Audit QQ16 integer consumer incidence and conditional budget

[Clear1nsight]

A module placed once at each first B1 cut has 16 proposed insertion sites across 144 local calls, a rate of 1/9, rather than the earlier pair rate of 1/2. This source-bound audit reconstructs the complete frozen hosted schedule and the corresponding nine-word QQ16 producer profile. The profile adds 136 H1 per replica (credited to Nick Petalas / #3339), giving mean n1 = 2412 + 136/9.

This changes the conditional budget decision: retaining the favorable assumptions W=896/9, centre64 and ideal rank51 idle completion, a 336-H1 module at that rate has G in [0.999977964672213086, 0.999977964672213087]. The strict integer native-module cap is 553 H1; 554 fails. The old pair-rate rejection therefore does not transport to this placement proposal.

The source contains zero actual #3952 invocations: all 144 calls still use gauge44. Sixteen counts proposed insertion sites, not proved consumer demands. The 482-event ledger identifies every relevant family/readset, the first missing provider connection, subsequent bridge/F2/terminal events and conditional output-retention peaks. It also checks that simply appending the projection's cleared output slots does not preserve arbitrary productive input under the selected family contract.

Scope is the frozen frontier1042-lineage hosted template, with selected frontier1094 kappa only as an arithmetic target. No QQ16 substitution, productive-family coverage transport, displaced idle credit, physical hosting or new kappa is admitted. The package has no executable or scientific dependency on the other parallel follow-ups. Parameter cases 208/224/304 are arithmetic values, not certifications of sibling candidates.

Validation on Python 3.12.3:

• Normal and optimized exact replay reproduce the complete typed RESULT.
• Seven optimized control groups cover provider/incidence/frame/family mutations, native rank and duplicate centre credit, reads before owned dirt is cleared, typed receipt mutations, source/manifests, and the 553/554 boundary.
• Minimal Git-free, read-only exports pass normal and optimized replay without changing file bytes.
• Registered package metadata/pins and full registered replay pass.
• Separate AI cross-review independently checked the 16/144 census, 384/1536 retention peaks, complete nine-stage profile, exact arithmetic and scope. Coordinating AI review found no substantive blocker. This is not human or proof-assistant review.

The published commit and source-history/byte preflight are reported below. Complete evidence, excluded obligations and next-step contract are in research/qq16-consumer-incidence/PROOF.md, RESULT.json, VERIFICATION.json, and REVIEWS.json.

Contribution workflow: original joint-query-controller card retained; priority revision 6; canonical main 20ecf2891d80b5acf18d980b2164fb44ea7faa90 remained unchanged at the 07:44 UTC checkpoint. Targeted live PR/Discussion overlap was refreshed at 07:45 UTC, with partial coverage explicitly acknowledged. This follows the canonical investigation claim.

Credit #3339 (Nick Petalas), #3952/#3963, Ali Romman's QQ16/source lineage, ywz65 #2512/#2707, Arham Tahir #2681, and the retained gauge44/T16/frontier1042 supplier/owner/coverage lineage. Chris R / Clear1nsight, with substantial OpenAI Codex assistance. Frozen source notices and the selected frontier are unchanged.

Publication: GitHub-created commit 137c557bbd5fb9f2221457fa400e3ed439481922 on Clear1nsight:research/qq16-consumer-incidence-20261019. PASS_RESEARCH_COMMIT_BYTES verified all 21 package/input files and source-history ancestry to canonical base 20ecf2891d80b5acf18d980b2164fb44ea7faa90; publication diff is confined to 14 new files in this package, and whitespace validation passes. GitHub CI status remains separately observable on this PR.

Hosted verification completed successfully: run39218396413, replay job117948351653. Logs confirm merge 5f4bdda0cef9201391b899cc5089c137aae382ad of head 137c557bbd5fb9f2221457fa400e3ed439481922 into base 20ecf2891d80b5acf18d980b2164fb44ea7faa90, CPython3.12.15, complete registered replay PASS, all seven control groups PASS, and the committed-inputs-unchanged step successful. This is hosted package replay, not complete supplier admission.