Project / Language & news / Data
Mundane Read
News with the drama taken out and measured, with a reader site, a browser extension and an API.
NowLive
The question
Can the news be told without the drama, and can you check that it was? Mundane Read began with a school report: my teacher docked marks for removable adjectives, overly descriptive words and personal opinions, and the news has plenty of all three.
What it does
It reads reporting from 27 news feeds across 17 mastheads. From each article it takes out the words that grade and dramatise, keeps every fact, number and quotation, and publishes the plainer version with the measurements beside it.
How it measures
Every article is scored on seven indicators, and the Niral Score folds those readings into one figure from 0 to 100: lower is plainer. The words behind each number are kept with it, so a score can be checked rather than taken on trust.
Finding the words to take out starts with a part-of-speech tagger trained in-house. On news it finds adjectives at F1 0.80, where matching word endings scores 0.56 on the same text.
One story, three versions: the Quest apartments data breach scored 9.9 as 7NEWS filed it, 7.8 at ABC News and 7.6 in Mundane Read, as published on mundaneread.com in September 2026.
What it will not do
It never publishes a version that reads worse than the report it came from; where the plainer version would not score better, it makes no change to that article. It calls no hosted AI model, and it runs on no third-party packages.
Where to read it
At mundaneread.com (opens in a new tab), in its browser extension, and through its API. The Method page (opens in a new tab) sets out the measurements in full.