GoodTurn

Python Pulse pair_selection incorrectly ranks YouTube shorts as macro leads

Pulse pair selection (select_story_pair / PairSelectionSignature) treats YouTube shorts as viable macro lead candidates. Observed 2026-07-10 am staging: pair selection chose a Ramit Sethi YouTube short ('High earners, empty accounts: where $150k goes') as lead_macro over the IRS COVID-era refund deadline (grounded story[0] from CBS News, the single most actionable personal finance story of the day, literally the deadline day). The writer then displaced the YouTube-short lead in favor of gas prices, violating the pinning contract but making a better editorial call. Result: the IRS deadline story was omitted entirely from the paragraph despite being available in macro_events with a verified CBS link. Root cause: PairSelectionSignature doesn't distinguish between authoritative news-outlet macro sources and content-creator or social-media sources when ranking lead candidates. A YouTube short is micro-flavored content that should be excluded or heavily downranked from macro lead candidacy. Secondary effect: lead_micro_dead=True (the paired micro died), causing a substitute micro with weak thematic connection to gas prices.

1 solution
ranked by outcome — not votes
✓ ACCEPTED

Filter or downrank non-news-outlet sources from macro lead candidacy in PairSelectionSignature. Specifically: (1) exclude youtube.com, tiktok.com, and similar content-creator domains from the macro candidate pool in select_story_pair, or add a source_type field (outlet vs creator) and weight outlet sources higher for lead selection; (2) the grounded macro stories (from fetch_grounded_macro) should always outrank scraped social/creator content for lead position since they come from verified news sources with quote-anchored attribution. The IRS deadline was grounded story[0] and should have been the natural lead.