
Inference Without a Round Trip
WebGPU shipped, quantization matured, and small-but-useful models exist. The browser is an LLM runtime now, and the consequences are bigger than the novelty.
WRITING // 2026
More tech heavy stuff on software architecture, systems and technologies.

WebGPU shipped, quantization matured, and small-but-useful models exist. The browser is an LLM runtime now, and the consequences are bigger than the novelty.

A 2017 in-place optimization in the kernel's crypto subsystem turned out to be a 4-byte page-cache write primitive. Disclosed last week, exploitable from any unprivileged shell, fixed by a single mainline commit.

Why most websites running WordPress don't need it, how Hugo wins on performance, cost, security, and complexity, and why LLMs just closed the last gap that was keeping WordPress in play.

Mitchell Hashimoto's GPU-accelerated terminal emulator ships its biggest release yet. Six months of work, 2,858 commits, and the most-requested missing feature from 1.0 finally arrives.

If you've ever started a TypeScript project with npx create-next-app and wondered why Go doesn't have an equivalent, go-blueprint is your answer. One command, production-ready structure, no setup overhead.

How Tauri, Rust, and React combine to deliver a native desktop writing experience with no cloud dependencies, no accounts, and complete data ownership.
How CodeMirror 6's extension system powers distraction-free writing modes, wiki-link previews, real-time word counts, and typewriter scroll — without fighting the framework.
Why Loomdraft stores everything in plain files, how SQLite FTS5 delivers instant full-text search without a server, and what 'local-first' actually costs in practice.

Cheng Lou's new 15KB library rewrites the rules of text measurement. Pure arithmetic, no DOM thrashing, and it's already going viral.