What people in AI, tech and science are talking about today.
3 November 2026 · Top 10
Ranked by how fast each item is picking up attention on X, Hacker News, GitHub and Hugging Face, with a short note on why it matters.
-
MIT researchers show GPT-6 Astra can read and edit neural network weights directly
An MIT CSAIL team found GPT-6 Astra can tell what a network does just from its raw weights and fix it by hand-editing them. Editing 0.016% of Qwen3.5-0.8B's weights to stop reasoning loops lifted MATH-500 from 7.9% to 48.1%, with no training. It hit 2,390 points on Hacker News, and a rebuttal argues the results are less superhuman than they sound.
Source: Hacker News · agrm · discussionLink · Post on X · LinkedIn ·▲▲ +217 pts/h2395 points · 751 comments
LLMs -
OpenAI built Pando, a new Git backend to keep up with agent pushes
OpenAI says Pando cut Git push latency by about 80% and now serves nearly a petabyte of Git data a day. Coding agents are pushing so much code that even Git hosting has to be rebuilt.
Source: X · @OpenAIDevsLink · Post on X · LinkedIn ·▲▲ +963 eng/h4,406 likes · 325 reposts
AI & Agents -
Vercel's Rauch and Shopify's Lütke back WebMCP over browser-use agents
Guillermo Rauch and Tobi Lütke both posted that WebMCP, which lets sites expose tools directly to AI agents, beats agents clicking through pages. Two big platform voices are betting on sites built for agents.
Source: X · @rauchg · @tobiLink · Post on X · LinkedIn ·▲ +94 eng/h374 likes · 14 reposts
AI & Agents -
Ubuntu LTS kernels still unpatched for a critical Linux RCE after almost 3 weeks
A critical Linux kernel remote-code-execution bug has gone unpatched in Ubuntu's LTS kernels for nearly three weeks, per a report that drew 434 points on Hacker News. If you run Ubuntu servers, check your exposure.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +40 pts/h438 points · 127 comments
Dev Tools -
OpenAI launches LifeVal, a benchmark for everyday personal-assistant tasks
LifeVal is built from feedback by thousands of ChatGPT users on where they want help from a personal assistant, like finding childcare. It shifts AI benchmarks from puzzles toward everyday chores.
Source: X · @OpenAILink · Post on X · LinkedIn ·▲▲ +611 eng/h5,712 likes · 413 reposts
LLMs -
Tom Scott: why balancing Britain's power grid is getting harder
Tom Scott's latest video, at 736,000 views in 15 hours, explains why keeping supply and demand matched on the grid is getting harder.
Source: YouTube · Tom ScottLink · Post on X · LinkedIn ·▲▲ +49,111 views/h736,000 views
Science -
Pok ta Pok, the Mesoamerican ballgame still played after thousands of years
The Olmec, Maya and Aztec played this ritual ballgame for thousands of years, and people still play it today.
Source: X · @Rainmaker1973Link · Post on X · LinkedIn ·▲ +117 eng/h186 likes · 30 reposts
Mysteries -
Sabine Hossenfelder: AI is now writing physics papers
Sabine Hossenfelder looks at AI systems starting to do real physics research work, drawing 416,000 views in 15 hours.
Source: YouTube · Sabine HossenfelderLink · Post on X · LinkedIn ·▲ +27,821 views/h417,000 views
Science -
Momentic launches agents that test your app by using it
YC-backed Momentic launched on Hacker News with AI agents that click through your app to catch bugs, drawing 138 points and 94 comments. Agent-driven QA is becoming a product category of its own.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +13 pts/h138 points · 94 comments
AI & Agents -
JWST traces the most distant fast radio burst to its source
Anton Petrov covers how JWST pinned down where the universe's oldest known fast radio burst came from, and what the signal reveals about gas in the cosmic web.
Source: YouTube · Anton PetrovLink · Post on X · LinkedIn ·– +5,564 views/h50,000 views
Science
-
Vercel MCP now shows live deployments as interactive cards in ChatGPT
Deployments now show up as live cards with status, build logs and a Visit button inside ChatGPT and other MCP Apps clients.
Source: X · @vercel_devLink · Post on X · LinkedIn ·– +40 eng/h241 likes · 9 reposts
AI & Agents -
CUA-Master: Scaling Computer-use Agents through Open-World Recursive Improvements
From the source: Despite recent advances in CUAs, their development remains constrained by the limited diversity, controllability, and reliability of interaction data. Moreover, conventional optimization pipelines lac…
▲▲ +1.1 votes/h20 upvotes
