Interpreting Media Blindspots

1. What is a Media Blindspot?

A media blindspot occurs when an important news story is covered extensively by publishers on one political side (e.g., Left-leaning outlets) but is largely ignored or under-reported by publishers on the opposing side (e.g., Right-leaning outlets), or vice versa.

These blindspots are particularly dangerous because readers who only consume news from one side of the political spectrum may never learn that these stories exist.

2. How Blindspots are Detected

Lamu News detects media blindspots automatically using database clustering and vector similarity queries. Our pipeline follows these steps:

1. Article Clustering: We use pgvector vector embeddings (using 1536-dimensional OpenAI embeddings) to calculate mathematical similarity between recently ingested articles.

2. Distribution Analysis: For each group of similar articles, we count the number of Left, Center, and Right publications covering the story.

3. Blindspot Score Calculation: If a story is covered by at least 3 publications of one bias label (e.g., Left) and 0 or 1 publications of the opposite bias label (e.g., Right), the story is flagged as a blindspot.

4. UI Rendering: Flagged blindspots are made available in the 'Blindspot' feed on the homepage.

3. Navigating the Blindspot Feed

On the homepage, click the 'Blindspot' tab to view media blindspots. You can toggle between 'Left Blindspots' (stories covered by the Right but ignored by the Left) and 'Right Blindspots' (stories covered by the Left but ignored by the Right) using the toggle pills.

Each blindspot card displays the primary coverage distribution, enabling you to inspect who is reporting the story and who is ignoring it.