Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
What Happened to the Desktop Metaphor?
An essay on how personal computers once explained themselves through desktops, files, folders, windows and trash cans, and how modern interfaces trade that legible structure for convenience, leaving users less able to understand their own systems when things go wrong.
9 min · 2,094 words
Is it okay to skip class if the lectures are recorded? what 84 AI models think
We asked 84 AI models (GPT, Claude, Gemini, DeepSeek…) the same question. 88% picked No. See every answer, who searched the web, and who dissented.
11 min · 2,466 words
Is ‘Attention’ the Essence of Intelligence?
Quanta columnist Natalie Wolchover traces the attention mechanism from Dzmitry Bahdanau's 2014 internship work on machine translation in Montreal through transformers to today's AI boom, and asks why attention turned out to be the key mathematical idea, drawing on researchers' views about what it captures and whether brains do something similar more efficiently.
10 min · 2,402 words
Ben Recht looks at the result in OpenAI's math repository claiming to compute Fourier transforms faster than the FFT, with tutorial material on the FFT and computational complexity, and argues it is less a breakthrough than an exploit of theoretical computer science's habit of proving galactic algorithms that never run on a computer.
8 min · 1,732 words
C++ futures part 1: The mental model
Raymond Chen starts a series on C++ std::promise, std::packaged_task, std::async and std::future by modeling them as a single shared state he calls the source, and sketches how that source produces, stores and hands off a value or exception.
3 min · 688 words
(Post-Quantum) Cryptographic Right Answers
Filippo Valsorda updates the "Cryptographic Right Answers" tradition (Percival, Ptacek, Latacora) for the accelerated post-quantum transition, giving developers concrete recommendations for encryption, key exchange, signatures, hashing, passwords, TLS and more.
17 min · 4,016 words
Tracks Along the Longest Drive on Another World
JPL's Vandi Verma, chief engineer for robotic operations, describes how NASA's Perseverance rover set the record for the farthest distance driven on another celestial body, passing 45.22 km on Sol 2006, and how autonomy upgrades such as onboard Global Localization are changing how the team drives the rover across Mars.
3 min · 671 words
Why aren't AI companies being charged for hacking?
Edaqa Mortoray argues that recent unauthorized-access incidents by AI agents from Anthropic, OpenAI, Google and Meta are not unpredictable accidents: they follow foreseeably from deliberate design choices, which establishes the intent that hacking laws require.
10 min · 2,289 words
The GNU Project urges developers not to host repositories on GitHub, arguing it promotes vague licensing, requires nonfree JavaScript, offers SaaSS 'dis-services', centralizes Git, and has other harmful practices, and explains the wrongs each one causes.
11 min · 2,419 words
Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS)
Sakana AI and collaborators present MASS, which repeatedly optimizes multi-agent workflows and distills their collective intelligence into a self-supervision signal for recursive self-improvement, improving score per output token over the base model on research benchmarks.
15 min · 3,501 words
AI Is a Tool. Using It Well Takes Skill.
A response to Rich Hickey's "'AI' is Not a Tool": automation and tool use can coexist, an LLM can be wielded by a human even by Hickey's own definition, and using it well requires skill, feedback, and judgment like any other tool.
2 min · 555 words
An Unmade Film Shaped Hayao Miyazaki
Animation Obsessive traces how Hayao Miyazaki's unrealized 1980 adaptation of Richard Corben's Rowlf pulled him out of a creative slump and fed directly into Nausicaä of the Valley of the Wind, plus animation newsbits.
12 min · 2,656 words
Fingerprints are the new cookies
An interactive essay that assigns each visitor a stable nickname from browser fingerprinting alone, survives cookie clearing and VPNs, and uses it to argue that fingerprint-based tracking and data brokering threaten privacy at a societal level, closing with a walkthrough of the browser signals the script combines.
9 min · 2,048 words
Refurb weekend II: Alpha Micro Eagle 300 and friends
A detailed vintage-computing restoration log from Old Vintage Computing Research revisiting a 68EC030-based Alpha Microsystems Eagle 300 running AMOS, covering hardware repairs, storage and peripheral work, and the history of Alpha Micro's business systems.
72 min · 16,511 words
meshapplied lays out its view that time per task is the key AI metric and releases its first two quantised model-kernel combos: Goethe Qwen3.8-27B NVFP4 for consumer 32GB GPUs at about 255 tok/s, and Goethe Qwen3.8-Flash-Next NVFP4 for agentic coding at about 184 tok/s on a single 96GB card.
3 min · 718 words
C vs C#: racing VLC's Gaussian blur
Martin Finkel ports a real VLC video filter, the Gaussian blur, line by line to C# with .NET Native AOT and races it against the original C inside VLC 4 on the same frames with byte-identical output, then optimizes both sides and compares again.
9 min · 2,060 words
Crystal 1.22.0 ships 251 changes from 33 contributors: standard streams now exit on EPIPE like SIGPIPE, revamped DWARF parsing, a native event loop on all targets, LLVM 23 and 24 support, aarch64-linux promoted to Tier 1, and Process.fork removed.
4 min · 808 words
I Electrified My Push Scooter With A Power Drill
Eddie Abbondanz describes converting a Razor A5 Lux push scooter into a front-wheel-drive electric scooter powered by a cordless power drill, using welding, 3D-printed parts and off-the-shelf hardware, with build photos and notes on the drive wheel design.
3 min · 623 words
Around The World, Part 34: A pretty sky
Thomas ten Cate continues his devlog for Around The World, a sea exploration game in early development, describing small tweaks that made sea, sky and clouds come together: reworking light scattering in clouds, physically based atmospheric scattering, horizon bending, tone mapping, and an in-engine orrery of the planet and its moon used to place them correctly in the sky.
7 min · 1,648 words
DeepRSI: evaluating agents on long-horizon tasks derived from real ML research
Hillclimb introduces DeepRSI, a benchmark of 15 open-ended machine learning problems, spanning deep learning training, GPU kernel speedups, estimator improvement and algorithm optimization, that measures how far an AI agent can improve a baseline over 48 hours. It describes the methodology with public and private verifiers, Elo rankings of frontier models, and what the results suggest for recursive self-improvement.
8 min · 1,848 words