wassname/moral-maps
1
0
Fork 0
mirror of https://github.com/wassname/moral-maps.git synced 2026-09-22 13:20:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

Record React WVS deployment review

Browse Source 
Co-Authored-By: PI[gpt-5.6-terra] <288921227+claudypoo@users.noreply.github.com>
This commit is contained in:
wassname
2026-09-17 05:38:50 +08:00
co-authored by PI[gpt-5.6-terra]
parent f657321ea7
commit d7fe52bb68
2 changed files with 25 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
slop/audits/20260916_wvs_react_deploy_verification.md
+10
View File
@@ -0,0 +1,10 @@
# React WVS deployment verification
Observed after deploy of `f657321`:
- public page: `https://wassname.github.io/moral-maps/wvs/react/`
- public asset reference: `./assets/index-CCNzz5z-.js`
- local and public JavaScript SHA-256: `8297b5c03db01b5d49610c0d39ea877abd0cd87e1c01b0372498232c26eaa03e`
- local and public `wvs_map_data.json` SHA-256: `08dc2a5681eac7305f6c75481fcceebf9a103fad86405c2b662119986910b518`
The deployed asset and shared data bytes therefore matched the local production build at this check. Local React UAT used production build output and browser default/Qwen-hidden captures; it checked 64 models, 90 countries, four zones, 13 latest labels, and the oriented axis labels. -- PI[gpt-5.6-terra]
slop/reviews/20260916_wvs_react_f657321_fresh_eyes_openai.md
+15
View File
@@ -0,0 +1,15 @@
# Independent image-only Goal 4 review
**Verdict: PASS for the pictured React stretch UAT.** No blocking visual regression found. Reviewed all four requested images and `AGENTS.md`, without relying on earlier reviews.
## Observed evidence
- **Crosshair orientation and origin match the static reference.** In `slop/research/wvs/20260916_openrouter/wvs_react_local.png`, "Self-expression" is left, "Survival" right, "Secular-Rational" above, and "Traditional" below, matching `docs/img/wvs/wvs_map_iw.png`. The crosshair intersects approximately at screenshot pixel `(780, 737)`, right and below the plot center. Relative to the country/model points, this agrees with the static reference's origin, rather than incorrectly centering the axes. The hidden-Qwen image preserves this crosshair.
- **Model positions and plotted family labels visually agree.** After accounting for plot scaling and page placement, the React image matches `slop/research/wvs/20260916_openrouter/wvs_page_local.png` and the static reference: the high blue star, neighboring "grok-4.3"/"muse-spark-1.3" stars, rightward "gemma-4-31b-it" and "qwen3.8-flash", and the remaining labeled family stars retain their relative positions. The plotted Claude label remains "fable-5.1". Minor chip wording difference: React says "claude: claude-fable-5.1", whereas the non-React page says "claude: fable-5.1"; this does not change the plotted label or placement.
- **Qwen-hidden state is correct visually.** In `slop/research/wvs/20260916_openrouter/wvs_react_qwen_hidden.png`, "qwen: qwen3.8-flash" becomes outlined and struck through. Qwen's purple star population and latest plotted label disappear. Other families, including the purple Muse and lavender Llama markers, remain. Country points, hulls, remaining labels, plot bounds, and crosshair do not visibly move.
## Limits and disproof checks
These screenshots establish visual parity and the supplied hidden state, not live click handling, exact numerical coordinates, artifact provenance, or other viewport behavior. A browser toggle test showing residual Qwen marks, removed non-Qwen marks, or layout movement would disprove the hidden-state conclusion. Coordinate comparison against the shared artifact could disprove the visually inferred origin/position agreement.