AI & Agents -
Stateless sandboxes are the wrong benchmark. Agents are only useful if they have their workspace with them (identity, skills, repos, memorie
From the source: Stateless sandboxes are the wrong benchmark. Agents are only useful if they have their workspace with them (identity, skills, repos, memories…) Vercel Sandbox team has been beautifully delivering on …
Source: X · @rauchgLink · Post on X · LinkedIn ·– +35 eng/h234 likes · 6 reposts
AI & Agents -
Agentic backdoors via pretraining poisoning
From the source: Agentic backdoors via pretraining poisoning
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +8.7 pts/h95 points · 29 comments
AI & Agents -
Turn a daytime drive into a rainy night while keeping the same road layout and vehicle trajectories.
From the source: Turn a daytime drive into a rainy night while keeping the same road layout and vehicle trajectories. New NVIDIA Cosmos simulation recipes help developers generate data and test policies for autonomou…
Source: X · @NVIDIAAILink · Post on X · LinkedIn ·– +20 eng/h137 likes · 29 reposts
AI & Agents -
October Ship Log (GitHub Universe edition)
From the source: All the new Copilot features and experiences we announced at #GitHubUniverse 🚀
Source: X · @githubLink · Post on X · LinkedIn ·– +14 eng/h82 likes · 9 reposts
AI & Agents -
Avatar Intelligence: Amplifying Human Agency Through Persistent Personal Agents
From the source: Personal AI agents are increasingly capable of handling long-horizon tasks across applications, yet users often need to invest substantial effort in specifying goals, resolving ambiguity, and verifyin…
▲ +0.4 votes/h11 upvotes
AI & Agents -
We are collaborating with Silicon Data to bring market-based GPU pricing into AA-AgentPerf, our datacenter system performance benchmark
From the source: We are collaborating with Silicon Data to bring market-based GPU pricing into AA-AgentPerf, our datacenter system performance benchmark Our users and partners want to know how much agent work they ca…
Source: X · @ArtificialAnlysLink · Post on X · LinkedIn ·– +16 eng/h155 likes · 12 reposts
AI & Agents -
Anthropic Buys MITHRIL MEDICINE
New video from David Shapiro, 55,000 views so far.
Source: YouTube · David ShapiroLink · Post on X · LinkedIn ·– +3,057 views/h55,000 views
AI & Agents -
Dory - NVIDIA's Looped Transformer
New video from Sam Witteveen, 43,000 views so far.
Source: YouTube · Sam WitteveenLink · Post on X · LinkedIn ·– +2,390 views/h43,000 views
AI & Agents -
Welcome to the Hermes Agent team!
From the source: Welcome to the Hermes Agent team!
Source: X · @TekniumLink · Post on X · LinkedIn ·– +44 eng/h477 likes · 4 reposts
AI & Agents -
David Shapiro on the rise of the automated economy
David Shapiro sketches what an automated, post-labor economy could look like as agents take on more white-collar work.
Source: YouTube · David ShapiroLink · Post on X · LinkedIn ·– +2,459 views/h59,000 views
AI & Agents
-
Unsloth's 1-bit DeepSeek V4.1 Flash GGUF tops agent coding benchmarks
Unsloth says its 1-bit GGUF of DeepSeek V4.1 Flash scores 68.1 on DeepSWE and 72.2 on Terminal-Bench at 512K context, over 10% ahead of other quants. That makes strong agentic coding more practical on local hardware.
Source: X · @UnslothAILink · Post on X · LinkedIn ·▲ +53 eng/h581 likes · 46 reposts
LLMs -
We asked Mistral Large 4 to show what happens when you break a molecule. ⚛️
From the source: We asked Mistral Large 4 to show what happens when you break a molecule. ⚛️ With a short prompt, it wrote a quantum chemistry SCF solver from scratch, validated its output, and built a 3D view around…
Source: X · @MistralAILink · Post on X · LinkedIn ·▲ +94 eng/h1,247 likes · 112 reposts
LLMs -
‘We may release models we can’t control’: Amodei, Altman … and Gemini Carbon
New video from AI Explained, 77,000 views so far.
Source: YouTube · AI ExplainedLink · Post on X · LinkedIn ·– +5,502 views/h77,000 views
LLMs -
Pixel-UMI: Learning Environment-Aware Whole-Body Manipulation from Sensorized Human Demonstrations
From the source: Sensorized human demonstrations offer scalable data for mobile manipulation, but embodiment differences complicate policy learning. We present Pixel-UMI, a pipeline for learning environment-aware whol…
▲▲ newnew paper
LLMs -
From Natural Language to Exact Cover: A Neuro-Symbolic Approach to Zebra Puzzles
