People are quietly withdrawing from active participation in social media. While they may not be deleting accounts en masse, many are:

Tourism transformed Mawlynnong's fortunes. But after decades of welcoming visitors, residents decided one day a week should belong to themselves.

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. Like many graphics programmers, I read Sebastian’s blog post and really enjoyed it - I was inspired by it, and decided to see how close to the API he describes I could get today, layered on top of existing platform APIs. The answer turns out to be “pretty darn close”. The result is a project I’m calling Loon GPU, and I’ve put it up on Github. While the library is still rough, poorly tested and surely filled with bugs, it is usable and I wanted to share it early in case other folks were interested. Currently it has a Vulkan 1.3 and Metal 4 backend, and here I want to do a high-level walkthrough to see how the API design maps on to those backends.

Organizations rolling out agentic command line tools like Anthropic's Claude Code and GitHub's Copilot CLI need to know who will try them, who will keep using them, and whether the tools produce enough output to justify their cost. At organizational scale, token spend can run into millions of dollars annually, so misreading adoption, retention, or impact can make a rollout expensive without changing engineering velocity. Studying tens of thousands of engineers at Microsoft over its early-2026 rollout, we find that first use spread primarily through social networks, retention was associated more with engineers' coding activity than with demographics, and adopters merged roughly 24% more pull requests than they would have otherwise. We use merged pull requests as our proxy for output -- acknowledging that a merged PR is not the same as the value it delivers -- and the lift persists across our four-month window. These results suggest that CLI coding agents are neither uniformly adopted nor mere novelty effects and that organizations should treat visible peer use as central to rollout strategy.

From coders to bikers to music lovers, the new enthusiasts of an ancient language are giving it a generational brio

How we built a pipeline to get exact gradients out of a Fortran thermal solver, then used it to solve real inverse problems from Python.

An interactive, sourced breakdown of what the SFFA v. Harvard case revealed about elite college admissions: hooks, price, demographics, and merit.

sx 2.0 is out: a native app for Mac, Windows, and Linux that lets anyone share AI skills with their team, no git and no terminal required.

OpenAI mandates hardware-backed passkeys for Trusted Access Cyber (TAC) members to log into ChatGPT accounts

A unified, unmanaged multimodal runtime. Zero-allocation native AI inference for Java 22+. - ProjectArgus-cc/libargus.cc

Previously: Kan Extensions in Double Categories. In programming, actegories play a central role in optics: lenses, prisms, traversals, etc. To understand actegories, let’s start with the defi…

From a small 1D array of floats to a proper autograd capable of training tinyML models, explaining every step.

dumb human. GitHub Gist: instantly share code, notes, and snippets.

Rejourney is a open source, self-hostable/hosted observability tool for web and mobile apps. Focus on lightweight and performance. - rejourneyco/rejourney

Contents Tolstoyan art Zero-knowledge proof A marriage of (or casual relationship between) the two Conclusion In this post I hope to prove to you (get it 😉) in rather uncertain terms, that:
Tolstoyan art has the structure of a zero knowledge proof.
Currently, this statement is a big nothing-burger, so first we will discuss the high level ideas of Tolstoyan art and zero knowledge proofs.
Tolstoyan art In Tolstoy’s 1897 book, what is art? 1, he discusses what it means for some piece of work to be art. In chapter V, he states that the activity of art is

Observability for AI-Native teams. Logs, metrics, traces, APM and LLMOps — drop-in replacement for Datadog, Langfuse, and Grafana at 1/5th the cost.

AI debugging for modern engineering teams. FixBugs is an AI agent that triages alerts, detects bugs, and resolves them automatically. From root cause analysis to validated fix, end-to-end.

IBM struggled over the past three months as customers scrambled to get their hands on components that AI datacenters are gobbling up, its CEO said.

A Jekyll plugin for fast server-side cached LaTeX rendering, with support for macros. Enjoy the comfort of LaTeX and Markdown without cluttering your site with bloated JavaScript. - yagarea/jektex

The women who served as munitions workers during the First World War endured dismal conditions and a plethora of dangers: illness, injury, and death. Why are they barely remembered?