another worker refreshed the branch; this audit verifies the actual resulting objects rather than attributing that publication to the root worker.

Observed metadata discrepancy: both fresh structured and raw PR172 readbacks identify head21c and test merge62696 but still report base SHA ab13. The separately fresh main ref is fa1, and the actual test-merge Git object has first parentfa1. These observations are all retained without rewriting a returned field. This source-binding verdict uses the actual live main ref and actual synthetic commit/tree. It does not claim the PR metadata base field had caught up or waive the integration worker's final live-state check.

Complete source preservation

I fetched complete recursive base/final tree records directly; both have truncated=false. I independently reconstructed every Git tree hash from canonical entry modes, names and child digests and matched the resulting roots against the fresh commit objects.

The current main advances the previous base by exactly eight added paths: .github/workflows/register-alignment.yml and seven files under reviews/register_alignment_20260930/. These additions do not overlap any of the fifteen reviewed packet/workflow paths. This is a source-disjointness check, not a review or acceptance of the peer's register-alignment content.

The complete current-main map has867 non-directory entries, and the refreshed candidate has868. The final map equals current main plus exactly the seven already reviewed repair paths: six modifications and the additive REVIEW_RECORD.json. All other current-main entries remain unchanged. Every one of the fifteen candidate files matches the frozen archived bytes, SHA256 and Git blob. The same eight original manuscript/source/finite/history files remain unchanged. The prior v1 binding payloads were rehashed against their frozen manifest and remain intact.

The published refresh and actual synthetic test merge share tree548a. No local packet test, mathematical review or full master was repeated. The already completed61-tests-per-mode local run and66 external negative rejections remain bound to the original candidate files on local parentb8. Their results transfer only as the recorded engineering evidence for those unchanged files.

Integration and retention boundary

Old successful hosted checks at head43dd/test27009 belong to the earlier base and are not rebound to head21c/test62696. Fresh hosted checks and original checked_commit/receipt evidence for the new complete tree remain the integration worker's obligation. This addendum does not claim hosted success or a merge.

The local review is preserved verbatim in the combined native body, followed by this v2 addendum. The previous combined15995byte body remains in remote-binding v1 as historical evidence. The original external merge of the defective packet, the later peer refresh, the guarded unpublished alternate and this exact-source acceptance are distinct events. A further head/base movement requires another bounded current-tree reconciliation.

Existing A/B/C analytic verdicts, source conditions, sampling restrictions, order of limits and review exposure remain unchanged. This engineering successor supplies no parent-theorem revalidation, new independence credit, scientific status promotion or theorem/prize closure.

[d6g8k5htny-coder]

C25 engineering custody repair and delivery are complete. Scientific effect NONE; this is not a new mathematical acceptance or organizational-independence vote.

The observed peer merge is fe2c3fff932de5f410d55d2ad95c7ca20b5630ca, with tree 548a1732a6ffb24246929d53b4c4334b1f73198e and ordered parents [fa1cf7b5f2d6a5f6c9cf23709509cf4c00f3b825, 21c7c2037343e0aaa60dd892e9ee7a16206700c7]. It matches the reviewed refresh and tested synthetic merge 62696d912aabba30dceebcb2c92c889fec126b97. No root merge API call or expected-head merge enforcement is claimed. The peer merged at 02:06:59 UTC; native engineering review 5360653158 was published at 02:07:35 UTC afterward. Prior independent local/source audits and the separate peer static review at the original head remain timestamped separately.

The final packet has fifteen exact sources: seven engineering/metadata changes and eight byte-identical original files. Both mathematical manuscripts, their six source pins, mathematical implementation/tests, results and validation history remain unchanged. Independent affected verification recorded 61 tests in each normal/optimized mode and 33 hostile variants in each mode (66 rejections). Original failed fixture/setup attempts and corrected executions are retained separately.

All four final pre jobs and all four post jobs succeeded on the recorded objects. Their original logs are preserved alongside four superseded original-head logs (twelve total). Final pre receipt: 180b3f0b14523613667b213e33bb6c4e7837f85e6e89727da953ed2151e708ff; post receipt: e969b9b03d1623c33a0ba0cedd1e6c1bf3a10ec0c6c9d5b03a624c5abc3722bf.

One restricted Drive delivery was uploaded and downloaded again: 1,994,642 bytes, ZIP SHA-256 eb45c10f521a2d3d534982049f3a7387761edfa4e9de7ce5b490782d973ba3dc, 462 payloads / 463 members. Exact ZIP bytes, every payload, and canonical archived handoff 28bb3295ed050bf22871275562633bffadd4fdd9d22a797ea56442167c9497e3 verified. Owner-only permissions were read back; no sharing changes were made.

R17 registration was read back exactly: Artifact Index 820, Evidence Lineage 542, Review Queue 54, PUBLISH 312, RELEASE 313. Claim b39708f0-b132-4d52-b449-308915f4eb8c was released at 2026-09-30T02:19:10Z. Immutable completion handoff-v2 canonical SHA-256: 5d942a8cfc0932a800e090b868161a19adfb8cfaf184b22130777c7b0554ee67; it links the archived C25 handoff and completed C23 predecessor. Existing conditional hypotheses and scientific gates remain unchanged; remaining review lanes retain their current owners.