> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gdeltcloud.com/llms.txt
> Use this file to discover all available pages before exploring further.

# magnitude

> How intense an event was, for its own type — on whichever of four ladders fits.

**Range 0–10.** `null` when intensity could not be established — never a fabricated `0`.

## It is not a body count

The log-fatality scale has the deepest academic pedigree here, so it is usually explained first — which
misleads, because it is the path our data uses **least**. In a 59-event sample from our own corpus the coder
chose the conflict path **zero times**: 34 economic, 14 hazard, 11 verbal.

`magnitude` asks one question — *how intense was this, for its own type?* — and answers it on whichever of
four ladders fits. The coder picks the `path` first and records why.

| `path`     | Used when                                                   | Anchored by                                         |
| ---------- | ----------------------------------------------------------- | --------------------------------------------------- |
| `conflict` | People were killed                                          | Deaths — the only closed-form formula               |
| `hazard`   | Physical damage or disruption, no deaths required           | Reach and severity of what stopped working          |
| `verbal`   | A statement, threat or coercive act with no physical effect | How coercive the act is (CAMEO/Goldstein tradition) |
| `economic` | Money, contracts, licences or legal obligations move        | Size and bindingness of what changed                |

## Formula

```
conflict:   magnitude = min(10, 2 + 2 · log₁₀(deaths))
otherwise:  magnitude = value on that path's anchored 0–10 ladder
```

Only the conflict path is arithmetic. The other three are ladders the coder places the event on.

<Warning>
  **Compare magnitude within a domain, not across.** A 7 on the verbal ladder and a 7 on the hazard ladder are
  not the same claim. To rank across domains use `significance`, which is family-scoped for exactly this reason.
</Warning>

## Grounding

Richardson's *Statistics of Deadly Quarrels* (1960) established that conflict severity is read on a
log-fatality scale rather than a linear one. The 0–10 framing follows the Center for Systemic Peace's MEPV
conflict-magnitude scale; the 25 and 1,000-death thresholds come from UCDP/COW. **Our adaptation:** the
anchors are re-derived from our own event distribution, and the three non-conflict ladders are ours — those
sources address armed conflict only.

## Worked example — hazard path

> Nationwide power grid blackout in Havana · INFRASTRUCTURE · 18 articles

```
path  = hazard
value = 7   "A nationwide infrastructure failure left approximately 9.6 million people without
             electricity, halted most public transport, and cancelled tens of thousands of surgeries."

magnitude = 7
```

No deaths were reported, so the conflict formula never applies — and the event is plainly severe. That is
what the other three paths exist for.

## Stability

Measured run-to-run noise is **±0.36**, so the smallest defensible difference is **0.72**. Almost all of that
instability sits on the `verbal` path — see [Limits](/metrics/limits).