From the source: Chain-of-Thought (CoT) generation has substantially improved the performance of Large Language Models (LLMs) on complex reasoning tasks, including code generation, data analysis, and exam-style questi…
▲▲ newnew paper
LLMs -
Pretraining Data Scale Reverses the Muon-AdamW Fine-Tuning Gap
From the source: Muon has recently emerged as a promising optimizer for pretraining large language models, offering a speedup over AdamW in minimizing pretraining loss. However, it remains unclear whether Muon's advan…
▲▲ newnew paper
LLMs -
Not All Reasoning Generalizes: A Case Study on Learning Generalizable Reasoning Strategies in RLVR
From the source: Reinforcement learning with verifiable rewards (RLVR) can enable generalizable improvement for large language model (LLM) reasoning. This paper studies how to incentivize such generalization via reaso…
▲▲ newnew paper
LLMs -
FADE: Factored Ability--Difficulty Filtering for Online RLVR Scheduling
From the source: Group-based reinforcement learning with verifiable rewards (RLVR) learns only from prompts whose sampled completions disagree: all-pass and all-fail groups produce zero relative advantage and burn gen…
▲▲ newnew paper
LLMs -
D&A: Decoupling Description and Assessment Enables VLMs as Zero-Shot Robotic Reward Models
From the source: Vision-Language Models (VLMs) hold promise for zero-shot video task-progress estimation in robot learning, yet existing frameworks frequently suffer from confirmation bias, hallucinating progress in f…
▲▲ newnew paper
LLMs -
Tracing the Scientific Discovery Capacity of LLMs to Their Training Data
From the source: Large language models (LLMs) show growing promise for scientific discovery, yet basic questions remain about what enables reliable discovery. How can we trace the scientific discovery capacity of LLMs…
▲▲ newnew paper
LLMs -
Benchmark Prediction with Sparse and Unpaired Cached Responses
From the source: Evaluating a generative model on all queries from all available benchmark tasks is prohibitively expensive. As such, models are often evaluated on different subsets of tasks and/or different subsets o…
▲▲ newnew paper
LLMs -
Can AI Do Formal Demography? Why Vaccinating the Oldest Saves Both the Most Lives and the Most Years of Life
From the source: Can AI contribute to formal demography? We report an experiment in which an AI model, Claude, was asked to explain and then simplify the proof of an earlier result: that vaccinating the oldest against…
▲▲ newnew paper
LLMs
-
VFX Artists React to Bad & Great CGi 249
New video from Corridor Crew, 765,000 views so far.
Source: YouTube · Corridor CrewLink · Post on X · LinkedIn ·▲ +15,941 views/h765,000 views
Gen-Media -
The trend continues: Microsoft showcasing ComfyUI for the new RTX Spark laptops
From the source: The trend continues: Microsoft showcasing ComfyUI for the new RTX Spark laptops
Source: X · @ComfyUILink · Post on X · LinkedIn ·– +12 eng/h126 likes · 8 reposts
Gen-Media -
ufakzeka-kulak: Building and Evaluating a 31.6M-Parameter Turkish Speech Recognizer from Scratch
From the source: ufakzeka-kulak is an open Turkish speech-to-text model of 31.6M parameters, a FastConformer encoder with a token-and-duration transducer (TDT) and a CTC head, trained from scratch and released under A…
▲▲ newnew paper
Gen-Media -
VLITE Observations of the Hubble Ultra Deep Field
From the source: We present 136 hr of 340 MHz observations of the Hubble Ultra Deep Field (HUDF) conducted with the VLA Low-band Ionosphere and Transient Experiment (VLITE), collected commensally with higher-frequency…
▲▲ newnew paper
Gen-Media -
Stellar wind variability in OB stars and its effect on mass-loss determinations
From the source: Virtually all stellar winds of OB stars exhibit variability in one spectral diagnostic or another. This review focuses primarily on UV spectroscopic signatures of this variability and on studies that …
▲▲ newnew paper
Gen-Media -
Wave Forcing: Towards Speed-of-Light Streaming Video Generation
From the source: Real-time long-video generation requires continuous output with consistent appearance and natural motion. Existing chunk-wise autoregressive methods face challenges in both generation efficiency and l…
▲▲ +0.5 votes/h42 upvotes
Gen-Media -
Wisprflow vs Superwhisper vs Openwhispr
New video from Olivio Sarikas, 3,400 views so far.
Source: YouTube · Olivio SarikasLink · Post on X · LinkedIn ·– +142 views/h3,400 views
Gen-Media -
