Body image coverage — 2026-05-18

DARE.CO.UK · BODY IMAGE COVERAGE · 2026-05-18

At a glance

245/690 pages have no body image (35.5% of the corpus).
Threshold: <20% green · 20–40% yellow · ≥40% red. Current: YELLOW.
Recovery path: ~/bin/dare_lost_image_audit.py queries Wayback for each page’s historical media, classifies whether the original image is still on R2 (just re-reference) or needs Wayback rewrite.

Distribution by section

Section	Pages without body image
`observations/`	74
`cinema/`	63
`daring-acts/`	19
`future-media/`	13
`architecture/`	12
`people/`	12
`albums/`	9
`photography/`	8
`fine-arts/`	8
`industrial-design/`	5
`user-interface/`	4
`psychology/`	4
`brands/`	2
`books/`	2
`narrative/`	2
`writing/`	2
`dmca-policy/`	1
`privacy-policy/`	1
`sitemap/`	1
`anti-spam-policy/`	1
`places/`	1
`new-york-city/`	1

First 50 candidate pages

… 195 more. Full list via the audit script.

What this tells us

Most low-grade pages are AI-migration casualties — the body text survived the template rewrite but the original <figure> blocks were stripped. The text often still references the missing images (e.g. “the smoky eyes in this portrait are particularly striking”), making the loss editorially detectable but visually easy to miss across hundreds of pages.

Watch items

A new AI-migration pass that doesn’t preserve <figure> tags would spike this metric. Tripwire catches it before the rollout completes.
If the % drops materially without a recovery sweep, investigate — could indicate page DELETIONS rather than image restorations.

Recommendations

When Wayback is reliable, run ~/bin/dare_lost_image_audit.py to get per-page recovery candidates classified by whether R2 still has the original.
Prioritise sections with high counts above — those are the densest pockets of recovery value.

Source: dare_body_image_coverage_2026-05-18.md · Rendered 2026-05-18 08:30