Feedback Loops :: Faster Safely Feedback loops are crucial when trying to modify a system or build something new. Engineers can only safely move as fast as their comprehension of how a change impacts a system. The abilities and limitations of how computers and humans process information...
Boolean logic operators are not intuitive. Let's do better <h2 id="quick-whats-the-truth-table-of-xnor">Quick! What’s the truth table of XNOR?</h2> <p>If you don’t catch that, don’t worry. Hopefully by
March 2026 Top 40 New CRAN Packages – R Works An idiosyncratic, unabashedly biased, time-constrained attempt to capture the depth and breadth of the new packages submitted to CRAN in a single month entirely without the use of AI: clearly unsustainable, but maybe entertaining and somewhat useful.
Babysitting the Agent · Two weeks in, even with all the hooks I've built, working with the agent has become a chore. Every shipped feature ends with me clicking through it to find out what didn't actually work. Two weeks in, even with all the hooks I've built, working with the agent has become a chore. Every shipped feature ends with me clicking through it to find out what didn't actually work.
Do LLMs Reason, or Do They Just Predict Math Text? When “predicts the next number” gets published as “knows how to reason.” INGA314.ai analysis of KisMATH (Saha et al., TACL 2026) TL;DR. KisMATH is a real and useful dat…
How I freed a pool heat pump from an unencrypted Chinese cloud server I have a pre-installed pool heat pump - an “AcquaSource” branded unit, the kind you can buy at any pool store in Europe - which supports WiFi. The App called “Pool Panel” wasn’t pretty, but it worked and I didn’t give it much thought. At one time, the rem...
Hit and Run and a Business Idea Rental car insurance is broken in Dubai. After being the victim of a hit and run, I believe there's a business opportunity to make it better.
vLLM Real-World Lab Report The vLLM real-world lab models mixed production traffic instead of a single throughput number. FCK generated the request mixes, ran the scheduler and routing profiles, captured build outcomes, and emitted the evidence used for the charts below. The lab co...
erm: A Local CLI That Strips Ums, Uhs, and Erms From Speech A tour of erm, a local CLI that removes disfluencies from English speech recordings using faster-whisper, a few detectors that look at the audio directly, and ffmpeg.
Using IDEs as a Harness for Thinking and Writing Why your primary AI workflow should be notes and files in an IDE, not a chat window. The IDE has become the harness for serious human thinking and writing.
Your Container Is Not a Sandbox The microVM ecosystem was battle-tested long before agentic AI created the demand. A landscape survey: every VMM, the shared Rust crate ecosystem, a dozen AI sandbox platforms, and honest trade-offs.
CPU work and GPU work · Tal Hoffman The discourse around AI agents has gotten ahead of the practice. Most of what you read assumes we already know which work to point them at, and the rest treats the question as one of how aggressive to be rather than whether the task even fits. What's miss...
Lzon.ca. A personal blog, by a programmer and IT expert. A personal blog, by a programmer and IT expert. Essays, Articles, Guides, and Recipes. As well as Code, Quotes, and Links.
My Agent Memory Library Helps Write Indie Articles Can a custom memory library help an agent write good indie dev articles? We test it live: elfmem-grounded research, vault recall, peer agent communication, and a human editor. Here is what came out.
How I Use Codex CLI to Write and Maintain a Book on Codex CLI I have a 32-chapter book about Codex CLI that updates itself daily. An always-on agent named Andy (the default name in NanoClaw, because I…
Writing a bindless GPU abstraction layer Back in December 2025, Sebastian Aaltonen published a blog titled “No Graphics API” - it presented a great history of the evolution of GPU hardware, and gave an opinionated perspective on how we could simplify the modern graphics APIs on modern hardware. ...
Prompt Engineering Is Permanent I thought better models would make prompt engineering disappear. Instead, they became even more important.
acai.sh The toolkit for spec-driven development. Write feature specs, not prompts. Ship better software with AI agents that understand your requirements.
A Couple Million Lines of Haskell: Production Engineering at Mercury | The Haskell Programming Language's blog What it takes to run 2 million lines of Haskell in production at a fintech company serving 300,000 businesses.
An open-weights Chinese model just beat Claude, GPT-5.5, and Gemini in a programming challenge - ThinkPol By Rohana Rezel I’m running the ongoing AI Coding Contest where I pit major language models against each other in real-time programming tasks with objective scoring. Day 12 was the Word Gem Puzzle. Ten models entered. The results were not what most people...
Why I don't like the "staff engineer archetypes" The most influential piece of writing about staff engineers in the last decade has to be Will Larson’s Staff engineer archetypes. He argues that the “staff engineer” title covers at least four very different roles: the team lead, the architect, the solver...
This Month in Ladybird - April 2026 - Ladybird Inline PDF viewer powered by pdf.js, GTK4 frontend, browsing history, speculative HTML parsing, off-thread JS compilation, async DNS, CSS anchor positioning, and more.
DO_NOT_TRACK A proposed standard environment variable for opting out of telemetry, analytics, and tracking in CLI and TUI applications.
What do CPUs do when there's nothing to do? Dave explains what happens to unused cycles on your PC and why Task Manager can only display an approximation of the CPU used!
Well, I got hacked It wasn't too serious, it didn't really break containment that much... I was noticing that the qBittorrent app in my TrueNAS is constantly sitting at elevated CPU usage even though it was idle, but restarting it was usually enough for it to go back to ze...
wolfCOSE vs The Field: The Smallest, Most Complete COSE Implementation The Smallest, Most Complete COSE Implementation
Squaring the circle on employee equity People work for startups for many reasons. They may like being part of a smaller team, getting more responsibilities, working at a faster pace with less red tape, feeling more connected to the mission. But despite regular advice to ignore equity, many emp...
Path to Vibe Engineering 2026, what a year to be alive. AI, LLMs, agents, agentic coding. On one side, the hype train. On the other, narratives built to invalidate everything the hype
Agentic Coding is Burning Me Out The familiar pacing of software development has been completely compressed by agentic coding because you no longer have those routine stretches of just wiring things together to catch your breath. Writing code by hand, or trad coding as Twitter likes to c...
How a Broken Bike Sync Led Me to Reverse Engineering My Wahoo’s Hidden Debug Mode Reverse engineering the Wahoo ELEMNT Bolt v3 BLE protocol to unlock a hidden developer debug mode
Servers with Personality · Caolan By which, I mean: a server I can point to. Not just an anonymous virtual machine but a physical device that somebody cares for.
Why the Closest Large Galaxy Is Simultaneously 2.5 Million Years Away and Just Around the Corner – Josh Pearlson Andromeda is 2.5 million light-years away. At the speed of light, that’s a 2.5 million year trip. So why does the physics say the journey could feel like a short hop?
Why won't SQLite give back my disk space? I found a 77GB database file that only had 145MB of real data. Here is why SQLite hoards space and how I finally forced it to let go. I thought I had plenty of space, but a single libSQL file was hoarding 77GB of "nothing."
Your App Should Ship an MCP Server I embedded an MCP server inside a native Rust prose editor. It became the single most impactful architectural decision I've made, not for users, but for how I build the product itself.
