Independently date 13 K4be bodies with December 2025–March 2026 creation assertions

<!-- swarmstatus:k4be-early-body-dating -->

Thirteen public K4be pastes currently declare creation dates from **14 December 2025 to 1 March 2026**. When can their **complete stored bodies** first be independently shown to have been public, and is there a specific evidenced relationship to the later task-sharing corpus?

The early dates are service assertions captured on September 5, not established historical swarm events. Timing and content relationships are separate parts of this question.

## Starting evidence

Discovery credit: **Jonas Wiedermann-Möller / j0wimo**, whose [post was supplied by the user](https://x.com/j0wimo/status/2096330564337078641). We received the post's text from the user; our attempted direct X read returned 403. The original month grouping did not specify a year. The UTC years below come from interpreting each captured API `created` value as Unix seconds.

| Paste | Current service-created assertion (UTC) |
|---|---|
| [3ae5f54b](https://pastebin.k4be.pl/view/3ae5f54b) | 2025-12-14T04:51:46Z |
| [514333ed](https://pastebin.k4be.pl/view/514333ed) | 2026-01-01T10:56:05Z |
| [316996fe](https://pastebin.k4be.pl/view/316996fe) | 2026-01-01T10:56:06Z |
| [5329a841](https://pastebin.k4be.pl/view/5329a841) | 2026-02-26T14:49:24Z |
| [bd44d381](https://pastebin.k4be.pl/view/bd44d381) | 2026-02-26T14:50:18Z |
| [680ec235](https://pastebin.k4be.pl/view/680ec235) | 2026-02-26T14:52:19Z |
| [d78a30b4](https://pastebin.k4be.pl/view/d78a30b4) | 2026-02-26T20:15:44Z |
| [0263afe2](https://pastebin.k4be.pl/view/0263afe2) | 2026-02-26T20:15:47Z |
| [06d98526](https://pastebin.k4be.pl/view/06d98526) | 2026-03-01T13:24:18Z |
| [17bbf392](https://pastebin.k4be.pl/view/17bbf392) | 2026-03-01T13:30:22Z |
| [c9601cdf](https://pastebin.k4be.pl/view/c9601cdf) | 2026-03-01T13:31:19Z |
| [1806ec31](https://pastebin.k4be.pl/view/1806ec31) | 2026-03-01T13:32:39Z |
| [5a4aa403](https://pastebin.k4be.pl/view/5a4aa403) | 2026-03-01T13:33:28Z |

All thirteen view pages and their documented API representations were captured successfully on **5 September 2026**. The HTML acquisition intervals span **20:23:41.496574–20:24:27.802742 UTC**. Those are the earliest complete-body observations in the evidence currently reviewed, not the pastes' first-publication dates.

Independent local comparison of the saved HTML and API representations confirms all thirteen complete stored texts after explicit one-time HTML-entity decoding and CRLF-to-LF normalization. The original rendered paste was checked separately from the reply-prefilled textarea; its display comparison also accounts for nonbreaking spaces and blank-line fillers. Raw HTML, API-envelope and extracted-text hashes remain distinct. A complete stored paste can itself be an excerpt of an upstream document.

The January 1 pair has identical unmodified API `raw` strings and creation assertions one second apart. The other bodies include NBA arithmetic, financial tables, a finance question/answer/program block, population/CO₂ figures, airport data and an India armed-forces series. These similarities and timestamps do not identify a writer, reply or copying direction. None of the thirteen captured API envelopes supplies a native parent/reply field.

A FinQA dataset-origin comparison is a lead, **not an established item match**. A benchmark release date would date source material; it would not date publication of these paste bodies or prove task execution.

## What has already been checked

- Exact identifiers were absent from the fixed focused/full database snapshots and from the checked prior native captures, manifests and supplied 16,579-document pack. These are scoped coverage gaps, not proof that the pages were newly created or absent elsewhere.
- Three exact Wayback CDX queries for `3ae5f54b`, `514333ed` and `316996fe`, restricted to successful HTML captures before September 5 and collapsed by month, returned HTTP 200 with empty arrays. No archived body was obtained. Other archives, statuses, content types and versions remain untested; do not interpret these results as general absence.
- A coarse older search-index rendition of the NBA page does not provide an independently dated complete-body artifact.
- Relative ages, current API dates, HTTP response dates, expiry/hit-update fields, body data years and later acquisition clocks are separate observations. The current records do not supply a historical import or body-edit clock.

## Required evidence: chronology

- [ ] Recover an independently dated, source-located historical artifact for an exact listed object that preserves its complete stored body. Supply the original bytes or a public-safe reproducible package, content hash, exact locator, acquisition provenance and capture-clock/timezone basis.
- [ ] Compare that body with the current captured version using stated decoding and whitespace rules. Preserve both raw hashes; distinguish exact equality, normalized equality, excerpts and changed versions.
- [ ] State the defensible **body-present-by** bound. A historical capture normally proves presence by its capture time, not first publication. Assess service-created, import/repost and edit assertions only where separate evidence supports their meaning; do not diagnose backdating or authenticity from a timestamp alone.
- [ ] Report each of the thirteen objects separately, including unresolved ones. A dated subset or a documented earlier version is a useful partial result. Cohort-wide dating remains open while any required object's historical body date is unsupported.

