Building Software Requires Digestion
Writing about the big beautiful mess that is making things for the world wide web.
How To Make Your Coding Agent Look Like An Idiot - Carlo Capocasa Tech Blog
Sharing 20 years worth of my best technical knowledge
Branding VS Good design · Reda Antar
Branding isn't merely orthogonal to good design, but opposed to it. Branding by definition has to be distinctive. But good design, like math or science, seeks the right answer, and right answers tend to converge.
Snowboard Kids 2 is 100% Decompiled
Snowboard Kids 2 for the Nintendo 64 has reached 100% matching decompilation. A brief note on what that means, how we got here, and what comes next.
Codex-maxxing - Jason Liu
How I use Codex as a place where long-running work can live.
How to Land a Frontier Lab Job
Vlad's Blog
The Floor Doesn't Exist
AI did not invent new attacks. It billed the old ones monthly. Crypto is the only place we can count it, and the early signals are in.
Three things about data
Some recent conversations have reminded me that I have opinions about data and its use inside organisations. Especially for marketing-type stuff. Here are three of those opinions Gather less of it A few years ago in a sales meeting some ad-tech person said…
The 96% that doesn't depend on you: why GEO is fundamentally an off-page discipline
On average, 96% of the citations that drive a brand's appearance in AI answers come from third-party sites. That single number is why GEO is fundamentally an off-page discipline — and why so many SEOs are struggling to embrace it.
10Gb/s Ethernet: using mini-heatsinks with a 10GBASE-T SFP+ module
A quick update to show the (smallish) benefit of adding mini-heatsinks to a 10GBASE-T SFP+ module
Semantic IDs for vulnerable code: finding 100× more cross-project clones than VUDDY — Shrikar Archak
Learned RQ-VAE Semantic IDs for C/C++ vulnerability clones. Borrowing the TIGER substrate from recsys: on a 5000-function CVE registry SecSid finds 112 cross-project clones; VUDDY finds 1.
Stripe seem friendly to "friendly fraud" - Gingerlime
Friendly fraud is the laundered name for something that the payment system is not really able to prevent. Even though I’m pretty sure they can do way better. Particularly big and sophisticated payment providers like Stripe, with a mountain of signals. I had a …
Keeping the Cultural Uplands Alive
Aerial view of Cape Coral, FL Sam Altman doesn’t understand the technology he’s building. In a recent New Yorker profile, neither he nor anyone …
Target Can't Measure Their Own Bins
I bought a plastic storage bin from Target, and the measurements listed were just incredibly off.
The HTML Sanitizer API
The HTML Sanitizer API is a new browser feature that helps developers prevent XSS vulnerabilities by safely sanitizing HTML content.
Handling the great code forge fragmentation
The kind of developer I hope to be someday It seems like there are a lot of people either leaving or talking about leaving Github, a very prominent one being Mitchell Hashimoto. Fragmentation seems inevitable, as people/companies start to distrib…
The Joys of Free Cloudflare
I’ve been tinkering around with the free tier on Cloudflare and have managed to churn out a couple of side projects. Of course, I had a little help with various AI systems, but it was assisted rather than vibe-coded.
![]()
Moving my app subscriptions to Europe
Having talked about it for so long, it's finally time to put my money where my mouth is The movement for European tech sovereignty is picking up speed nic...
I've Been Shipping 'Multi-Tenant' Wrong for a Decade | Adrià Cidre
Tenant-aware is not tenant-isolated. Postgres row-level security, a restricted database role, and default-deny: the four things I missed every previous time.
Scripts and aliases for your linux
Every person who spends more than five minutes in a terminal runs into the same thing: the same long commands have to be typed over and over, and routine act...
Anthropic acquires Stainless
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
Let's talk about AI slop
Is it the end of open source we know and love?
Particle Census
GitHub - zakirullin/files.md: 🌱 Your life in plain .md files
🌱 Your life in plain .md files. Contribute to zakirullin/files.md development by creating an account on GitHub.
