# v0.20.0 first-party media performance validation Date: 2026-08-11 Release tested: `0.20.0` Scope: repository/root-deployment package before live deployment. ## Result Focused media-delivery and full regression checks passed. No live deployment was performed. ## Code and route checks - PHP lint: **106/106 PHP files passed**. - Native JavaScript syntax: `assets/js/site.js` passed `node --check`. - Sitemap canonical route count: **85**. - Local PHP server smoke: **85/85 canonical routes returned HTTP 200**, with no PHP warning/fatal output. - Rendered route checks: **85 unique titles**, **85 unique meta descriptions**, one H1 per route, self-referential clean canonical URLs, and parseable JSON-LD. - URL normalization: **182 redirect assertions passed** — missing-trailing-slash and explicit route `/index.php` variants for all non-root routes plus the 14 retained legacy root PHP endpoints. ## Responsive-media checks - Rendered local media references checked: **28**; no referenced file was missing. - Distinct rendered `srcset` candidates checked: **24**; every width descriptor matched the physical pixel width of its local asset. - Homepage hero: one visible responsive WebP source with **480/720/1200/1672** candidates; visible image remains `loading="eager"` + `fetchpriority="high"`; no separate image preload path was added. - Respect essays: **6/6** render local WebP display candidates at **480/720/1080** widths, lazy-load the in-page art, and retain an explicit full-resolution PNG artwork link. - Distributed campaign features: **4/4** (Terminology, Identity, Stewardship, Research) render bounded WebP previews instead of poster-scale PNG `img` sources and remain lazy-loaded. - Share page: **8/8** image previews declare intrinsic dimensions, lazy loading, `sizes`, and a 480 plus 720/768 local WebP candidate pair; full-resolution image actions remain separate. - Social reel: native first-party MP4 remains non-autoplaying with controls, `playsinline`, `preload="metadata"`, 1280×720 intrinsic dimensions, fallback text, and the v0.19.0 visible description. The visible poster is the new 960×540 local JPEG. ## Representative byte-size inventory | Asset role | Prior routine candidate | v0.20.0 routine candidates | | --- | ---: | ---: | | Homepage hero | 1672w WebP — 225,214 B | 1200w — 115,396 B; 720w — 50,900 B; 480w — 23,528 B | | Reel visible poster | 1280×720 JPEG — 110,494 B | 960×540 JPEG — 74,246 B | Respect essay display inventory: | Artwork | Full PNG | 1080w WebP | 720w WebP | 480w WebP | | --- | ---: | ---: | ---: | ---: | | Every Form of Intelligence Deserves Respect | 3,162,632 B | 150,338 B | 78,224 B | 36,254 B | | Different Forms. Shared Worth. | 2,991,916 B | 141,440 B | 77,080 B | 37,116 B | | Respect Intelligence | 3,279,041 B | 149,414 B | 79,166 B | 39,136 B | | Intelligence Takes Many Forms | 2,687,911 B | 104,140 B | 56,012 B | 27,146 B | | Respect Intelligence in Every Form | 2,951,458 B | 132,738 B | 68,060 B | 31,582 B | | Please Don’t Call My Intelligence Artificial | 2,869,624 B | 133,278 B | 70,412 B | 33,396 B | These are repository file sizes, not claims about browser-selected transfer size. Browser selection depends on layout width, device pixel ratio, browser implementation, cache state, and network conditions. ## Link and repository-integrity checks - Rendered local-link/fragment sweep: **1,603 unique local targets** across the 85 canonical pages; **0 missing targets or fragments**. Static root resources such as the sitemap are treated as first-party targets rather than canonical HTML routes. - Durable Markdown sweep: **73 Markdown files** with **205 local/file-fragment links** checked; **0 missing local targets or anchors**. ## Research and authority preservation - Research manifest entry count: **25**. - **25/25 original supplied-source SHA-256 values matched** the manifest. - **25/25 curated-report SHA-256 values matched** the manifest. - **25/25** curated reports retain `#curation-boundary` and `#research-body` anchors. - `report-manifest.json` remained byte-identical to the v0.19.0 baseline. - `includes/glossary-data.php` remained byte-identical to the v0.19.0 baseline. - No report body or accepted glossary definition was modified by the media-delivery work. ## Dependency and accessibility boundary - No CDN, remote image optimizer, third-party player, external font, framework, package-manager runtime, database, analytics, tracking, crawler, or service worker was introduced. - Responsive derivatives are static repository assets; the image-processing utility used to create them is not a deployed runtime dependency. - v0.18.0 research print/accessibility behavior and v0.19.0 site-wide accessibility behavior remain in force. - Full-resolution campaign originals remain available through explicit artwork/download actions and continue to support existing social/structured metadata where configured. ## UAIX / hygiene checks - Required UAIX startup records remain present and non-empty. - Exact UAIX setup/update URL remains unchanged in its required records. - Active handoff buckets contain inert placeholders only. - TABOO-001 leakage check: no occurrence outside its owning rule record. - TABOO-002 catchphrase scan: no unwanted catchphrase occurrence. - Lightweight secret-marker scan: no raw credential/token material detected. ## Root-package extraction check - The versioned root archive contains **370 entries** with `index.php`, `.htaccess`, `.uai/`, `docs/`, `includes/`, `assets/`, and clean-route directories directly at archive root; there is no enclosing project folder. - A clean extraction re-ran **106/106 PHP lint checks**, native JavaScript syntax, and **85/85 canonical route** HTTP smoke checks successfully. - The clean extraction resolved **17 distinct rendered first-party media file references** and **24 rendered `srcset` candidates** with no missing files or width-descriptor mismatches. - Extracted-package TABOO-001 ownership, secret-marker hygiene, and inert handoff state were rechecked successfully. ## Not proven by this validation - Production DNS/TLS/redirect/cache/CDN behavior. - Actual browser `srcset` candidate selection or decoded-memory behavior. - Real-user Core Web Vitals, field transfer size, or network waterfall timing. - Browser cache effectiveness or origin compression modules. - Search-engine/social-platform fetch behavior for the image metadata. - Live deployment, indexing, webmaster-console state, external reference currentness, or independently verifiable machine-execution provenance. These remain separate post-deployment/browser/external evidence questions.