Skip to content
HHackerTube.videoSurprise me

Data & Method

HackerTube's data comes from the HackerTube dataset (exported October 2026) and was last imported in Nov 2026. It covers 6,454 videos, of which 4,636 were recommended at least once and are not tied to a single news event.

Pipeline

  1. Every HN comment from 2007 to 2026 was scanned for YouTube links; a comment counts only with at least 100 characters of its own text.
  2. A video is kept if at least three different people mentioned it in at least three different months; up to 20 mentions per video are kept.
  3. Each mention was labelled by a language model with a stance (strong recommendation, recommendation, reference, joke, criticism), a short reason and a verbatim quote.
  4. Each video got a summary, a one-line reason HN likes it, a content type, and one to three of 150 categories.

Labels written by language models are good but not perfect; expect a few percent of stances or summaries to be off.

What we hide by default

Videos that were only cited and never recommended, videos that matter only around one news event, and duplicate uploads of the same talk stay reachable but are kept out of rankings, topic pages and search engines.

Licence and attribution

The data is released under the Open Data Commons Attribution License (ODC-By) v1.0. Credit: HackerTube (hackertube.org) and the open-index/hacker-news dataset. Hacker News data is provided by Y Combinator, and comments remain the work of their authors. Video titles and channel names belong to their creators.