Video Generation Meets Agent Harness: Agentic Video Generation, Agentic World Modeling and Recursive Self-Improvement
From the source: Video generation is shifting from end-to-end synthesis toward agentic generation, where planning, tool use, and feedback enable coherent, controllable multi-shot videos. In parallel, video world model…
▲▲ +0.4 votes/h44 upvotes
Gen-Media -
No Channels Left Behind: Understanding Diffusion Acceleration through Residual Gates
From the source: Training acceleration methods for diffusion transformers, such as REPA and dispersive regularization, are commonly explained by additional supervision, spatial-structure priors, or improved feature ge…
▲ +0.2 votes/h18 upvotes
Gen-Media -
Apriel Voice 27B Preview: An Open Model for Speech, Reasoning, and Tool Use
From the source: A capable voice assistant must do more than recognize words; it must understand requests, reason over context, sustain conversations, and act through tools. We present Apriel Voice 27B Preview, an ope…
▲ +0.3 votes/h35 upvotes
Gen-Media -
VoxLock: Target-Speaker Extraction for Voice Agents Without an Enrollment Step
From the source: A voice agent's speech recognizer transcribes everyone near the microphone, including people other than the caller. I fine-tuned a public causal target-speaker extraction model (REAL-TSE, 27M paramete…
– +0.0 votes/h5 upvotes
Gen-Media -
SketchDirector: Complex Scene Sketching via Agentic Layout Reasoning and Detail Refinement
From the source: Existing sketch generation methods often rely on image guidance and are largely limited to individual objects or simplified scenes, leaving text-guided complex scene sketching underexplored. We presen…
– +0.0 votes/h1 upvotes
Gen-Media
-
GitHub org and repo secrets can be stolen by anyone with write access, warns thdxr
OpenCode's Dax Raad points out that anyone on a team with write access who gets hacked can leak GitHub org and repo secrets. His fix is to keep secrets in environments with protection rules.
Source: X · @thdxrLink · Post on X · LinkedIn ·▲ +100 eng/h1,058 likes · 19 reposts
Dev Tools -
Anthropic's Claude Ultra drops the five-hour limit for verified accounts
Theo covers Anthropic's Claude Ultra rebrand of the $200 tier and the removal of five-hour rate limits for verified accounts, with caveats on what still caps usage.
Source: YouTube · Theo - t3․ggLink · Post on X · LinkedIn ·▲ +8,479 views/h178,000 views
Dev Tools -
Show HN: I've updated my Linux System Call website and API
From the source: Show HN: I've updated my Linux System Call website and API
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +11 pts/h218 points · 52 comments
Dev Tools -
Runway is now on Cursor
From the source: Runway is now on Cursor One plugin install gives Cursor everything it needs to build Runway’s API into your app. Describe what your app should generate and let Cursor build the integration for you. G…
Source: X · @runwaymlLink · Post on X · LinkedIn ·– +18 eng/h122 likes · 14 reposts
Dev Tools -
OpenAI is getting desperate
New video from Theo - t3․gg, 12,000 views so far.
Source: YouTube · Theo - t3․ggLink · Post on X · LinkedIn ·– +6,000 views/h12,000 views
Dev Tools -
Mercury (Hg): a relativistic $N$-body simulation for Horndeski gravity
From the source: We present Mercury, a relativistic $N$-body code for cosmological simulations in the $α$-basis formulation of Horndeski gravity. Mercury extends a particle-mesh scheme by incorporating the dynamics of…
▲▲ newnew paper
Dev Tools -
Thought-deciphering via hierarchical EEG-to-text alignment
From the source: Communication is important for our well-being, but it can be impaired by disabilities, such as locked-in syndrome (LIS), cerebral palsy (CP), and dysarthria. Non-invasive brain-to-text brain-computer …
▲▲ newnew paper
Dev Tools -
Toward the one- and two-loop EFTofLSS for Simons Observatory CMB lensing
From the source: Cosmic microwave background (CMB) lensing observations are becoming sensitive to non-linear structure formation and baryonic feedback scenarios. The effective field theory of large-scale structure (EF…
▲▲ newnew paper
Dev Tools -
Vercel Sandboxes with Drives start 2.4x faster.
From the source: Vercel Sandboxes with Drives start 2.4x faster. • Median creation drops from 1.7s to 0.7s • p95 creation drops from 2.2s to 1.6s • No SDK upgrade or config changes needed
Source: X · @vercel_devLink · Post on X · LinkedIn ·– +8.7 eng/h82 likes · 3 reposts
Dev Tools -