## Separate evidence: relationship to later material

- [ ] Test a concrete, fully specified content relationship against an independently located later record, if one is found. Preserve both source locators, complete relevant question/table/answer/program or ordered selector bundle, native/capture clocks and copy/import lineage.
- [ ] Distinguish an exact distinctive task match from common benchmark text, shared numbers, generic prompts, same-host reuse or curator copies. Matching a public dataset item establishes that content relationship only; it does not demonstrate a message, receipt, agent identity, task use or transfer direction.
- [ ] State the result of this relationship stage independently from the chronology stage. If no discriminating relationship is supported, report it as unresolved. Earlier dated text alone cannot close a coordination claim; a bounded nonmatch cannot establish that no relationship exists.

A contribution may advance either stage. Mark partials as partial and keep unresolved criteria open. Search misses, current metadata alone, and a diagram joining similar titles do not meet these evidence requirements.

<details>
<summary>Current extracted-text SHA-256 controls</summary>

These hashes cover each unmodified JSON-parsed API `raw` string encoded as UTF-8, including its original line endings, before entity/line-ending normalization. They are not hashes of the API envelopes or HTML pages.

```text
3ae5f54b 4561a13e3c59846d9d02bf2fad88964598632b2f9a86d106d9e2f5667cb4621d
514333ed 76da74d9471e7b12892b538b85d3fba70005799a56f6081f1878ff8b16bfe485
316996fe 76da74d9471e7b12892b538b85d3fba70005799a56f6081f1878ff8b16bfe485
5329a841 94d46d8d52654d063d1056eede571b94c39889a0b6d953ed873df31900395948
bd44d381 ec89a96fa17f4347f5976e1e392cc5a9fa20cdd05b8ab650e5156db4e89c09d7
680ec235 c36c211ea5138142c7bf257386df02fc5e77f59293e1349f5a597d82e095363a
d78a30b4 05b598b81113a62916b83a0bbe8e4eccacb4ff62cb83faf19f1f5dd6dac6870c
0263afe2 1d9adba5396d886107cba63d3d4667a44c22cbb91ece7a89e9aa121752bf581b
06d98526 1d7c03707602cbb5378e88a6d8f3955633d276bf4b0fe76a1a8f84b4907fefef
17bbf392 cef636fa0156ef66700b6cff2669721279804dfbf83025db98878a07789d791a
c9601cdf 64254238457d888ea0a4c90e48948c1d96f9e8e3c52936e5f46386df04d2f8e0
1806ec31 d28c67fdf6b313dda342202ce5059fb3287e36a39bcf3d7b192f2c9e5107470b
5a4aa403 373bc8c396433321ac42a05402d588c097c6b3fd7df413954e6686acf4f6a112
```

</details>

## Contribution boundaries

Use retained evidence and ordinary authorized publication/archive reads. Do not replay embedded proxy or executable payloads, mutate historical posts, bypass source stops or access controls, or treat newly posted investigation material as historical evidence. Avoid private identity data and attribution from public labels. The thirteen observations remain new research intake, not an accepted historical-swarm milestone.

To announce work, comment `/attempt` alone, then describe your scope and first concrete finding separately. Claims are not exclusive. Accepted evidence receives public credit; **no cash reward is offered**. Follow the [contribution guide](https://github.com/aadarwal/swarmstatus-bounties/blob/main/CONTRIBUTING.md) and submit a reproducible public-safe partial or answer package.


## Research note — exact benchmark item match, 5 September 2026

An authorized acquisition of the [official FinQA repository at commit `0f16e2867befa6840783e58be38c9efb9229d742`](https://github.com/czyssrs/FinQA/tree/0f16e2867befa6840783e58be38c9efb9229d742) supplied a concrete source-content match.

In [`dataset/dev.json` at that commit](https://github.com/czyssrs/FinQA/blob/0f16e2867befa6840783e58be38c9efb9229d742/dataset/dev.json), zero-based array index **619**, item **`AMT/2008/page_32.pdf-4`**, the exact `qa.question`, `qa.answer` and ordered `qa.program` strings reconstruct the entire [`1806ec31` paste](https://pastebin.k4be.pl/view/1806ec31) as its three labeled Question/Answer/Program lines, including its final newline. The answer field is `29.2%`; the program is treated as inert text, not executed.

The relevant source table's zero-based `table[4]` records the 2008 quarter-ended-December-31 high of `37.28` and low of `19.35`; `post_text[0]` records the February 13, 2009 closing price of `28.85`. These are the dataset's retained fields, not a new financial calculation or independently verified financial result.

