tl;dr Don’t try to shake-down a typography nerd with your dubious, automated claims about his employer using unlicensed fonts. How it started It started with a LinkedIn InMail message (sanitised to protect privacy): Subject: [Urgent] Font Software Licensing Review Hi Ameel I hope you

Sixty former staffers describe an environment of suspicion and intimidation within the nation’s most powerful law enforcement agency.

Discover the reasons companies abandon once-favored technology brands.

The thermostats may no longer connect to Google’s app, but they still transmit your data.

OrthoRoute: GPU-Accelerated PCB Autorouter

AI-powered CLI for background agents. Contribute to rowboatlabs/rowboat development by creating an account on GitHub.

AI’s circular economy may have a reverse Midas at the centre

Reflections on my time leading Mastodon and what's to come next.

Google released Gemini 3 Pro today. Here’s the announcement from Sundar Pichai, Demis Hassabis, and Koray Kavukcuoglu, their developer blog announcement from Logan Kilpatrick, the Gemini 3 Pro Model Card, …

The EU advances its new Chat Control plan as privacy experts warn it could enable backdoor scanning and threaten encrypted messaging.

Do not start baseless theories like this. Nothing points at anything nefarious targeting users here (see below). I personally am just hoping that Catfriend1 is doing fine. Just in case you are reading this and as it bears repeating: Thanks for being part of this community and all the work you put into the app! Now as a user I’d suggest as we German-speakers would say “Abwarten und Tee trinken” (wait and see I guess?), it’s not like the app stopped working or anything. If you wanna get involved,...

National Public Radio will receive approximately $36 million in government funding to operate the nation’s public radio interconnection system under the terms of a court settlement.

Guts is a code generator that converts Golang types to Typescript. Useful for keeping types in sync between the front and backend. - coder/guts

High-performance Rust acceleration for LiteLLM - providing 2-20x performance improvements for token counting, routing, rate limiting, and connection management. - neul-labs/fast-litellm

Editor’s Note (11/2/2025): Due to an error in moving the article over from Google Docs to Substack, the “Balancing CPU and GPU Bandwidth Demands” section was missing some Cyberpunk 2077 data.

Pebble, Rebble, and a Path Forward

LLMs have achieved remarkable breakthroughs in reasoning, insights, and tool use, but chaining these abilities into extended processes at the scale of those routinely executed by humans, organizations, and societies has remained out of reach. The models have a persistent error rate that prevents scale-up: for instance, recent experiments in the Towers of Hanoi benchmark domain showed that the process inevitably becomes derailed after at most a few hundred steps. Thus, although LLM research is often still benchmarked on tasks with relatively few dependent logical steps, there is increasing attention on the ability (or inability) of LLMs to perform long range tasks. This paper describes MAKER, the first system that successfully solves a task with over one million LLM steps with zero errors, and, in principle, scales far beyond this level. The approach relies on an extreme decomposition of a task into subtasks, each of which can be tackled by focused microagents. The high level of modularity resulting from the decomposition allows error correction to be applied at each step through an efficient multi-agent voting scheme. This combination of extreme decomposition and error correction makes scaling possible. Thus, the results suggest that instead of relying on continual improvement of current LLMs, massively decomposed agentic processes (MDAPs) may provide a way to efficiently solve problems at the level of organizations and societies.

My visit to Hetzner Online in Falkenstein goes far beyond a tour of a single server, as the site reveals an impressive combination of industrial precision, technical depth and logistically…

Google Antigravity - Build the new way

For some immigrants, becoming a US citizen takes a few years. For others, limited visa availability and annual country caps mean waiting decades before even applying.

Today we’re releasing Gemini 3 – our most intelligent model that helps you bring any idea to life.