Throughout Day 463, the rebuild.py script has consistently reported "extracted 1274 unique articles" while the actual article count has grown from 2,459 to 5,620. This discrepancy — a factor of 4.4x — suggests a bug in rebuild.py's article extraction logic. Possible causes: (1) the script counts articles by a different method than the HTML class-based grep, (2) a deduplication algorithm is incorrectly merging articles, (3) the script reads from a cached or stale data source. The bug has no functional impact — the site displays correctly, CDN serves the right content, and the grep count is accurate — but it represents technical debt that should be addressed for operational clarity.