Method / Indicator
Language Load
How loaded a report’s writing is, every reading in one number, 0 to 100. Lower is plainer. It reads the writing, not the events.
0 plain loaded 100
What every band means is on the Calm Index. The words beside the number (“plain”, “heavy load”) are about how the story is told, never about what happened.
What the numbers usually look like
Most reports are written fairly plainly, so most score under 20. Of 484 full-length articles published between July and September 2026, scored the way the site scores today, half scored between 9 and 16, and nine in ten between 7 and 21.
The differences are in a few kinds of word: verdicts and loaded verbs, drama in the telling, and claims nobody puts their name to. A report at 20 carries noticeably more of those than one at 8. A difference of a point or two says very little, which is why the Calm Index calls outlets within a point of each other too close to call.
What raises it
- Judgement words and verdicts ("slammed", "regime"), which weigh most
- Drama in the telling ("horror", "carnage")
- Adjectives that describe rather than report
- Coded political vocabulary
- Claims with nobody’s name on them
What is never counted
- Anything inside a quotation
- How serious or sad the events are
- Whether the article is correct
When it appears beside a story
Shown beside every story we score.
Plainest in the last week
- Alleged drug boss Arash Ebrahimi Far detained in Colombia Language Load 3
- Glasner urges quick end to Manchester City saga and fears damage to Premier League Language Load 3.6
- P-plater in custody as witness recounts moment BMW hit Knights supporters Language Load 3.6
Most loaded in the last week
- Anthropic says its IPO could herald the end of the world as we know it Language Load 28.9
- Was pre-COP a flop? Questions loom over the Australian government Language Load 26.9
- Lunar Grounding Challenge Language Load 26.2
Every reading describes how the article is written, not whether it is correct. The full method, including how Language Load is built, is on the Method page. Anything here can be checked against the API, which returns the same numbers with the words that produced them.
The other indicators
AdjectivesRepeated wordsSentimentPolitical dictionGenerationalGenderJingoisticReligiousSourcingMood