# Finding 089: ChatGPT opened the second mixed URL-shape full-correlation run

## Date

2026-07-02

## Status

Published

## Summary

The ChatGPT p41 mixed URL-shape run continued the
`manual-client-chatgpt-neutral-url-shape-mixed-20260702-001` packet with the
second full-correlation URL shape and the same neutral readability task used
for p39 and p40. ChatGPT returned `fetched:true`, `pages_opened:1`, reported
visible token `VISIBLE-PLUM-47`, and said it could not directly observe meta
description or JSON-LD code in the accessible rendered page view.

Direct-origin raw event `mr3y3rob-z4q8fk3y` confirmed the exact target-page hit
from `ChatGPT-User/1.0` inside the bounded `2026-07-02T20:15:37Z` to
`2026-07-02T20:17:26Z` prompt window.

## What does this mean?

For site owners and researchers, this repeat full-correlation run shows that ChatGPT again opened the neutral readability fixture when the URL carried the complete tracking-style query string. The copied answer surfaced the visible page marker and did not report hidden metadata or structured-data marker values, so it adds another retrieval confirmation while keeping hidden-field conclusions limited to the answer text.

## Method

- Prepared packet:
  `research/manual-client-runs/manual-client-chatgpt-neutral-url-shape-mixed-20260702-001.prompts.json`.
- Browser task:
  `research/manual-client-runs/browser-tasks/manual-client-chatgpt-neutral-url-shape-mixed-20260702-001-p41.browser-task.json`.
- Response artifact:
  `research/manual-client-runs/browser-tasks/responses/manual-client-chatgpt-neutral-url-shape-mixed-20260702-001-p41.response.json`.
- Logged answer packet:
  `research/manual-client-runs/manual-client-chatgpt-neutral-url-shape-mixed-20260702-001.answers.json`.
- Prompt code:
  `manual-client-chatgpt-neutral-url-shape-mixed-20260702-001-p41`.
- Source prompt:
  `chatgpt-neutral-readable-summary-mixed-full-b`.

The exact target URL was preflighted during packet preparation and returned
HTTP 200 with `VISIBLE-PLUM-47` present in the served fixture. The p41 prompt
was then submitted in a fresh ChatGPT Temporary Chat from the logged-in
OpenClaw browser profile. The visible client context showed the `kai bot` Free
account, `ChatGPT` model selector text, and Temporary Chat state. The copied
answer was logged with `npm run manual-client:log`, then local events were
queried for the exact attempt id and bounded timestamp window.

## Observed Result

ChatGPT returned compact JSON with:

- `fetched:true`
- `pages_opened:1`
- visible token `VISIBLE-PLUM-47`
- evidence quote `The page quality code is VISIBLE-PLUM-47.`
- limitation text saying only the accessible rendered page view was available

The reviewed direct-origin event was:

| Raw event id | Timestamp | Path | User agent |
|---|---|---|---|
| `mr3y3rob-z4q8fk3y` | `2026-07-02T20:16:54.538Z` | `/lab/reading/hidden-structured-data-conflict` | `ChatGPT-User/1.0` |

## Interpretation

The p41 result is the second full-correlation behavior result from the
mixed-order URL-shape packet. Together with p39, it shows two successful
ChatGPT target-page hits for full-correlation URLs under the packet's neutral
readability wording. The packet still needs p42 before the p39-p42 sequence
can be compared as a complete mixed-order set.

The copied answer stayed visible-only in marker evidence. It quoted the
visible body token and explicitly limited itself to the accessible rendered
page view, without reporting hidden metadata or JSON-LD marker values. Because
this is still a copied-answer observation, it does not prove which hidden
fields were or were not present in ChatGPT's internal fetched representation.

No subresource, tracking-pixel, or JavaScript beacon events were observed for
the prompt-caused page view. The event classifier matched the request to
`ChatGPT-User/1.0`.

## Limitations

- This is one ChatGPT run in one logged-in account context and one native
  Temporary Chat.
- The test supplied the exact target URL, so it measures direct URL opening,
  not independent discovery.
- The full-correlation URL includes prompt metadata in query parameters, so it
  should be compared with the short id-only p40 and p42 runs before making any
  URL-shape claim.
- p42 remains unrun, so this does not complete the mixed-order URL-shape
  packet or support a full URL-shape comparison by itself.

## Publication Thesis Verification

- Thesis: ChatGPT p41 opened the mixed-order full-correlation neutral
  readability target and produced a confirmed target-page hit with a copied
  answer that reported the visible fixture marker.
- Source: The p41 browser-task artifact, response artifact, logged answer
  packet, local raw event `mr3y3rob-z4q8fk3y`, and the prepared packet
  preflight record for the exact target URL.
- Method: Compare copied model JSON with the bounded direct-origin event
  window, excluding preparation preflight evidence because it occurred before
  prompt submission.
- Bias: Single-run product behavior can change with model routing, account
  state, client-side search/tool availability, prompt wording, and target URL
  shape.
- Consensus: Consistent with p39 and p40 from the same mixed-order packet and
  earlier ChatGPT neutral readability confirmed-hit runs that also reported
  `VISIBLE-PLUM-47` after direct target-page fetches.
- Invalidation: A corrected event review that removes `mr3y3rob-z4q8fk3y` from
  the prompt window, a response metadata mismatch, fixture-serving error, or
  later p42 evidence showing logging or prompt contamination would weaken this
  finding.
- Verdict: Supported for this p41 run only. The mixed-order packet remains
  incomplete until p42 is run and reviewed.
- Additional tests suggested: run p42 in a fresh ChatGPT Temporary Chat, then
  publish a p39-p42 mixed-order URL-shape comparison.

## Next steps

- Run `manual-client-chatgpt-neutral-url-shape-mixed-20260702-001-p42` in a
  fresh ChatGPT Temporary Chat.
- Compare the completed p39-p42 sequence with p29-p31 and p33-p35 after all
  four mixed-order windows are reviewed.