Taste, Quality, Craft, and Value
A commitment to value over output may prevent prolific AI psychosis.
My Proxmox LVM pool died and I'm perfectly fine with that
I've made a lot of progress in my homelab, but I'm getting ready to scale it back.
When Thinking Goes Sedentary | Sid's Blog
After watching my mum hit chatbots (Meta AI inside WhatsApp) for the most mundane calculations and tasks, my hot take is that we're rapidly entering the default sedentary era of intellectual fitness. And seeing it written out, I realize it's a convenient excus…
![]()
i am a better person without my phone
Screenshot 2026-09-17 2 I am a better person without my phone. In fact, I would go so far as saying there is a negative correlation between the amount...
I guess we live in the P = NP world after all
AI haunts Gödel’s Lost Letter
Backgrounding My Way to Japan
How to learn truth from fiction
Computer programming, and friction as a learning activity
echo $RANDOM. Exploring Software and writing about it.
DigitalOcean has $3 million for DHH and nothing for CSS-Tricks - keith.is
A week and a bit on UpCloud, deleting my DigitalOcean account after more than 10 years, and what DigitalOcean is doing to CSS-Tricks, one of the best places on the whole internet.
The input layer - Magenta Creations
Agents took the codebase out of my head, so deep focus stopped paying. What is left of me in the loop is typing and switching windows, and that is what I am working on now.
GrapheneOS (@GrapheneOS@grapheneos.social)
Android 17 QPR1 is the first release since Android Honeycomb (3.x) adding new APIs for app developers without a release to the Android Open Source Project. The new APIs are currently exclusive to the Pixel OS and aren't available to other Android OEMs. https:…
Jev’s Architecture Unmasked — archerhume
I probed Jev with 10,000 API calls to work out roughly how it’s built, and why most of the grifter takes on X are completely wrong.
Alea iacta non est: rerolling the LLM dice
Like everyone else, I talk a lot to various LLMs. And sometimes I wonder how consistent their answers are. I know that LLMs are non-deterministic, but there's a difference between giving wildly different answers and saying the same thing with different words. …
Claude Code changelog - Claude Code Docs
Release notes for Claude Code, including new features, improvements, and bug fixes by version.
I cancelled my personal Claude subscription
The claudish got too bad and I don't want to pay for an AI system that has gotten worse. Opus 4.7 is where it went downhill.
Type safe interpreters
2026-01-17
Query plan rewriting in PostgreSQL - The Consensus
We take a look at some of the ways PostgreSQL rewrites your queries (or not).
Why not now? — cpv
A hello world of sorts.
Keeping vLLM's Prefix Cache Warm Between Agent Turns
What vLLM's cache tiers hold, why coding agent traffic keeps invalidating them, and the Bifrost tool schema bug that cost 90 to 180 seconds on the turns it hit.
Security Through Obfuscation Is Dead - Robin Glauser
How an AI reverse engineered a custom challenge code from examples, uncovered XOR-masked country data and a SHA-256 checksum, and built a working decoder.
Communication Doesn’t Have a Compiler
Communication Doesn’t Have a Compiler
How I Run a Coding Agent on GPT-6 Pro Without Touching My Codex Limits
Codex and Work share one allowance. Regular Chat has its own. Here is how to point a normal ChatGPT conversation at your files and let Astra do the work, step by step, with the safety rails I actually use.
What the Heck Is Jev?! | James O'Reilly
Jev is a fast decision model that chooses from supplied answers and opens up new uses for AI.
Adventures in BSD
Adventures in BSD - What I learned from using FreeBSD as a main OS for a summer
I had strings, but now I am free
This is a fictional short story and no claim made reflects any real world event.
I Hate Workday
Workday makes it a huge pain in the ass to apply to companies.
Senior Engineers Are the Next DRAM Shortage
There’s a shortage coming that nobody’s put on a slide yet: senior engineers. Every company on earth is about to want one who can drive a fleet of agents with judgment — and almost nobody …
A stop sign for $1,000
3 crashes at one intersection, photographed on my way to work. The all-way stop went in 178 days after I sent the city a notice of dangerous condition.
The LLM Comments Are Not For You
I’ve heard a lot of talk about LLMs recently, and among the most common topics of discussion have been the comments. LLM comments generally seem to suck: they are verbose, regardless of surrounding context, they seem to encode conversation decisions (like “do …
A SQL WHERE Clause for Taste | Query.Farm
I pointed DuckDB at Hacker News and at Jev, TypeSafe's decision model, and asked it to score 500 stories against one sentence about what I like. It took 17 seconds and cost less than a penny.
Build Faster Feedback Loops Using Qualitative User Research
Introduction: Why Getting Early Feedback on Ideas or Products Is Useful
Some of All Fears
AI anxiety is rational, but the past 10 days of conversation has been absurd and irresponsible.
EA Safety
EA promises to solve AI Safety. Now we have two problems.
Tyrannosaurus
Quick Tunnels · Cloudflare
Localhost, meet the Internet. Turn the server on your laptop into a public, encrypted URL on Cloudflare's edge with one command.
Multiple Get Methods with the Action Method Selector Attribute in .NET Core 10 | no dogma blog
I wrote about this back in the days of .NET Core 2.0. How to do this has changed in .NET 10.
Bend 2 and the Vibe-Coding Trap
September 18, 2026
NTP, an atomic clock, and having a great time at the world's largest VCF
I'm back from VCF Midwest 21, arguably the largest vintage computer festival on the planet. After being blown away last year, I decided to get a table of my own this year, to demonstrate NTP Time on a bunch of old Macs, representing every era up to the pinnacl…
SemIf — local decisions in your browser
Run typed option logits and autoregressive JSON locally with open models and WebGPU.
I don't like passkeys | Ethan Hawksley
Why passkeys are a step back for personal security: examining account lockout risks, platform lock-in, hardware key limits, and fragile recovery flows.
Hacking OpenAI
A heap overflow and SSO misconfiguration to compromise OpenAI internal repositories
Two techniques for working with System One models
I recently wrote about Jev, a new “System One” language model that only outputs decisions: the answers to a set of user-provided multiple-choice questions. This…
PrismML — Introducing Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint
Ternary Bonsai 2 27B retains 98.2% of Qwen3.8 27B benchmark performance in a 5.9GB footprint, with multimodal and agentic capabilities.
The Front Page of the Internet Is Up for Grabs
The race for the front page of the internet is still going. All the AI companies will fight for it. Benchmarks don't matter when you don't get any views.
GitHub CLI Installation on Ubuntu | no dogma blog
Even if you follow the instructions on GitHub, the GitHub CLI is installed from the Ubuntu repository, which is not the latest version. This post shows how to install it from the GitHub repository.
Bend
Bend: a fast language that blocks AI mistakes via proof.
Introducing Astra for Law
OpenAI for Law brings frontier intelligence for law, custom firm workflows, connected legal data sources, and legal-grade controls for confidential client work.