GitHub - stephenlthorn/auto-identity-remove: Automated data broker opt-out runner — removes your personal info from 30+ people-search sites on a monthly schedule
Automated data broker opt-out runner — removes your personal info from 30+ people-search sites on a monthly schedule - stephenlthorn/auto-identity-remove
Don't call yourself a Software Engineer, you are an AI Enabled Engineer.
I can only imagine what it's like learning the skill of programming in this day and age. What does an average college class look like? What is the CS professor teaching? And students, how do you recon
GenCAD: Image-conditioned Computer-Aided Design Generation with Transformer-based Contrastive Representation and Diffusion Priors
GenCAD
The just-say-no engineer was a ZIRP phenomenon
The engineer who says no all the time is a real archetype among senior and staff engineers. Their role is to slow things down, to block the development of features that add complexity, and to ensure that as little code gets written as possible (since code is a…
GitHub - MinishLab/semble: Fast and Accurate Code Search for Agents. Uses ~98% fewer tokens than grep+read
Fast and Accurate Code Search for Agents. Uses ~98% fewer tokens than grep+read - MinishLab/semble
GitHub - tech4bot/rk3562deb
Contribute to tech4bot/rk3562deb development by creating an account on GitHub.
The Pirated Corpus Was Always a Balance-Sheet Item | Abhishek Shankar's Blog
Anthropic's $1.5 billion settlement is being read as a deterrent. It is much closer to a tariff — a price tag on an arbitrage that produced an asset worth more than the tariff itself, and an arbitrage that is now closed for everyone else. The corpus is gone; t…
Reducing OpenClaw token usage
TL;DR Link to heading I had reduced my OpenClaw heartbeat cadence a few weeks ago expecting a big drop in LLM credit usage. It barely moved the bill. Digging into the per-call usage logs, I found that every persistent session had compactionCount: 0, so each cr…
Don't Outsource the Learning
Right now, it's too easy to let AI write the code while you skip the learning. The bug gets fixed. Your mental model doesn't move. We are silently trading fu...
Towards approachable observability with wide events
A tragedy in four acts
Killing a `Cow` made my JSON formatter 42% faster
Early optimization is bad
Mitigating Ongoing Registry Supply Chain Attacks | Ethan Hawksley
They are only becoming more frequent and more damaging.
Bun's problem may be developing in the open
As soon as people found a Bun branch mentioning an experiment to use an LLM to port the existing Zig code to Rust, they went mad.
![]()
Learning to write (again)
I saw a post on twitter/X that had a screenshot of a prompt that they gave to Claude. The prompt was something like "If you were a human, what is some small ...
Tokenomics: the 62.5-minute rule for Claude's cache | Ryan Skidmore
Is it more efficient to refresh the 5-min cache, let it expire, or just rely compaction?
What Do You Actually Want?
Many companies treat AI as a tooling question. Often they cannot even state clearly what they actually want.
Effective use of HTML with Agents
My thoughts after experimenting with HTML as the output format for AI.
We lost our imagination
We lost our imagination. It’s not because of AI. Not everybody had it before AI, but the people who had it lost it too.
What Matters in Production RAG
Most of us build RAG the same way: follow a tutorial that embeds a handful of PDFs, stores the vectors in a local Chroma instance, and chains everything together with LangChain (if that's still a thing). The demo works. The answer looks reasonable. Then you ta…
The Wild Cyberwest - Trimming Circles
Learning-focused CTFs are Facing a Restructure
As LLMs supercharge CTF automation, the community must adapt to ensure learning incentives remain effective.
AI Fundamentals: World Models for Planning Agents | Michal Pandy
Michal Pandy
Coding on Paper
About three months ago, I bought the Onyx BOOX 25.3” Mira Pro Color, an e-ink monitor for desktop use. I’ve used it as my primary monitor since, and I’ve had a lot of questions about it. This is my experience report, from the perspective of a working, still mo…
Julien Palard - A checkbox to enable the Django debug toolbar
Python Developer and Trainer