Methodology

Every discourse report follows the same recipe. The recipe is the product.

1. Topic selection

Six times per day, our pipeline scans X for trending conversations across seven verticals: AI, coding, politics, health, business, news, and culture. We pick the topic with the highest signal across credible accounts in each vertical.

2. Source pull

For each topic we pull hundreds of posts from accounts with verifiable context — practitioners, researchers, journalists. We do not summarize articles; we work with primary posts.

3. Perspective clustering

We cluster the posts by perspective — distinct positions on the topic — and weight each cluster by the number of credible voices it contains. This is what enforces balance.

4. Quote selection

From each cluster we pick verbatim quotes that carry the position cleanly. Every quote links back to the original post.

5. Editorial review

A human editor reviews the report before publication. The editor can reject any draft that fails the balance, sourcing, or accuracy checks.

6. Corrections

If we publish something wrong, the correction is filed publicly with a timestamp. We never silently edit. See corrections.

FAQ

Read the FAQ on the landing page for the most common questions.