How we score

Every story carries a trust score.

Importance tells you how big a story is. Trust tells you how sure we are it's true. Here's exactly how we work it out, with the same numbers our newsroom uses.

Who reported it

Every story starts at 50. If it comes from one of the 29 outlets on our trusted (tier-1) list, it starts at 70 instead. Our desk picks tier-1 outlets for original reporting and a habit of correcting themselves. Source quality matters more than speed.

How many confirm it

Each extra independent outlet reporting the same story adds 8 points, up to 3 outlets (+24). Several pages from the same outlet count once, so three newsrooms saying the same thing beat one loud one.

Does the reader's catch check out

When a reader catches a story first, we read their source, check its credibility and match it against the primary reporting. That check counts for 50% of the final score; the sources count for the rest. If the catch checks out, the reader is credited by name on the post and earns real stock.

What the number means

The final score sits between 10 and 99. We never claim 100: nothing is certain the moment it breaks.

Breaking grade: 70–99; Well sourced. A big story here posts as BREAKING.
Developing: 50–69; Real, but still being confirmed by more outlets.
Weak: 10–49; Thin or doubtful sourcing. We don't push it.

From DEVELOPING to CONFIRMED

A story needs importance 85 or more to reach our X account. Under 70 trust it posts as DEVELOPING. We keep checking it for 48 hours, and when more outlets lift it to 70+ we post a CONFIRMED follow-up quoting the original.

Fact checks are scored too

When we check a claim, the trust score reflects our verdict and how confident we are in it. A bot never posts FALSE or MISLEADING on its own: a person on our desk approves every correction.

True: 50–99; Higher means we're more sure it's true.
Unverified: 30–50; Not enough independent reporting either way.
False: 1–50; Lower means we're more sure it's false.
Misleading: 15–40; Built on something real, framed wrong.

Worked example

A trusted (tier-1) outlet breaks it: 70
2 more outlets confirm it: +16
Score from sources: 86
A reader caught it first and their source checks out: 80
Blend: 50% sources, 50% reader check: 83

83 clears the 70 bar, so a big story like this posts as BREAKING.

What the labels mean

BREAKING: A story with importance 85+ and a trust score of 70 or more. Big and well sourced.
DEVELOPING: Big enough to post, but under 70 trust so far. Usually one or two outlets. Treat it as early reporting and expect updates.
CONFIRMED: A follow-up on a DEVELOPING post once more outlets push it to 70+ within 48 hours. It quotes the original post.
FACT CHECK: Our verdict on a claim going around: TRUE or UNVERIFIED.
CORRECTION: A claim we found FALSE or MISLEADING, with what actually happened.
OUR ERROR: When we get it wrong. The original stays up, marked and corrected.

Corrections

Every fact check, every correction and every mistake of ours is logged in public. When we get a story wrong we post OUR ERROR on X: the verdict, what we said, what is actually true and the score the original carried. The original post stays up and is marked, never quietly deleted. Read the corrections log. Nothing on Paparazzi is financial advice.

Fact checks are scored too

When we check a claim, the trust score reflects our verdict and how confident we are in it. A bot never posts FALSE or MISLEADING on its own: a person on our desk approves every correction.

True: 50–99; Higher means we're more sure it's true.
Unverified: 30–50; Not enough independent reporting either way.
False: 1–50; Lower means we're more sure it's false.
Misleading: 15–40; Built on something real, framed wrong.