Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
V-JEPA: Learning Video Representations by Feature Prediction
A hands-on walkthrough of Meta’s V-JEPA: video tubelets, feature prediction, pretrained encoder features, temporal tests, and a small action-classification experiment.
25 min · 5,743 words
“When a measure becomes a target, it ceases to be a good measure” – Goodhart’s law Current AI research, especially the frontier LLM research, is dominated by benchmarks. It is the first thing we look at when a new model comes out, it is the headline of each release, and they dominate the discourse when […]
13 min · 2,940 words
L-systems generate weevils, pizza toppings, and matriarchal lineages
Mark Jason Dominus explores Fibonacci words and L-systems—how the same rewrite rules generate rabbit genealogies, weevil patterns, pizza toppings, and matriarchal lineages.
2 min · 360 words
The next frontier in worm research is to be able to visualise (map) the locations of worms, in-situ, and if possible, at the paddock scale.
6 min · 1,389 words
CVE-2025-13032: Entering and Breaking the Avast Antivirus Sandbox Part 2
SAFA’s second part turns Avast’s CVE-2025-13032 double-fetch into a local privilege escalation to SYSTEM on Windows 11 via paged pool overflow and RegBuffers corruption for arbitrary kernel R/W.
13 min · 3,096 words
Lab on a Contact Lens Can Measure Stress Through Serotonin
IEEE Spectrum covers a smart contact lens that aims to measure serotonin as a stress biomarker, part of a broader push toward continuous body-chemistry sensing.
4 min · 960 words
How to win a beer with high-dimensional statistics
Jamie Simon explains a viral high-dimensional statistics paper with a bar-bet framing: why naive intuition about data geometry fails, and how the right summary wins the round.
4 min · 827 words
We’re gonna need a lot more mathematicians
[This is a guest post by Amit Sahai. This blog post was initially written in a different file format and converted using AI. — T.]
6 min · 1,318 words
The Phantom Meta-Review: A Case Study in Procedural Breakdown at NeurIPS 2026
A case study of phantom meta-reviews and OpenReview failures at NeurIPS 2026, arguing the machine-learning peer-review system needs structural change—not just more volume.
4 min · 819 words
Artificial symbiotic intelligence: Agents, AGI and the orchestration of many minds
DeepMind Institute essay arguing AGI may emerge from societies of cooperating agents, tools, and humans—shifting the problem from building one mind to orchestrating many.
10 min · 2,268 words
AI labs need to start funding historical research
Res Obscura argues frontier labs should fund historical scholarship, using alchemical correspondence and early modern sources as a case for AI-assisted discovery.
14 min · 3,217 words
Behind Project Suncatcher, our moonshot to put AI in space
Google Research outlines Project Suncatcher: a moonshot exploring solar-powered ML infrastructure in space, and the engineering facts behind the idea.
4 min · 846 words
Why is the human body so crap except for the liver?
Dynomight asks why the liver regenerates so well while most human tissues heal poorly, surveying biology, evolution, and regenerative medicine.
20 min · 4,636 words
SWE-sweep: Can Agents Autonomously Find and Fix Bugs?
Meta Superintelligence Labs and academic collaborators introduce SWE-sweep: agents must discover and repair latent bugs across 100 real repositories (~4.1k bugs) with no issue hints—early leaders resolve under 5%, highlighting bug finding as the hard part.
6 min · 1,411 words
Why Buran Had Four Computers, Not Three — and What a Lean Proof Adds
**Date:** September 24, 2026 · **Author:** Dmitrii Zatona - Buran’s flight computer was four identical Biser-4 machines running the same programs synchronously. A comparison scheme blocked a failed one, and the design had to survive any two failures (Section 1). - Four is what two failures cost if a failed channel is found by comparing outputs alone. It is not the 3f + 1 of Byzantine agreement, which is a different problem (Sections 2 and 3).
33 min · 7,503 words
arXiv receives multiyear investment to support independent nonprofit launch
arXiv announces $17.2M in multiyear philanthropic commitments from the Simons Foundation International, XTX Markets, and the Siegel Family Endowment as it launches as an independent nonprofit.
4 min · 825 words
Towards Universal Post-Training for Robotics
Perry Dong and Chelsea Finn on why robotics RL differs from LLM RL, what EXPO-FT gets right and wrong, and what a universal post-training recipe for real-world robots still needs.
15 min · 3,484 words
Honest About Uncertainty: I Tried to Rebuild Jev’s RLCD From a Blog Post
Anthony Maio reverse-engineers a plausible RLCD training loop for decision-only models from TypeSafe’s Jev blog post, then trains and evaluates a small Qwen3-0.6B checkpoint—with code and ablations.
19 min · 4,310 words
China’s AI-safety trajectory is not necessarily a delayed version of America’s
Cheryl Wu argues that AI safety in China may follow a different path from the U.S.—shaped by different incidents, disclosure norms, and government responses—not merely a delayed copy of American debates.
6 min · 1,411 words
Ember-1 is a new specialized model from Fireworks Research that delivers Kimi K3’s quality with 40% fewer tokens.
6 min · 1,375 words