← All posts

A thousand and one topics a day

You no longer have to ban the news: the agenda can be drowned in sheer volume of information, without a single lie.

Russian outlets lead on the number of news items per day: in digesta they routinely put out about 2.5–3 times more news a day than Western ones, and twice the world average. But that volume is produced by a few "giants" (state agencies and wires), not by "the whole market". Let us look at one striking example.

On 3 September 2026, TASS put out 1,456 items — roughly one a minute, around the clock. They spread across 990 different stories.

1 item822

2103

326

418

57

6–2413

1161

822 stories got exactly one item. And one got a hundred and sixteen: "Putin at the Eastern Economic Forum: no mobilisation, a chance for peace with Ukraine", a story that ended up gathering 68 outlets. Eight percent of the agency's daily output went into a single story.

So this table is about a mechanism talked about less often than censorship, though it is more visible.

01A pointless ranking

The first thing you want to do with data like this is sort outlets by volume. We did, and here is the top of the list (median records a day over 30 days):

This ranking has nothing to do with manipulation. An agency legally publishes fifteen hundred items a day, a weekly publishes five. Comparing them by volume is like comparing a truck with a bicycle. So what you have to measure is not the size of the output but its shape. We tried four measures of shape. Three of them did not work, and that too is a result.

02Items spread in an even layer

The first two attempts failed the same way. "Agenda entropy" (how evenly the output is smeared across stories) turned out to be almost an exact copy of volume — it measured not shape but the same old size. "Flatness" did not distinguish anything at all: for almost everyone it hits the ceiling, because most stories get exactly one record from a source. The figures and formulas are in the "Method".

The third measure gave an unexpected twist. "Stories per record" points the other way. The measure is genuinely spread out — from 0.41 to 1.00 — but at the bottom of the scale stand Astra (0.64), Interfax (0.69) and TASS (0.71), that is, outlets that return to a story and file a second, third, tenth note. And at the top, exactly at 1.00, are N+1, Naked Science, StopGame: thin thematic wires where every record is about its own thing. The hypothesis "volume blurs the agenda" would have predicted the opposite arrangement.

03The share of "noise"

The only measure independent of volume (its correlation is near zero, −0.09) is the share of the output that goes into stories that did not gather enough coverage. The threshold is set at the point where a cluster forms: a story reaches the feed if at least ten outlets from at least two countries wrote about it. Everything below that we count as "noise".

The median outlet spends 68% of its output on such noise. The spread is enormous: from 23% to 100%.

And here a caveat is needed, without which the figure lies. It is half about us, not about the outlet.

A hundred-percent noise for The Korea Times or ScienceDaily means not "the outlet drowns the agenda" but "we carry almost none of its neighbours". This quantity can only be read inside a dense country or language audience, and only relative to its own peers.

04What can be seen after all

822 stories a day, each of them mentioned exactly once. This is not repetition, nor propaganda in the usual sense. It is a smooth, continuous flow of items about everything at once, in which each individual item is true, yet together they add up to nothing you can remember.

Half of the entire flow — 48.3% of records — lives in stories that one or two outlets picked up. A significant part of this is not "junk" but diversity: local, niche, original stories that no one else has (for The Hindu or the Sydney Morning Herald such stories are the majority, and across the whole pool the singletons are distributed widely rather than pouring from one source). But for the same reason there is nothing to check them against: a single-source story is not independently confirmed — and in the flow you cannot tell a unique angle from something simply no one re-checked.

This does not prove anyone's strategy, but it shows the shape: there is now so much news that the question "what was left unsaid" has given way to the question "what of this can you even hold in your head" — and the second question is protected from exposure, because formally not a single untrue word was said.

05And now across all countries

We ran the same across the whole pool — how much of each country's output goes into single-source stories and how much reaches the feed at all (30 days):

Russia181k

Serbia93k

Brazil75k

UK51k

France50k

US40k

• Blurring is universal. ~70–82% of the output in any country falls short of the feed threshold — that is the structure of the media market, not a trait of one newsroom.
• Russia leads on volume, not on "noisiness". Russia alone produces twice as many records as the US and the UK combined (181k against 91k a month).
• The naive guess did not hold. The share of single-source stories for Russia is even lower than the West's (26% against 39–47%) — but that is not diversity, it is reprinting: one agency wire duplicated by dozens of storefronts.

06How not to drown in volume

Blurring is not one country's trick but the shape of the market: three quarters of everything published reaches nowhere, and it is like that everywhere. Russia has its own role within that shape — not "noisier" but larger and more uniform: Russia alone produces twice as many items as the US and the UK combined, and its stories are less often single-source not out of diversity but out of reprinting — dozens of storefronts duplicate one agency wire. A hundred storefronts, one voice.

Hence a simple conclusion: "read more" does not work against volume. The opposite works — not to discard the long tail, but to mark the level of confirmation: where a story was confirmed by ten independent outlets, and where it is a single voice. That is what digesta does.

An open question — dynamics. Our data show the shape of the flow, but not its behaviour over time: whether a media space floods the feed with the irrelevant precisely when something inconvenient happens. That is a separate test on a long baseline; we began taking daily snapshots on 04.09.2026 and will return to it once the baseline accumulates.

---