Skip to the content
The Indexsites.llc

Entry IX-07 · News shelf

Newsy

A live reader built from eleven publishers' own feeds, in time order, with nothing choosing for you. We read every wire through the relay and tested its duplicate filter.

Tested 27 Sep 2026, 20:04–20:08 UTC · 2 comparators · labs.llc/newsysite/

The Newsy front page on labs.llc at desktop width: the headline “Beyond the Noise”, a live snapshot panel counting stories and wires, and a row of headlines labelled with their outlets and ages.
labs.llc/newsysite/, captured from the local copy of build 537 on 27 September 2026.

01What it is

Newsy is a live news reader built from publishers' own RSS feeds: eleven wires across five desks, Today, Business, Intl, Middle East and USA. Repeated headlines are dropped, the rest are set newest first, and the page re-reads every 100 seconds. There is no personalisation and no ranking model, and every figure on the page is counted from the stories it holds. The wires are BBC World, The New York Times, NPR, DW, Al Jazeera, WSJ Markets, CNBC, Fortune, BBC Middle East, The Hill and Politico (newsy2.js:47-78).

It is for readers who want the latest headlines from named outlets, in time order, without an algorithm or an account.

02How it works

The engine (newsysite/newsy2.js) first tries a pre-built feed.json, made on a schedule, and falls back to the eleven wires if that file is missing, stale or thin. Each wire goes through the site's own relay, rss-proxy.php, which only fetches hosts drawn from the feed list, re-checks every redirect hop and refuses literal IPs and private ranges, so an allowed host cannot bounce it to a cloud metadata address (rss-proxy.php:11-22). The relay's fresh window is 30 seconds and the page polls every 100, so each poll can bring something new (newsy2.js:24-29, 91).

Headlines are de-duplicated in the browser on a key: lower-cased, outlet suffix, punctuation and short common words removed, cut to 72 characters. Then they are sorted newest first, with undated items last (newsy2.js:288-316). One detail shows the upkeep: the old WSJ address still answered 200, but its newest item was twenty months old, so the page moved to Dow Jones' live host (newsy2.js:56-59).

03The test

Test log, 27 September 2026 (times UTC)
UTCWhat we didWhat came back
20:04:48Page, local 537 copyHTTP 200, 69,887 bytes.
20:04:48Pre-built feed.jsonAbsent from the local tree (an HTML page came back), so this copy falls back to the wires.
20:04:48All 11 wires through the local relay, fresh=1All HTTP 200 in 0.03–0.17 s: 383 items, from Fortune's 10 to DW's 133.
20:05:01The duplicate rule on the Today desk203 headlines from five outlets; 0 dropped as duplicates.

That last row matters. The rule catches near-identical wording only, so the same event told five ways by five outlets stays five items. The relay itself was quick: all eleven wires answered in under a fifth of a second, and DW alone supplied over a third of the 383 items, worth knowing when you read the Today desk.

04Against the field

Each comparator was read on its own site on the date shown. We credit what each does better before saying where Newsy goes further.

Google News

Help article 9005749 fetched 27 Sep 2026, 20:08 UTC; Full Coverage confirmed in help article 9005601 at 20:47 UTC.

Google's aggregator, where, by Google's own account, algorithms decide which stories appear and in what order, partly personalised from your interests and activity.

Where it does better

  • Groups coverage of one event from many outlets, which Newsy's literal filter cannot.
  • Far more publishers and languages, with local news.

Where Newsy goes further

  • No algorithmic selection and no personalisation: the order is publication time (newsy2.js:311-314).
  • Names every feed down to its address.

Ground News

Fetched 27 Sep 2026, 20:08 UTC.

A news platform that merges articles about the same event from over 50,000 sources into single stories, with bias, factuality and ownership information.

Where it does better

  • Real clustering of one story across outlets.
  • Context on each outlet: bias, factuality, ownership.
  • Tens of thousands of sources against eleven.

Where Newsy goes further

  • Straight from publishers' feeds every 100 seconds, with no rating layer between you and the headline.
  • Every number counted live from the stories held (newsy2.js:18-22).

05Where it falls short

  1. Duplicate removal is literal. On 203 Today headlines it found no matches, so one event can appear five times.
  2. Eleven English-language outlets, mostly American and British, and no way to add or drop a source.
  3. It needs the site's relay; where the relay is missing the page cannot read the wires, and says so.
  4. Headlines and summaries only, cut at 260 characters; reading the story means going to the publisher.

06Filing note

File underHeadlines in the order they were published, from outlets you can name.

Newsy keeps a simple promise and does real work to keep it true: a fenced relay, a poll matched to its cache, a dead feed caught and replaced, counts that are counted. Its weak spot is exactly what the big aggregators do well, grouping one story across outlets. Read it for the clock; use Ground News or Google News for the shape of a story.

Open Newsy on labs.llc

07Spec sheet

IX-07 · Newsy · as tested
Addresslabs.llc/newsysite/
Where it runsYour browser plus the labs.llc relay (rss-proxy.php)
Sources11 publisher RSS feeds, named in the code
OrderPublication time, newest first
RefreshEvery 100 s; relay fresh window 30 s
DuplicatesHeadline key, 72 characters
AccountNone
Measured27 Sep 2026, 20:04–20:05 UTC
Set againstGoogle News; Ground News