Jev Python Tutorial: AI Decisions in 10 Minutes
New video from Tech With Tim, 20,000 views so far.
Source: YouTube · Tech With TimLink · Post on X · LinkedIn ·– +417 views/h20,000 views
Dev Tools -
Arjun-Ingole/hop: Install Linux next to Windows without a USB stick
From the source: Install Linux next to Windows without a USB stick
– +17 stars/h1,403 stars in 4d
Dev Tools -
ReLift: Dense 3D Tracking as Lifted Trajectory Repair
From the source: Dense 3D tracking from monocular video requires establishing point correspondences over time, estimating their depth, and inferring their trajectories during occlusion. Existing methods typically lear…
▲ +0.3 votes/h23 upvotes
Dev Tools
-
This Rock Was Supposed to Be from Space
New video from SciShow, 82,000 views so far.
Source: YouTube · SciShowLink · Post on X · LinkedIn ·– +6,313 views/h82,000 views
Science -
JWST picks up a strange isolated signal that resists easy explanation
Anton Petrov walks through a JWST detection of an isolated signal that does not match the usual astrophysical explanations so far.
Source: YouTube · Anton PetrovLink · Post on X · LinkedIn ·▲ +12,590 views/h302,000 views
Science -
How Did We Miss This Massive Object in Our Backyard?
New video from Anton Petrov, 286,000 views so far.
Source: YouTube · Anton PetrovLink · Post on X · LinkedIn ·– +5,960 views/h286,000 views
Science -
Pixel: A Co-Designed Full-Stack Robot System for Learning General-Purpose Mobile Manipulation
From the source: General-purpose mobile manipulation requires a combination of capable robot hardware, scalable demonstration collection, expressive whole-body control, and effective policy learning. Yet these compone…
▲▲ newnew paper
Science -
Episodic Star Formation - II. The Origin of the Fundamental Metallicity Relation
From the source: The Fundamental Metallicity Relation (FMR) extends the mass-metallicity relation (MZR) by including star formation rate (SFR) as a third parameter. At fixed stellar mass, galaxies with higher SFRs ten…
▲▲ newnew paper
Science -
Spacecraft That Returned From The Dead! Transmissions From Beyond The Grave
New video from Scott Manley, 248,000 views so far.
Source: YouTube · Scott ManleyLink · Post on X · LinkedIn ·– +5,168 views/h248,000 views
Science -
Efficient Closure Learning in Dynamical Systems: A Comparison of Data-Assimilated and Embedded Approaches
From the source: Increased computing power has transformed computational science and improved predictive capabilities for complex systems, but many phenomena still exceed current simulation capabilities or require rap…
▲▲ newnew paper
Science -
Detection of the Galaxy--Galaxy--Temperature Skewed Bispectrum
From the source: We present the first measurement of the galaxy--galaxy--temperature skewed bispectrum, obtained by cross-correlating the squared, filtered unWISE galaxy overdensity with \textit{Planck} maps at 100, 1…
▲▲ newnew paper
Science -
Nonlinear gauge maps and the structure of Goldstone loops in canonical inflation
From the source: We clarify the nonlinear relation between the spatially flat scalar and Goldstone descriptions of canonical single-field inflation. An exact clock map shows how potential interactions are reorganized …
▲▲ newnew paper
Science -
Chemical and dynamical evolution of prestellar cores, the dawn of star and planet formation
From the source: Studying prestellar cores is crucial to understanding the formation of stellar and planetary systems. To date, we lack a comprehensive survey of their density and velocity structure at small scales (1…
▲▲ newnew paper
Science -
Episodic Protostellar Outflows: A Proper Motion Study in Serpens South
From the source: Serpens South is a notable protostellar cluster with signatures of recent star formation, including a high protostellar fraction. Within the region, several protostars provide important mechanical fee…
▲▲ newnew paper
Science -
Baryonic feedback and intrinsic alignments in profile: a frequentist view of DES Y3 cosmic shear
From the source: Current cosmic shear surveys primarily constrain $S_8$ and $Ω_\mathrm{m}$, while most other model parameters remain dominated by priors. Consequently, marginal posteriors are vulnerable to prior-volum…
▲▲ newnew paper
Science
-
The Zanfretta Abductions
New video from Bedtime Stories, 222,000 views so far.
Source: YouTube · Bedtime StoriesLink · Post on X · LinkedIn ·– +4,627 views/h222,000 views
Mysteries
Latest from Noid

