How Huduku keeps balance
The rules that decide which headlines the rolling edition shows, and in what order; why they are what they are; and how they are working today. The same rules for every outlet and every reader. Nothing here is a judgement of an outlet.
The data
Huduku shows headlines only from outlets it has chosen by hand and approved for its public pages; the list of outlets says which, and each outlet’s page says why it is there. It reads their own feeds, never Google News.
News feeds are fetched every hour. The other public feeds (fact checks, opinion desks, longer reads) get an hourly run of their own, so they don’t wait behind the private feeds of Huduku’s research workbench. A headline dated in the future is given the time it was fetched.
Who owns each outlet comes from published facts, each with its source, on the outlet’s page: company filings, rating reports, the outlet’s own pages. An owner is the group behind an outlet, or else its publisher.
The rules
Each rule in one sentence, with its numbers, as the code applies them.
- Owners, not outlets
- Sister outlets, and an outlet’s YouTube channel, share one owner’s room.
- One turn each
- A section takes its cards in rounds, highest weight first: every owner present gets its best card in before any owner gets a second.
- Limits per section
- At most 2 cards from one owner in a section and 3 in the lead section (India in English, Karnataka in Kannada); at most 8 cards in a section and 12 in the lead.
- The weight
- A card’s weight is the number of owners that carried its story, divided by 1 + (hours since its newest report ÷ 8). A story quiet for 8 hours counts half; a single report counts as one owner.
- A soft ceiling
- Once an owner has 10 cards on the page, it waits until every other owner has had its turns; then its cards may fill what is left, so no section goes thin.
- Shown by whoever has room
- A story several outlets carried is shown by the first outlet to report it whose owner still has room. Compare coverage still lists every outlet, first to report first.
- The order in a section
- Highest weight first. Among neighbouring cards whose stories the same number of owners carried (4 or more count alike), the owner with fewer cards higher on the page goes first, then the newest.
- Top stories
- Up to 8 stories, weighted the same way, each carried by at least 3 owners in English and 2 in Kannada. Their order isn’t changed for balance, only who shows each one: the first to report it among the owners with the fewest cards so far.
- Latest
- In each time band, at most 3 cards from one owner and 30 in all, one turn each; an owner with 6 cards in the view waits for the others.
- Topics and search
- Newest first. A topic page lists at most 6 cards from one owner, and search 8 among its first 100; the rest are behind a “More from …” line at the end.
- “N reports”
- Counts outlets: an outlet’s video and its article on one story are one report.
- Nothing else
- No outlet is lifted or held back by hand, nothing depends on who is reading, and nothing is ranked by clicks.
Why these rules
Posting more doesn’t buy more room. Some outlets publish dozens of headlines an hour. On a list sorted by newest they would fill every section. Audits of other news aggregators found just that: a handful of outlets taking most of the room, helped by freshness (the research).
Owners, not brands. When one company runs several outlets, a story carried by all of them is one company’s coverage, not several confirmations. So the rules count owners, and each outlet’s page gives the source for its owner.
Equal turns, not room in proportion to output. This is a choice, and researchers disagree about it: some argue that exposure should follow merit. Huduku’s view is that output volume isn’t merit. Within each turn it still orders by how many owners carried a story, which isn’t volume either.
What it costs. When one owner breaks several big stories, its second-best can sit below another owner’s weaker one, and a small owner with one weak item still gets a full turn. Equal turns balance sources, not views: two owners can say the same thing.
The yardstick. Huduku isn’t bound by the EU’s Digital Services Act or the Council of Europe’s guidance, but uses them as a test. The Act asks platforms to explain the main parameters of their ranking in plain words (DSA Observatory); the Council of Europe, writing about content given prominence in the public interest, asks for criteria that are clear, non-discriminatory, viewpoint neutral, transparent and objectively justifiable (Council of Europe, 2021). This page follows the four layers Diakopoulos and Koliska suggest for explaining a news algorithm: its data, its rules, how it decides, and what it shows (Diakopoulos and Koliska, 2017).
How it’s working now
Counted from the rolling edition as this page was made, the way Huduku’s balance check counts the live pages every day. A card counts for the owner of the outlet that shows it. A top slot is one of the first three Top stories, or the first card of a section. The effective number of owners is how many owners with equal shares would make the page as concentrated; weighted by position, a card counts 1 ÷ log₂(1 + its place in its section), the discount the research uses for how far down a list readers look.
Huduku’s own line is 15% of a page’s cards for one owner: past it, the daily check fails. No standard line exists for a news page; this one is Huduku’s choice.
| Measure | English edition | Kannada edition |
|---|---|---|
| Cards on the page | 99 | 90 |
| Owners | 19 | 14 |
| Biggest owner | India Today Group 11 of 99 (11%) | Asianet News Network 12 of 90 (13%) |
| Top five owners’ share | 54% | 59% |
| Top slots | 7 owners in 14; at most 3 for one | 8 owners in 14; at most 3 for one |
| Effective number of owners | 12.3 | 10.6 |
| Biggest owner’s share, weighted by position | 13% | 14% |
| Effective number of owners, weighted by position | 11.3 | 10.4 |
Counted at 10:25 am.
What isn’t solved
Feeds that land in one section. Some outlets’ feeds name no desk in their links, or only one, so most of their headlines compete for one section’s places. In October 2026 these included The Print, News18 and Onmanorama in English, and Public TV, Sanjevani and Hosadigantha in Kannada.
Late feeds. The Print’s feed listed items one to 24 hours after they were published (October 2026), so its headlines arrive late and count for less. That is the publisher’s feed; one turn each softens it, and nothing here fixes it.
Ownership is only as good as its sources. Kannada Prabha shows the limit. A May 2026 rating report lists its company as a subsidiary of Asianet News Network, which also runs Asianet Suvarna News, while its website says a company on the other shareholder’s side runs the site. Huduku counts it with Asianet Suvarna News and says so on its page.
Top stories keep their order. Who shows each one is balanced; which stories lead, and in what order, follow how many owners carried them.
The order rule reaches only neighbours. It reorders cards carried by the same number of owners that sit next to each other. A fresh single report next to an older, widely carried story stays where its weight puts it.
Busy feeds. The fetcher takes at most 50 new items from a feed in one run, so the busiest feeds can lose some headlines in a busy hour.
Further reading
The research behind it: what published audits and studies found, and how these rules compare.
Huduku’s own choices, in the reading guide.