The complexity lives underneath so the surface can stay simple.

RESON is built with semantic HTML, modern CSS, vanilla JavaScript, and a small Python standard-library compiler. There is no frontend framework, package manager, database, analytics service, or client-side router.

Writing and project case studies live in portable Markdown. A piece needs three things — a title, a date, and what it is about — and the build works out the rest: its address, its metadata, which interests it belongs to, what to suggest reading next, and where it appears across the site. One file can sit in several places without ever being copied.

Photographs are referenced by filename and sized from their own proportions, so a landscape photograph is given room and an upright one is not stretched to fill it. Real dimensions are read from each file, which is why nothing shifts around while a page loads.

Inter Variable carries the interface. Newsreader Variable is reserved for long-form reading. Both are self-hosted as licensed WOFF2 files, and the browser downloads Newsreader only when an article needs it.

GitHub preserves the source and its history. Cloudflare Pages delivers the generated static site.

The site is designed around progressive enhancement, keyboard access, readable type, one warm light editorial identity, and useful HTML before JavaScript runs.

Source

The source repository is available for anyone interested in how the site is put together.