
Hacker Tube: Stop Watching Slop.
Hacker Tube (HackerTube) is the anti-algorithm video platform for hackers, engineers, and deep thinkers. Every video on Hacker Tube was organically recommended by real humans in Hacker News discussions—with zero clickbait thumbnails, zero tracking, and pure technical signal.

Simple Made Easy
Hacker Tube vs. The Algorithmic Trap
Commercial video platforms optimize for eyeballs and dopamine spikes. Hacker Tube (HackerTube) optimizes for long-term intellectual leverage.

- Exaggerated Clickbait: Open-mouth screaming reaction thumbnails designed to trigger cortisol and impulsive clicks.
- Watch-Time Optimization: Endless recommendations loop the same 5 topics you clicked yesterday, trapping you in an echo chamber.
- Shallow Formats: 60-second shorts and over-edited hype videos that leave zero lasting comprehension.
- 50% Mid-frame Previews: Honest static frames taken from the middle of the lecture. What you see is what you learn.
- 20-Year Human Consensus: Filtered from 40,000+ organic recommendations by engineers with skin in the game.
- Deep 1–2 Hour Masterclasses: MIT lectures, computer architecture deep dives, and foundational mathematical intuition.
Interactive Demonstration: Thumbnail Sanitizer
Toggle below to experience the psychological difference between clickbait baiting and honest 50% timestamp preview.
d_loss / d_weights = (X.T @ d_output) / batch_size
weights -= learning_rate * d_weights # pure math, zero hypeHacker Tube Curated Video Archive


Inventing on Principle
![Mid-frame preview of [1hr Talk] Intro to Large Language Models](https://img.youtube.com/vi/zjkBMFhNj_g/hqdefault.jpg)
[1hr Talk] Intro to Large Language Models

Let's build GPT: from scratch, in code, spelled out

The Future of Programming

Fast Inverse Square Root: A Quake III Algorithm Deep Dive

The Spelled-Out Intro to Neural Networks and Backpropagation: Building micrograd

The Mother of All Demos, presented by Douglas Engelbart (1968)

How to Speak - Patrick Winston (MIT)

'Clean' Code, Horrible Performance

Zero Knowledge Proofs and Modern Cryptography Foundations

The Essence of Calculus, Chapter 1
Hacker Tube FAQ: Curated Video Philosophy
What is Hacker Tube (HackerTube) and why was it created?
Hacker Tube (HackerTube) is an anti-algorithm, human-curated video discovery directory created in response to mainstream YouTube's algorithmic 'enshittified wall of slop'. Instead of optimizing for watch time or clickbait engagement, Hacker Tube populates its feed exclusively with talks, lectures, and documentaries recommended by software engineers across 20 years of Hacker News comments (2007–2026).
What is Andrej Karpathy's role in Hacker Tube?
Andrej Karpathy (former OpenAI co-founder and Tesla AI Director) originated the concept in late October 2026. Frustrated by sensationalized YouTube feeds on a Friday night, he directed Claude Code to parse millions of Hacker News comments and spin up hackertube.org in roughly two hours, sparking a global wave of community-curated frontends.
Why does Hacker Tube use 50% timestamp previews?
Mainstream YouTube creators are pressured by recommendation algorithms to upload exaggerated reaction faces, screaming red arrows, and clickbait graphics. Hacker Tube bypasses this by capturing a neutral, honest snapshot directly from the exact middle (50% mark) of the video, showing you what the lecture actually looks like.
Is there a Hacker Tube Apple TV app or mobile client?
Yes! Developer @jnaina released 'HackerTube for Apple TV' (hackertube-tvos on GitHub), allowing users to browse curated feeds on their TV while leveraging the native YouTube app for seamless streaming. Hacker Tube is also a responsive Progressive Web App (PWA) on mobile.
Can I download the dataset for AI agents or my own projects?
Yes. In keeping with the open-source spirit, the complete structured dataset is available via our public JSON API at /api/dataset, complete with video IDs, categories, Hacker News citation excerpts, and AI key takeaways.
How does Hacker Tube protect user privacy?
Hacker Tube requires zero logins or accounts. No tracking cookies or analytics telemetry are stored on our servers. Your watch history, bookmarks, and preferences stay 100% client-side inside your browser's localStorage.