Method

How this front page is built

One small program, a fixed list of feeds, and a schedule. No algorithm deciding what you should care about, and nobody’s engagement model in the loop.

The pipeline

1. Every twenty minutes a builder on our own server fetches each feed on the published list over HTTPS, with an honest user-agent that says who we are and where to ask questions.
2. Titles, publish times and snippets are read from the feed itself — RSS pubDate or Atom published — never guessed. HTML tags are stripped, entities decoded, and snippets trimmed to about forty words.
3. Items are deduplicated by link, so one story syndicated by three wires appears once. Stories older than three weeks drop off the front page but keep their page, because a permanent address that 404s is worse than an old one.
4. Pages are ranked strictly by publish time within each section. There is no promotion, no sponsored slot and no personalisation: two visitors see the same page.
5. The pages are plain files served by nginx. There is no database behind a page, no JavaScript required to read a headline, and no request to any other domain.

What a feed is not

A feed tells us a publisher believes an item is worth announcing, which is a real signal and a limited one. Outlets change their feeds, sometimes drop them, and a few block robots by default. When a feed fails we keep the last good copy rather than empty a page, and the status page shows which sources were fetched in the last run and which did not answer.

Money

The only commercial links on this site go to bananacoconut.com, our own store, and they carry tags that tell us which of these pages sent the click. We do not take commission on the news we link to, so no outlet has an incentive to be featured, and none is punished for being inconvenient.

If you want a feed added or dropped, ask us at hello@adaylabs.com and we will test it and say honestly what happened.