Date: 2026-08-11 Scope: the protocol's starred anchors (1–8), with anchor 4 / KC1 first; §1 baseline skeleton (SSDI ≠ SSI; concurrent caution); stage-resolved adjudication for the verification year; GBMT-11 seam (diagnostic awards + stock); Ticket and BOND evaluation PDFs located. This is a gate pass, not a scorecard or a whitepaper. It changes no site page.
Verdict: GO — KC1 fires; KC2 and KC3 do not
The desk wager survives the cheap tests that can kill it.
- KC1 fires. The circulating integrity figure is an improper-payment rate from SSA stewardship reviews of non-medical eligibility factors — not an intentional-fraud series. FY2023 SSI improper payments were 10.62% of outlays (~6.5B); OASDIcombinedwas * *0.304.1B). Leading SSI causes in the Agency Financial Report are financial accounts, wages, and in-kind support and maintenance — reporting/complexity/status change, not a published intentional-fraud breakout at those magnitudes. Under the protocol, that mislabel is the filing's measurement headline: every architecture scored on O1 inherits a band, not a point.
- KC2 does not fire. Initial vs. hearing waits and allowance rates are stage-resolvable from public SSA publications for FY2024 (initial allow 38%; hearing allow 51%; initial APT 231 days; hearing APT 342 days annual average — above SSA's 270-day goal).
- KC3 does not fire for 1619(b). Take-up is published: 108,825 section 1619(b) participants aged 18–64 in December 2023 — 2.6% of blind and disabled SSI recipients in that age band. Ticket participation denominators live in the SSA-commissioned evaluation series; §5 still owes a clean eligible-pool fraction for H4's Ticket leg.
- Anchor 5 is corrected hard (GBMT-11 seam). Mental disorders are 12.7% of 2023 disabled-worker awards, not 25–35%. Disabled-worker stock in December 2023 is a different construct: depressive/bipolar 12.4% + intellectual 3.8% + all other mental 12.4% ≈ 28.6%. SSI under-65 mental share is larger still (agency highlight: six of ten). Do not reintroduce the false one-third-of-awards prior.
No kill condition stops the filing. KC1 reframes it: measurement honesty and adjudication/cliff workstreams stay load-bearing; fraud-first architectures enter demoted pending a clean intentional-fraud series (KC4 re-rank path).
1. Definitions locked (SSDI ≠ SSI; improper ≠ fraud)
- SSDI (Title II DI): social insurance for insured workers (and eligible dependents). Disabled-worker counts are not "all disability beneficiaries" (widow(er)s and adult children are separate). Medicare after the waiting period.
- SSI (Title XVI): means-tested cash for aged, blind, or disabled with limited income/resources. Often Medicaid from day one; state supplements vary. Blind and disabled recipients ≠ aged; do not sum SSDI worker rolls with SSI blind/disabled without subtracting concurrent dual entitlement.
- Concurrent: persons aged 18–64 receiving both Social Security disability and SSI — 949,971 in December 2023 (DI ASR / SSI ASR Table 15 family).
- Improper payment (Payment Integrity / AFR): payment that should not have been made or was made in the wrong amount under program rules, estimated from stewardship reviews of non-medical factors. Not a fraud rate. Intentional fraud is a separate OIG/prosecution object.
- Allowance rate: share of disability determinations allowed at a stage (initial / reconsideration / hearing). Hearing rates are materially higher than initial; they are not independent draws from the same applicant pool.
- Age-sex-adjusted DI incidence: awards per thousand disability-exposed workers (Trustees Report Figure V.C3 construct) — the right series for H1's "explosion" test, not raw roll counts.
Band rule for this filing: SSDI and SSI are never interchangeable in a scorecard cell. Awards ≠ stock. Improper ≠ fraud. Hearing allowance ≠ initial allowance.
2. Gate results by starred anchor
| # | Protocol prior | Phase 0 result | Confidence / source type | Effect |
|---|---|---|---|---|
| 1 ★ SSDI workers ≈ 7.5–8.5M; SSI disabled ≈ 6–7M | Recent-year rolls; do not sum casually | Mostly holds with construct locks. Dec 2023: disabled workers 7,365,987; all DI disabled beneficiaries 8,709,006. SSI total 7,425,331; blind and disabled 6,264,542; aged 18–64 SSI (all categories) 4,039,319. Concurrent aged 18–64 on both programs 949,971. Worker count sits just below the prior's 7.5M floor. | Strong — SSA ASIDI / SSI ASR / SSI-by-state. | Carry Dec 2023 figures; forbid casual SSDI+SSI sums. |
| 2 ★ Initial DDS allow ≈ 30–40%; hearing higher | Outcomes / workload | Verified (FY2024). Initial allow 38% (deny 62%) on 2,086,885 decisions; reconsideration 16%; ALJ hearing allow 51% (dismiss 33%, deny 16%) on 289,492 hearings. SAOR state table: national initial 38.3%. | Strong — SSA ODSSI workload PDF + SAOR. | H7's "initial <40%" leg is live. H2 adjudication flagship stays. |
| 3 ★ ALJ wait hundreds of days | Hearing workload / GAO | Verified. FY2024 annual average hearing processing time 342 days (FY2023 450); pending hearings ~262k (down 19% YoY). Still above SSA goal of 270 days. Initial APT 231 days (up from 218). Stage-resolved → KC2 does not fire. | Strong — SSA OIG major-management challenges FY2024; SSA budget APM. | H2's wait leg supported for multi-year span above goal. |