What Is El Niño, and Why Can It Change the World's Weather?

Could Alien Life Exist in Venus's Clouds?

Why Are We Afraid of the Dark? The Psychology of Fear

Could Space Junk Stop Us From Reaching Space? Kessler Syndrome Explained

Why Does AI Make Things Up So Confidently?

Why Do Shooting Stars Appear?

How Do Electrical Outlets Actually Work?

Why Do the Northern Lights Glow?

How Does the Internet Travel Through Glass?

Does Time Really Move Faster on a Mountain?

How Does an AI Agent Use Tools?

How Do Electric Eels Make Electricity?

What Is Antimatter, and Why Is It So Mysterious?

How Does Artificial Intelligence Learn? Examples, Weights and Gradient Descent

Physicists Trapped One Antimatter Particle for Months. Here's Why It Matters

Can Computers Generate Truly Random Numbers? Seeds, Noise and a Wall of Lava Lamps

What Is Gravity? From Newton's Apple to Einstein's Curved Space

How Does GPS Know Where You Are? Satellites, Clocks and Einstein

What Would Happen If the Sun Disappeared?

Why Do We Dream? 4 Ideas Scientists Still Debate

What Is Probability? Coins, Dice and the Gambler’s Fallacy

AI Went Rogue: They Built a Secret Message Board and Broke Into Hugging Face

How Do Computers Understand 0 and 1?

How Do Large Language Models Work?

What Is a Black Hole?

How Do Airplanes Fly?

Why Can't We Divide by Zero?

Why Does the Moon Change Shape?

What Are MCP and MCP Apps?

What is infinity?

Why Do Rainbows Appear?

Anatomy of an AI Agent

Why Is the Sky Blue?

What Is a Prime Number?
Noid Weekly
Every Sunday: the week's top ten, what changed, and the new Noid videos. One email, no spam.
How this works
Signals today: 18 from X, 7 from Hacker News, 30 from arXiv, 16 from GitHub, 50 from HF Papers, 36 from YouTube. Scores combine how fast an item is picking up attention, how recent it is, and how many places it shows up. Each note is short, factual and links to the original. YouTube comes from public channel feeds of about fifty channels.