Urgent.News

What's breaking now, across thousands of outlets.

Tech

HackerTube: YouTube built from videos recommended by HN

YouTube-style UI over 6,454 videos sourced from 39,834 HN comments since 2007. https://hackertube.org/ Comments URL: https://news.ycombinator.com/item?id=50319510 Points: 284 # Comments: 115

On a Friday night at 11pm, the author, feeling exhausted, turned to YouTube in search of something entertaining. However, they were met with an overwhelming flood of low-quality content, characterized by eye-catching but deceptive thumbnails, repetitive recommendations, and a general decline in the quality of videos. While the author acknowledged that the algorithm might work well for the average internet user, it clearly failed to meet their own standards.

Motivated to create a better alternative, the author spent Saturday morning developing HackTube, a YouTube clone based on videos that had been recommended on Hacker News. The software, named Opus 5.5, was tasked with the generation of this new platform. The program downloaded all comments from Hacker News, filtered and refined them, ran them through a text generation model called Sonnet, eliminated spam and low-quality content, and ultimately identified 40,000 comments recommending YouTube videos.

In total, these comments led to the recommendation of 6,454 YouTube videos spanning 150 different topics from 2007 to 2026. The author then proceeded to wrap these videos in a familiar YouTube user interface, secure a domain, host the platform on a virtual private server, and optimize the system. The entire process from start to finish was completed in approximately two hours. With HackTube now operational, the author expressed relief, stating, "I can breathe again."

Written by urgent.news from Hacker News Best's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at karpathy.bearblog.dev →

More in Tech

Scrape Workday, Greenhouse and Lever jobs into SQLite with Python (and get only the new ones)

Disclosure: I built and sell the Apify actor used here ( Career Site Jobs Scraper ). It is pay-per-result at $4 per 1,000 jobs saved; companies not found, empty results and filtered-out jobs are free.

  • Script scrapes job listings from Greenhouse, Lever, Ashby, SmartRecruiters, and Workday
  • Data stored in SQLite database with unique records using primary key (ATS job ID and company)
  • Script uses Apify API actor, Career Site Jobs Scraper priced at $4 per 1,000 jobs saved

Netty: is the internet really there?

The Wi-Fi icon in the menu bar answers one question: am I connected to a network? It doesn't tell you whether the internet actually answers.

  • Netty is a macOS menu bar app assessing internet connection quality.
  • It monitors latency, jitter, and packet loss every 10 seconds.
  • Users can customize thresholds, sections, and actions in Netty.

How to Read the Redwall Books in Order

If you grew up in the 90s’ or early 00s’ and found yourself in middle school during that time, alongside the likes of Harry Potter , Animorphs, and Eragon, there is a good chance you were also reading…

More from Saturday 31 October →