| 4 ★ IP mid/high single digits or low teens; mostly not intentional fraud | AFR / PaymentAccuracy / OIG | Verified; KC1 fires. FY2023 SSI IP 10.62% (OP 9.18% / UP 1.44%); OASDI combined IP ~0.30%. Causes: SSI — financial accounts, wages, ISM; OASDI — relationship/dependency, SGA, computations. Stewardship reviews are non-medical. No AFR table equates these rates to intentional fraud. | Strong — SSA FY2024 AFR Other Reporting Requirements; OIG PIIA. | Measurement headline. Fraud-first architectures demoted pending KC4. |
| 5 ★ Mental disorders ≈ 25–35% of working-age SSDI awards | ASIDI diagnostic tables | Corrected (seam with GBMT-11). 2023 disabled-worker awards: mental disorders 12.7% (musculoskeletal 34.0%, neoplasms 13.6%). Dec 2023 disabled-worker stock: mental categories sum ≈ 28.6%. SSI under-65: agency profile says six of ten have a mental disorder diagnosis — a different program and denominator. | Strong on awards and worker stock; SSI "6/10" is highlight-level pending table archive in §8. | Prior awards band is false. H8 award-share threshold (≥25%) fails on awards; stock/SSI legs remain live. |
| 6 ★ Ticket employment effects small at population scale | Mathematica / SSA TTW evals | Supported. SSA Social Security Bulletin synthesis of the Mathematica evaluation: intent-to-treat effects on benefit exit for work are null or undetectable at population scale; service enrollment rises, employment/exit impacts do not. Mamun & Page (2014): substantial service-enrollment impacts, no consistent months-without-benefits impacts. | Strong on published evaluation consensus; exact <2pp employment bar for H5 still needs §5 table pull. | H5 stays live; Ticket redesign architectures enter with small effect sizes. |
| 7 ★ 1619(b) small fraction of working-age SSI | SSI ASR; Red Book | Verified. Dec 2023: 108,825 1619(b) participants aged 18–64 = 2.6% of 4,148,143 blind/disabled recipients 18–64. All-ages 1619(b) 116,684. Working blind/disabled (incl. 1619) 5.1% of blind/disabled. | Strong — SSI ASR Tables 40, 43. | H4 take-up leg supported; KC3 does not fire for 1619(b). |
| 8 ★ Age-adjusted SSDI incidence declined/flattened after mid-2010s peak | ORES / actuarial | Verified. Trustees Report Fig. V.C3 age-sex-adjusted incidence: peak 6.4 per thousand exposed in 2010 → 4.0 (2019) → 2.9 (2022–23) → 3.3 (2024). Sustained post-peak decline, not a mystery surge. | Strong — 2025 Trustees Report data table. | H1's recent-decade leg supported; growth narrative must use adjusted incidence, not raw rolls. |
3. Baseline skeleton — non-substitutions
- SSDI ≠ SSI. Mixing unlabeled "disability rolls" is a discourse finding, not a baseline.
- Do not sum casually. Concurrent dual entitlement (~0.95M aged 18–64) means SSDI workers + SSI blind/disabled double-counts.
- Improper ≠ fraud. Citing the SSI 10.62% IP rate as a fraud rate fails the record. OIG CDI/prosecution counts are a different object (anchor 10 / H7 — workstream).
- Awards ≠ stock ≠ SSI diagnosis share. 12.7% / ~28.6% / ~60% under-65 SSI are three true statements about three constructs. The false prior glued them into "one-third of awards."
- Initial ≠ hearing allowance. 38% vs 51% is selection through denial and appeal, not proof that ALJs are "soft" without case-mix controls (H2).
- Incidence ≠ prevalence. Age-sex-adjusted awards per thousand exposed fell after 2010 even while absolute beneficiary stocks remained large.
4. Kill / re-rank triggers after this gate
| Trigger | Status after Phase 0 | Filing consequence |
|---|---|---|
| KC1 — integrity debate on mislabeled / unmeasured fraud number | Fires | Whitepaper leads with measurement honesty; O1 cells carry bands; demote fraud-theater architectures. |
| KC2 — backlog not stage-resolvable | Does not fire | §2 can score O2 quantitatively on FY2024 stage data. |
| KC3 — work-incentive take-up unpublished | Does not fire for 1619(b); Ticket eligible-pool fraction still to tighten in §5 | O4 can use 1619(b) 2.6% as the published cliff-take-up anchor. |
| KC4 — clean intentional-fraud series at circulating magnitude | Does not fire | Integrity stays secondary to adjudication + cliffs unless §3 finds a separable fraud series. |
| H7 early support (initial <40%; waits; CDRs vs prosecutions) | Partial — (a) holds at 38%; (b)/(c) untested | Steelman stays equal-effort in §7; CDR vs prosecution count still owed. |
5. Prospective follow-up (not done here)
| Item | Why it matters | Owner |
|---|---|---|
| Separable intentional-fraud $ or person counts vs IP dollars | Completes H3 / KC4 path | §3 |
| ALJ allowance variance after case-mix controls | Completes H2 | §2 |
| Ticket eligible-pool participation % and <2pp employment table extract | Completes H5 bar | §5 |
| BOND Stage 1/2 earnings nulls in one citation block | Cliff redesign scorecard | §4 / §5 |
| CDR cessation person-counts vs OIG fraud prosecutions | H7 (c) | §7 |
| SSI under-65 mental-disorder table (not just highlight) | GBMT-11 seam stock | §8 |
| Child SSI caseload (~1M prior) verification | H6 | §6 |
| Representative-payee share among mental-disorder beneficiaries | Seam with GBMT-11 | §8 |
Freeze
Protocol at docs/research-inquiry.md is the commitment as of Phase 0 start (2026-08-11). Anchor table verified values above supersede the unverified priors for anchors 1–8. Deviations logged in research/deviations-log.md.