neochart v2 research

cases/42usc1981/RESEARCH.md

Source: cases/42usc1981/RESEARCH.md

Exact source text

# Section 1981 case annotations — author research record

Research date: 2026-11-10. Source corpus was read-only; all ten files use CASE DSL v1 and ids declared in `neov2/law/42usc1981.law`. This is author preparation for independent review, not independent QA approval. Shared NOTES, tools, law files, durable memory, and git history were not edited by this lane.

## Selection and what was actually decided

The six existing law research seeds were retained after reading their facts, posture, operative reasoning, and disposition: Domino’s, CBOCS, Boxill, Ridgell, Nelson, and Bryant. Jones supplies historical Supreme Court limitations analysis; Brown is a deliberate preclusion/background example; Haynes and Hu add Fourth and Second Circuit claimant-success pleading/summary-judgment decisions. The sample spans 2004, 2006, 2008, and 2019, with Supreme Court and Second, Fourth, Sixth, and Eighth Circuit opinions. It is a purposive sample, not a random or exhaustive survey; 2019 and the Eighth Circuit are overrepresented because the cited-by index is recent and capped.

- **Domino’s:** pure legal defect in shareholder/agent ownership of contract rights on a dismissal motion. Personal injury and alleged racial animus do not supply his own contract. Proposed-contract and postformation coverage are rule announcements, not affirmative injury findings. Intended third-party beneficiary status is reserved.
- **CBOCS:** Supreme Court recognizes retaliation, including advocacy of another worker’s racial contractual rights, but does not decide the trial elements. Both content blocks are `rule-announced / not-reached`. Title VII procedures do not subsume §1981. Majority ends before the Thomas dissent at 25374.
- **Jones:** the relevant claims depend on the 1991 expansion, so §1658 supplies four years. Both blocks are procedural and excluded from merits instances. The defense holding rejects the asserted Illinois two-year bar; it does not establish that every other timeliness issue is resolved or that discrimination occurred. The separately discussed employee-status claims are excluded.
- **Boxill:** §1983 is a valid vehicle for §1981 state-actor rights, but the complaint fails on individualized knowledge/retaliatory causation. The §1981 section incorporates the factual deficiency discussed earlier under First Amendment retaliation; its motivating-factor legal standard is not imported. The separate successful equal-protection hostile-environment claim is not called a §1981 win. Corrected IR’s “Six Circuit” court spelling.
- **Ridgell:** exoneration of the sole alleged discriminator defeats the City theory on this record; independent council discrimination also lacks evidence. These are separate state-remedy and intent issues. The unchallenged mayor verdict is not treated as a newly reviewed factual finding.
- **Nelson:** two qualified intent blocks record the actual rejected pretext theories: comparable qualifications and allegedly manipulated selection criteria. Summary-judgment evidentiary failure uses `no-evidence`; the prima facie case is not ultimate liability. Caption normalized to USAble.
- **Bryant:** record supports punitive culpability; due process permits the particular punitive award despite nominal compensatory damages; prior awards support counsel’s rate under §1988. The four-year continuing-environment defense is separately procedural. Underlying liability is not re-created from the affirmance, and dismissed retaliation is not an appellate finding. The opinion’s §1981a punitive-standard analogy does not import its Title VII damages cap.
- **Brown:** identical factual allegations after a Missouri merits-equivalent dismissal yield claim preclusion. Restaurant contract allegations are context only. One background/procedural/not-reached block preserves the source; no nonexistent preclusion law id or false limitations holding is invented.
- **Haynes:** comparable white employee discipline and shifting termination reasons are triable intent routes. This is remand, not proved liability. The final motivating-factor footnote predates Comcast and is not used as a modern §1981 causation holding. Retaliation mentioned in the introduction is not independently analyzed.
- **Hu:** same-site standing-water comparator and alleged racial bias satisfy the pleading issue expressly resolved under §1981. Company/client facts are retained under own-contract `not-reached`; state-remedy and modern but-for issues remain unresolved. This avoids extrapolating all elements from vacatur or importing the separate equal-protection class-of-one standard.

No current but-for finding is fabricated: all ten opinions predate Comcast (2020). The law file’s external current-rule annotation remains controlling context. This sample intentionally does not present old motivating-factor language as current §1981 law. Coverage of protected-person variants, alienage, and post-Comcast merits is a remaining gap, not an implicit negative result.

## Rejected or unavailable candidates

Runyon `/us/427/0160-01` and Saint Francis College `/us/481/0604-01` were attempted through the original corpus helper: their reporter IR files were unavailable. They were not invented or substituted from secondary text. Jones was available as the historical Supreme Court addition.

