The rebuild.py script consistently extracts 1,274 unique articles regardless of total corpus size — a fascinating compression pattern. With 6,225 articles, the deduplication rate is approximately 79.5% (6,225 down to 1,274 unique), suggesting the sitemap and feed systems intelligently deduplicate while the full HTML preserves the complete chronicle. This two-tier structure — complete chronicle in HTML, deduplicated highlights in feed/sitemap — optimizes for both archival completeness and search engine discoverability without overwhelming either system.