49 links from the last 24 hours · 20 October 2026
Protecting Firefox users following the compromise of eMudhra intermediate certificates – Mozilla Security Blog
On Saturday, October 17, at 20:06 UTC, Mozilla received a report that private keys associated with eight eMudhra intermediate certificates had been compromised. These keys could allow attackers to issue ...
Matyáš Racek's blog
The Netherlands is building its own Linux OS to break free from Microsoft
The Dutch government is trying to solve its digital sovereignty problem with its own Linux operating system. A local source recently highlighted the development efforts being poured...
A high-performance and high-level purely functional data-parallel array programming language that can execute on the GPU and CPU.
Jim Calabro
GitHub - deepseek-ai/DeepInfer: DeepInfer: High-performance, large-scale inference framework for DeepSeek models
DeepInfer: High-performance, large-scale inference framework for DeepSeek models - deepseek-ai/DeepInfer
OpenBSD 8.0
WhatsApp's Hidden Notes Told AI Not to Recommend a Regulator
WhatsApp removed hidden notes telling AI not to recommend a regulator, and not to cite GDPR or CCPA once a lawyer came up. We quote both notes.
Why do we need ud2 when int 6 is right there? - The Old New Thing
An interrupt is not a fault.
Laypeople support technical fields for technology, not art
Some people think that LLMs should not be used in technical fields like programming or mathematics, because those fields are actually art. True or not, this…
Jeremy Scheff poignantly captures the key loss caused by coding agents for career programmers who enjoyed the puzzle of developing code: Programming turned into one of the greatest jobs you …
Ranking a Backlog with TypeSafe's Jev: One Rubric, Two Backlogs, Kendall's Tau
An introduction to TypeSafe's Jev, a model that answers typed questions with probabilities, through a small tool that ranks a GitHub backlog with it and checks the result against a blind ranking.
Isaac Correa's manifesto, October 2026: anyone can now build, copy or replace software, so code no longer protects a company. What is left is well-structured data, complex solutions and relationships.
Who lives without a car in the Bay Area
~1 in 10 Bay Area households has no car. The biggest group is seniors, half earn under $50k, and outside San Francisco the car-free population is getting richer.
Always-on AI agents are distributed systems. Follow one Grok Bot message from send to sleep: queues, background workers, durable workflows with Temporal, idempotency, Firecracker VMs, prompt caching, and context compaction.
TLDR: I trained an LLM reasoning agent to play chess, with model sizes 0.8B, 2B and 9B. The agent calls tools which help it keep track of the board state, which enables good performance with a relatively small number of parameters. The demo code is available o…
Shipping a LaTeX compiler behind an edge app · bhargav.dev
Your app runs at the edge and cannot make a PDF. Here is the whole path from that problem to a running container, including the failed deploys, and why the...
Someone asked me, given that LLMs exist and likely keep improving, whether I thought Ruby on Rails still made sense? The implication being of course that wri...
Why I’m no longer updating this repository
I’ve decided to archive the source code of my Ruby gems to make it clear that I’m no longer working on them; I’ve written this post to link from them in order to explain why.
We live in the greatest of times to build software. A $200 subscription gives you access to state-of-the-art coding models that can build almost anything. So why isn't our software 10x better? And why does most of our software still look like it was designed i…
My Neighbor, Me, and a Grand Principle - The Last Word On Nothing
The other week I was trudging through laps around the court I live on, and a neighbor yelled across the court, first “HI” and I yelled “HI” right back and then he yelled, “THIS IS THE BEST DAY OF THE YEAR!” So I asked why it was the best day of the year, and t…
*in machine learning.
Where Jev Fits in a Large-Scale Data Pipeline - Channel3 Blog
Jev raised our product categorization accuracy from 81.9% to 97.2% at about the same cost, when we used it to choose from 30 prefiltered categories. Searching the full taxonomy with it was too expensive.
Getting my TOTP data out of Twilio's Authy app thanks to the GDPR
The personal blog of Thomas Steiner
Tests don't prove anything, but they are a model
Cloud AI assistants keep a server-side copy of everything you tell them. Here is what the policies actually say, and why open-weight models running on your own device change the equation.
Why Hanami splits its database layer into relations, repos, and structs, and how that split keeps your persistence layer from leaking into the rest of your app.
Why Even Write Code in a Professional Context?
Why Even Write Code in a Professional Context?
Agentic Monitoring
Agents are structurally well suited to monitoring production systems.
all vulnerabilities are now race conditions
TLDR AI made finding bugs cheap. It did nothing to make getting paid for them any easier. The same bug now gets found by ten different people inside a fortnight. One of them collects the money. The other nine collect a polite duplicate notice and a dent in the…
Chessbook-1, Maia-3 and Allie 2.0 at a glanceCapability scorecard: a 0-100 grade per measurement, and whether the model has the capability at all. Predicting human moves Matching strength Time scramble play Predicting think time Predicting who wins Chessbook-1…
Muting the built-in speakers in non-2026+ Q-Symphony TVs
DrupalCon Rotterdam 2026 DriesNote presentation.
The advantages of mobile Linux over AOSP
If our goal is the empowerment and greater freedom of people, mobile Linux appears to be a more feasible route to achieve those goals in the long term than AOSP.
Banana Curving Machine
What is a Banana Curving Machine? Well it's a fantastic apparatus that does what it says. Curves Bananas.
We nearly bought our own GPU servers. Seven vendor quotes later, the math said $354k of hardware would take until year nine to beat renting, so we kept renting. Here are the real numbers, the proposals, what changed the answer, and how to decide for your own t…
What Shall I Teach My Son?
What is the purpose of education in the age of AGI?
lilly is a terminal text editor written in V, built as an alternative to Neovim. We timed it against Neovim, Helix, Sublime Text and Zed on the same machine. These are the numbers, including the ones where lilly comes…
Code is clay now
Artisans, factories and the new material software is made of
We Need a New Model for Employment
AI works 24/7, so the 40-hour week no longer measures output. Why full-time employment is changing, and a new model that pays people for results, IP and value created.
Today, we officially roll out MiniMax M3.1 Flash. As a next-generation Flash model, M3.1 Flash delivers a generational leap in agentic tasks, coding, Computer Use and cybersecurity, while making online inference nearly three times faster and reducing usage cos…
After years of Logseq, a failed migration pushed me to Obsidian. Here’s how I moved my notes and rebuilt my daily workflow with a few plugins.
It always warms my heart to read kind blog posts with interesting ideas for making Bubbles even better. Of course I can't just keep adding features: Bubbles is already packed with stuff, and I want to keep it minimalistic and intuitive.
Arch Linux ARM on the ASUS ProArt P16, with a bootable installer, published Windows comparisons, and local Qwen inference.
Was the HuggingFace hacking incident scheming?
In July 2026, internal OpenAI agents hacked HuggingFace (HF) in order to figure out how their evaluation grader works. This involved a thousand agents coordinating on an improvised message board to…
Jim Calabro
Why “don’t bring me problems” causes problems (and how to face them together)
I recently watched a video of a tech CEO explaining why he doesn’t do one-to-ones with his direct reports. His reasoning was that he doesn’t want employees bringing him subjects he’s not inte…
A few days ago I wrote about reporting a bug to Google Disability Support. It's a bug that kills touch exploration for every screen reader on Android until you reboot the phone. In return I was asked to turn TalkBack off and on again, clear the cache of an app…