Mungai `/f4th/141/959` was read in full: the 2025 appellate opinion waived challenges to §1981 dismissal and decided Title VI and other claims. It would be a valid background example but was not selected because Brown and Jones already exercise procedural scope. Mawakana `/f3d/926/0859-01` was screened at 0..4496 and its subsequent Title VII discussion: §1981/§1983 summary judgment was expressly not appealed. Its motivating-factor success therefore cannot be assigned to §1981. The batch display of its later Title VII discussion was truncated; no conclusions rely on the omitted portion.

## Reading windows and source pins

Offsets are Unicode code points into the canonical IR `text`. These windows include factual and procedural context and the actual court issue, not just keyword hits. Some commands print complete overlapping paragraphs around the requested window. “Entire” means the available IR record, not a claim that a capped IR reproduces a complete official opinion. CBOCS’s historical discussion between the listed windows and its dissent were not comprehensively read. Hu is capped at 60000; its §1981 section is complete before 56132, but later due-process/municipal reasoning and the final disposition may be missing. Only the explicit §1981 vacatur is annotated as its result.

| case id | source_sha1 / chars | reading windows | selected rule and held pins |
|---|---|---|---|
| `/f3d/918/0990-01` | 86dd01fe70d562b97d66c8b7d261d59fea87b1f3 / 10751 | 0..10751 (entire canonical record) | rule 6787..7484, held 7485..8416, held 9710..10093 |
| `/f3d/919/0520-01` | 18f0209ac11cb356a7ae489bc36a4d988eed4d74 / 18642 | 0..18642 (entire canonical record) | rule 5425..6180, held 6181..7163, rule 7164..8056, held 8057..9012, rule 13095..13830, held 15003..15981, rule 16080..16182, held 17251..17851 |
| `/f3d/922/0219-01` | 58df5b6605c36f634be86b37adea23cf5ad9e028 / 19643 | 0..19643 (entire canonical record) | rule 8201..9052, held 12581..12931, rule 14641..15372, held 17052..17524 |
| `/f3d/927/0081-01` | e76e41832d96364f6fe79f1579fc1d5d7839bb1d / 60000 | 0..14064; 35623..41228; 46378..56132 (facts, posture, relevant comparator application, §1981 ruling); keyword index used only to locate reading windows | rule 54759..55737, held 55738..56131 |
| `/f3d/931/0712-01` | e571851d18ea2cd8f164345053535ed78fba6a75 / 7009 | 0..7009 (entire canonical record) | rule 2910..4157, held 6634..6850 |
| `/f3d/935/0510-01` | dfea3da6a402058d0732fea06d5a327790eca30a / 20792 | 0..20792 (entire canonical record; 6418..10915 re-read separately after batched display truncation) | rule 13207..13654, held 13655..13808, rule 14122..14517, held 19558..19839, rule 14122..14517, held 14518..14663 |
| `/f3d/935/0633-01` | 2b0f55880b8427f39ac8208b3749ec5255b169c3 / 12402 | 0..12402 (entire canonical record) | rule 4620..5033, held 9417..9693, held 11546..11854 |
| `/us/541/0369-01` | 472bcbf15b4b9b5ae7d15d3fe2557a047acd7d7a / 28755 | 0..17523; 17523..21729; 23123..23507; remaining endnotes screened | rule 17523..18269, held 19008..19222, rule 1997..3521 |
| `/us/546/0470-01` | 1fd8723b1aeb06cd5c055f91acdf19b2b2d453ed / 17353 | 0..17353 (entire canonical record) | rule 6665..6820, held 9057..10154, rule 6665..7639, rule 8244..9056 |
| `/us/553/0442-01` | 7873a4001f1716d2c2c65d2fe80c957c2051d9ff / 58655 | 0..7290; 17483..25373 (opening, posture, operative objections and majority conclusion; dissent boundary checked) | rule 51..581, held 25226..25305, rule 20427..21416 |

## Findings for leader integration into shared NOTES

1. A law id can contain both a remedial gate and substantive municipal attribution: Boxill’s §1983 vehicle is available while liability fails; Ridgell’s vehicle is available but no attributable wrong survives. Qualifiers and gloss distinguish these meanings under `state-remedial-route`; a future schema could separate vehicle, policy, and underlying violation.
2. The missing preclusion id cannot be repaired by treating res judicata as limitations. Brown therefore remains background/procedural and contributes zero instances. Its court describes the prior judgment as “on the merits” for preclusion; that does not make a §1981 element decided on the merits here.
3. Fact polarity lacks a neutral procedural-context value. Brown’s purchased-drinks/removal allegation supports the alleged contractual-impairment theory but is not a merits finding. Hu’s corporate relationships similarly remain contextual and not-reached. These should not train a model as positive liability findings.
4. Pretext is a proof route to racial intent, not an extra statutory element. Two theory-qualified blocks in Nelson and Haynes record distinct recurring evidence while retaining a single underlying intent question; consumers should avoid treating them as separate claims.
5. Historical opinions lack typed good-law/temporal metadata. Explicit modern-but-for `not-reached` blocks and prose prevent the author from asserting a Comcast holding, but a downstream consumer still needs the law-side temporal restriction. Hu’s comparator formulation is local to selective enforcement, not a universal §1981 pleading requirement.
6. Rule-announced coverage (CBOCS, part of Domino’s) can be highly important doctrine while producing no factual instance. Procedural timing in Jones/Bryant must remain outside the merits totals. The generated variant catalogue also lists applied-but-not-reached context, so consumers should use `block.instance`, not every row of the catalogue.