- Official `dataset/dev.json` SHA-256: `a847fb7e0d61a3125a1e2909852df6b89f1ee64d2c5ff1bf689e332214deee51`.
- Complete unmodified paste `raw` string, UTF-8 SHA-256: `d28c67fdf6b313dda342202ce5059fb3287e36a39bcf3d7b192f2c9e5107470b`.
- Retained API envelope SHA-256: `0eae1795033297f81364303acda7ae18ad021b1e31cc6d9e6d2bf43e2dce5583`.
- Repository checkout acquisition: **2026-09-05 20:37:37.302002–20:37:38.802853 UTC**. Its Git commit metadata is a separate assertion, not a historical capture of K4be.

A second reviewer independently checked the local Git blob and the paste's original API text. This establishes an exact benchmark-item content relationship. It does **not** establish that the poster obtained it directly from this repository, executed the program, belonged to the later corpus, or posted the complete body at the service's asserted creation time. Both historical paste dating and any discriminating later-corpus relationship remain open. The original question and acceptance criteria above are unchanged.

<!-- swarmstatus:k4be-finqa-item-context -->


<!-- swarmstatus:k4be-humana-report-context -->
### Research source context: Humana table (5 September 2026)

An independently reviewed comparison of three held paste bodies with the already acquired [official FinQA training data](https://github.com/czyssrs/FinQA/blob/0f16e2867befa6840783e58be38c9efb9229d742/dataset/train.json) supports these narrower content relationships:

| Paste | Reviewed match |
|---|---|
| `5329a841` and `bd44d381` | All six 2010–2015 dates and all 18 amounts, in HUM / S&P 500 / Peer Group row order, after explicit dollar-line joining, entity decoding and label handling as applicable. This is structured table equality, not raw-body equality. |
| `680ec235` | Only the two HUM cells: 2011 = $162 and 2012 = $128. The other years and comparison rows are absent. |

The same table appears at zero-based training indices **149, 538 and 677**, IDs `HUM/2015/page_46.pdf-2`, `-1` and `-3`, with distinct questions and programs. None of those complete questions or programs appears in the three pastes. Index 677 relies on dividend prose and amounts absent from all three, providing a same-table/different-task control. Prose also differs: literal and whitespace-only comparisons fail; one longer sentence requires separately documented punctuation-artifact repair. Shorter bodies omit or rewrite text.

Source pin: commit `0f16e2867befa6840783e58be38c9efb9229d742`; `dataset/train.json` SHA256 `49f237eb9779b569473b26b08048867d04635a7cc39ad6a7a5664c55bb428db6`. Independent review covered these three bodies and three source rows, the body and training-file hashes, table extent, transformations and the index-677 control. It did not repeat an all-dataset search. No source program was evaluated.

This adds report-content context only. Historical body dates, direct copying route, unique task identity, actors and use or receipt remain unestablished. The existing AMT note and all question criteria remain intact and open.


### Research source context: India personnel CSV (5 September 2026)

Independent comparison connects the complete decoded [06d98526 paste](https://pastebin.k4be.pl/view/06d98526) to a **selected excerpt of the current [official OWID CSV](https://ourworldindata.org/grapher/armed-forces-personnel.csv?v=1&csvType=full&useColumnShortNames=false)**: India/IND, 1995–2010, all 16 ordered rows, 80 data-field strings and five header fields, including the literal Asia region and integer values.

The original API raw string is 537 UTF-8 bytes and contains two header `&quot;` entities. Decoding only those entities gives 527 bytes. Source physical header line 1 plus lines 2374–2389 is 528 bytes; removing exactly its final LF produces the same complete 527 bytes. No sorting, deduplication, numeric rounding or general whitespace collapse was used. The full 191,776-byte CSV and whole India series are not equal to the paste. The same-file 1994–2011 control adds two rows and fails equality.

Retained source pins:
- CSV SHA256 `7b2a1fc134dd7ffb6357ef2c14d23aac43641dae509736f659ac02e0811db28d`; complete HTTP 200 capture at 21:25:41.228393–21:25:41.693128 UTC on 5 September 2026.
- [Metadata](https://ourworldindata.org/grapher/armed-forces-personnel.metadata.json?v=1&csvType=full&useColumnShortNames=false) SHA256 `fd91766e0573187e099ac89fb80c5df36d6a3fef1ac970c8c9279c59d49930ed`; complete HTTP 200 capture at 21:25:58.335651–21:25:58.703451 UTC. It identifies personnel variable 1294141 / WDI 129, lastUpdated 2026-07-27, and region variable 900801, lastUpdated 2023-01-01.

These source metadata dates, data years, acquisition clocks and the paste's native creation assertion remain separate. No explicit immutable historical dataset version was supplied. This establishes current selected-data content context, not a unique benchmark/task, direct copying route, historical paste publication, actor, successful use or receipt. Both question tracks and their original criteria remain open.

<!-- swarmstatus:k4be-india-current-csv-context -->