/dev/null daily

How this page works7 sources readRead every 15 minutes

What this is

Every 15 minutes this page reads the front pages of a few tech news sources and shows what is getting attention. It keeps a story's title, its links and three numbers: points, comments and front-page rank. It never copies article text or images.

Each reading stores those numbers again, so a story's rise and fall can be drawn. Nothing is thrown away: any day since launch can be opened from the list of days. Days are UTC days.

Until 19 November 2026 Dev Null Daily was called Trending Tech News, at /trending-tech-news. Links to the old address still lead here.

How stories are ordered

The hottest story comes first. A story's size is its points and comments against what is usual on its own source's front page, so a small source and a large one share a scale. Smaller sources then count for less, so that their best post does not outrank the day's news: half for Lemmy Technology, Mastodon and GitHub Trending, and a quarter for programming.dev and Tildes. A story on two sources adds both, and size fades with age. The last hour's gain is then added, counted twice, so a story taking off passes bigger ones that have gone quiet.

Posts on different sources are one story when they link to the same article within 30 days, or when their headlines say the same thing within 72 hours.

Rising puts the fastest gainers first, whatever their size. Newest puts the latest arrivals first. A past day is ordered by what each story gained that day, on the same scale, and its rows show the totals when the story was last seen that day. Choosing that day again, or Clear beside the days, goes back to what is on the front pages now.

Topics and trends

Topics come from written rules, applied to a story's title and the site it links to. No model is involved and no article is read. The rules still miss some stories and misread others. When they change, every stored story is tagged again.

The trends page gives each topic's and each linked site's share of front-page stories, by day and by week, and marks it rising, steady or fading against the days or weeks before.

Mutes and privacy

Mutes hide the sources, topics and sites you do not want to see. The stories page also marks what has reached a front page since your last visit.

Both rest on two things kept in this browser's own storage: your mutes and the time of your last visit. Neither is sent to this site or to anyone else. No accounts. No cookies. Clearing this site's data in your browser removes both.

The page can ask this site for a list of stories and no one else for anything, and that request carries nothing of yours.

Sources

SourceHow it is readStatusLast read (UTC)
Hacker News Official Hacker News API Active, read every 15 minutes
Slashdot Slashdot RSS feed Active, read every 30 minutes
Lemmy Technology Lemmy API Active, read every 15 minutes
programming.dev Lemmy API Active, read every 15 minutes
Tildes Tildes group feeds (~tech and ~comp) Active, read every 30 minutes
Mastodon Mastodon API: trending links, technology stories only Active, read every 15 minutes
GitHub Trending GitHub's Trending page: repositories that gained a twentieth of their stars in the last day Active, read every 30 minutes
X X API (paid): news stories, technology stories only Not read: no key or spending limit is set for it never

Lemmy is read as two sources, Technology and Programming, and Tildes as its ~tech and ~comp groups. Their front pages hold a post for days, so only posts from the last day are taken.

Mastodon is not a technology site, and most links that trend there are general news. A link is kept only when its headline or its site is plainly about technology, by rules like the topic rules, or when another source here has posted it. Some stories are missed and a few slip through. Its points are the accounts that shared the link in the last seven days.

GitHub Trending is not a place where people post stories. It is GitHub's list of the repositories gaining the most stars. A repository is taken only when the last day's stars are at least a twentieth of all it has: a new project taking off, not an old favourite. Its points are the last day's stars, and its age runs from its first sighting on the list. Neither it nor Mastodon has a comment count.

X is read through its paid API, for the stories it lists as breaking news on a few technology subjects. Each headline is written by X's own system from the posts, not by a publisher, and can be wrong. Only technology stories are kept, and a story's points are the posts X has gathered under it. X is read only while a key and a monthly spending limit are set, and reading stops when the limit is reached or the balance cannot be checked.

History before launch

Collection began on 17 November 2026. Earlier days were reconstructed afterwards from what three sources still hold about their own past. Each was asked only for a story's id, title, link, points, comment count and posting time.

  • Hacker News, through its search API, which Algolia runs: one request per past day, for the stories that reached 20 points.
  • Lemmy, through its own API: one request a minute, as its servers ask.
  • Tildes, through its group feeds.

A reconstructed day is thinner than a live one, and is marked wherever it appears. It has no Slashdot, Mastodon, GitHub Trending or X, because none of them publishes its past. Its points and comments are today's, not that day's, and a story sits on the day it was posted. On the trends page that stretch of a line is dashed and is not compared with the live part.

Lemmy, Tildes and Mastodon joined on 18 November 2026, and GitHub Trending a day later, so the first live days have fewer sources.

The crawler

Requests come from a bot that identifies itself as DevNullDailyBot/1.0 (+https://launchdayadvisors.com/dev-null-daily/about).

  • It reads a site's robots.txt first, caches it for 24 hours and obeys it, Crawl-delay included. If robots.txt says no, the source is not read.
  • It uses each source's published API or feed, not its web pages, with one exception. GitHub publishes neither for its Trending list, so that one page is read as a web page, at most once every 30 minutes, for each repository's name, description and star counts. GitHub's robots.txt allows it.
  • X is the one place where robots.txt is not consulted, because X is not crawled. Its paid API is called with a key X issued, under X's terms for developers, and none of its pages are fetched.
  • It does not read the articles that stories link to. A site named on these pages is there because stories link to it.
  • It asks Slashdot and Tildes for their feeds at most once every 30 minutes, and each Lemmy server for one page per crawl.
  • It asks a site for its icon once a month, and only if that site's robots.txt allows it. No other images are fetched.
  • When a source returns errors it backs off, doubling the wait each time, up to six hours.

If you run one of these sites and want the bot to stop or behave differently, write to projects@launchdayadvisors.com.

Credits

Inspired by Progscrape, which is published under the MIT licence.