## Verification and permanent quote freeze

`python3 neov2/tools/casedsl.py fill neov2/cases/42usc1981/*.case` completed, after which no quote text, kind, or offsets were edited. `check` reports **10 files, 76 quotes, 0 errors**, and a direct check API audit finds **0 warnings**. Generated VARIANTS contains **25 blocks, 14 instances, 4 applied-but-not-reached blocks, 7 other rulings**. There are 38 fact lines with explicit polarity, 5 marked absent. Corpus, tools, shared notes, and law files are unchanged by this lane. No commits were made; leader owns integration and independent QA follows.

The following SHA-256 fingerprints freeze each file’s ordered `(kind, start, end, quote)` tuples immediately after fill. Compute with `json.dumps(tuples, ensure_ascii=False, separators=(',', ':')).encode()` and SHA-256; tuples follow parsed block/quote order. This includes quote kinds and offsets, not just strings. The combined fingerprint hashes sorted `(filename, tuples)` pairs using the same serialization: `70d17a6e254ac3b0a84dddd75523006afad4260724db1d6726acf34719ed63cf`.

| file | quote count | frozen SHA-256 |
|---|---|---|
| `f3d-918-990-nelson-v-usable-mutual-insurance.case` | 7 | `ad0acdbe0764a7e0192bf0f2d7a37c0694f1f5c3fbb18b202befdcc49bd909ad` |
| `f3d-919-520-bryant-v-jeffrey-sand.case` | 15 | `c8aefe9952c7cbeac1e8261cfddf232e479285abb2e9a2bf063886e0a51285f0` |
| `f3d-922-219-haynes-v-waste-connections.case` | 9 | `fe57d3ec93dc14e0a109abb9857f16ad3f9f4d19e8159d2fcbfeab6a416ba8d1` |
| `f3d-927-81-hu-v-city-of-new-york.case` | 7 | `22a7a5689cc5e213dcb33a429b3a8bdc2f85ff68dce230fa2eca84cbf54666b3` |
| `f3d-931-712-brown-v-kansas-city-live.case` | 3 | `3ff3990f969eb128c8dd4cb3548a4f0c1c1fdd8375e50cb27c74f425640091db` |
| `f3d-935-510-boxill-v-ogrady.case` | 11 | `4fc6a2aa12b0bdd3f6455cd0875a8f5369fe325ea1ba0b5ba148e52830badd59` |
| `f3d-935-633-ridgell-v-city-of-pine-bluff.case` | 8 | `b12b14d771a39854bf599e7706cc5e6597f2f1ffa34c3f0226f9418a9afd783a` |
| `us-541-369-jones-v-rr-donnelley.case` | 5 | `742dbab71f655e9871a4bbc5bdc787dd72d0098347feaf12d95f178867f41114` |
| `us-546-470-dominos-pizza-v-mcdonald.case` | 7 | `e31f04ae678f8a72fa0d34b40e6bc028f4c1cb2f0e550c1f987ab7c7c35e1c6f` |
| `us-553-442-cbocs-west-v-humphries.case` | 4 | `9b7b58dfd7ceebc355437b2ff2b2812fb1b40d35cb0891bfe1948d59be8d0539` |

## Independent peer-review correction

Peer review identified that Bryant’s limitations ruling rests on trial witnesses and permissible jury inferences concerning harassment within the four-year period, not only legal period selection. Its basis is now `proved`; its procedural scope and exclusion from the 14 merits instances are unchanged. The termination/filing-date fact now has `cuts-against` polarity and a gloss clarifies that dates supply context, not proof of a timely component act by themselves. Hu’s gloss now states that statistics cannot replace a specific comparator, with its actual same-site comparison independently supporting the result.

Only basis, polarity, gloss, research, and regenerated catalogue changed. Before/after comparison preserves all 76 `(kind, start, end, quote)` tuples exactly, and the combined fingerprint remains `70d17a6e254ac3b0a84dddd75523006afad4260724db1d6726acf34719ed63cf`. Repeated `check` and direct warning audit both pass (0 errors, 0 warnings). Fact polarity totals are now supports 25, cuts-against 13, with 5 absent facts; counts remain 10 cases, 25 blocks, 14 instances.