M
Military Intensity
14/20Frequency and intensity of hostilities (operations, attacks, clashes).
Partial score: 14/20
CSI
IAW's standard for measuring crisis severity.
In development · refined as it is applied to each conflict
Scroll
Before reading a number
What the CSI measures — and what it does not
What it does
It measures severity
How grave a situation is across five observable axes, on a scale comparable between different conflicts.
It measures change
How that gravity moves over time. A 70 on the way down and a 70 on the way up are not the same 70.
It is published with a date
Every score belongs to a moment and to the evidence available then. It gets revised; it is not a verdict.
What it does not
It does not measure legitimacy
The index does not say who is right. Two parties can produce the same severity for morally opposite reasons.
It does not assign blame
No dimension asks who started it. Responsibility is a different argument, and a number does not settle it.
It does not forecast
Escalation risk describes present conditions, not a prediction. A high score does not mean something will happen.
The CSI is IAW’s own indicator, in development, published only where the evidence supports it. Where it does not, the profile carries no score — and that is information too.
Five dimensions · one score
CSI=++++
Scroll to explore each dimension · tap a letter to jump
M — Military Intensity
01/05
Frequency and intensity of hostilities (operations, attacks, clashes).
What the score is made of
Twenty sub-indicators, five dimensions, one hundred points
Each dimension is built from four weighted sub-indicators. The five columns end at the same height: no dimension outweighs another. That is a choice, not a consequence — and probably the most debatable one in the index.
M
Military Intensity
20Combat Frequency
5 pts
Weaponry Used
5 pts
Battle-Related Deaths
5 pts
Disputed Territory
5 pts
C
Civilian Impact
20Direct Civilian Casualties
7 pts
Forced Displacement
6 pts
Infrastructure Destruction
4 pts
Deliberate Attacks on Civilians
3 pts
E
Escalation Risk
20Rhetoric & Threats
4 pts
Military Mobilization
6 pts
New Actors
5 pts
Diplomatic Failure
5 pts
H
Humanitarian Access
20Access Restriction
6 pts
Food Security (IPC)
5 pts
Health Infrastructure
5 pts
Siege / Blockade
4 pts
I
Internationalization
20External Military Support
6 pts
Sanctions
4 pts
International Presence
5 pts
Media Coverage
5 pts
This table is generated from the same configuration used to score every conflict in the internal system. If a weight changes, it changes here.
100 points possible
Severity spectrum
0–24Low
Limited or sporadic violence. Restricted civilian impact. Contained escalation risk.
25–49Medium
Sustained tension with regular violent episodes. Moderate humanitarian impact. Escalation risk present.
50–74High
High violence or significant humanitarian deterioration. Restrictions on humanitarian access. Elevated escalation risk.
75–100Critical
High lethality, severe humanitarian crisis, or probable escalation. Requires urgent international attention.
Anatomy of a crisis
Anatomy of a score
How a CSI is built up dimension by dimension. The case is hypothetical and the figures illustrative: the real score for each conflict is on its own page.
C
Civilian Impact
16/20Direct harm to civilian population and effects on daily life.
Partial score: 16/20
E
Escalation Risk
11/20Probability of conflict intensification in the short term.
Partial score: 11/20
H
Humanitarian Access
17/20Restrictions and actual capacity for assistance (food, health, safe corridors).
Partial score: 17/20
I
Internationalization
9/20Level of participation or involvement of external actors (military, financial, diplomatic).
Partial score: 9/20
Composite index
67Critical
The CSI sums five dimensions (0–20 each) into a 0–100 total score.
Interactive
CSI simulator
CSI
60Adjust each dimension (0–20) and watch the total index change.
M12/20
C14/20
E11/20
H13/20
I10/20
LIVE · CSI
CSI of Monitored Conflicts
Severity index for the profiles under weekly monitoring. Conflicts we analyse without continuous monitoring are not scored, and their profile says so.
Last update: February 21, 2026
0
Crisis estatal, violencia de pandillas y colapso institucional en Haití
Latin America · Feb 21, 2026
M
12
C
16
E
13
H
16
I
13
0
Insurgencia armada y violencia terrorista en Nigeria
West Africa
Dimension breakdown not published yet
Data chain
Intelligence pipeline
From heterogeneous sources to a comparable score, with the evidence for each dimension recorded.
Open sources and verified media
International body and NGO reporting
Conflict event datasets
IAW scoring
IAW
Output
CSI score with its evidence
Who writes this
The model proposes. The person decides.
IAW uses language models to read and organize large volumes of information — never to decide what is true. No number, briefing or analysis reaches the public without review and sign-off by an analyst.
- 01
AI reads and proposes
Models classify events, extract entities and draft summaries from collected evidence. They always work over verifiable material, never from their own memory.
- 02
Every claim cites its source
Generated briefings cite the specific evidence behind each paragraph. An unsupported paragraph is automatically flagged and cannot be published without correction.
- 03
The analyst decides
All AI-assisted content is born as a draft. A human analyst reviews, corrects and signs it before anything is published.
- 04
Anti-hallucination guardrails
Systems reject figures, names or links the model cannot trace back to source evidence. The discipline is structural, not an editorial promise.
This protocol applies across the pipeline: news ingestion and classification, CSI scoring suggestions, and situation briefing generation.
How far it goes
The evidence, the judgement and the margin of error
01
Where the evidence comes from
Open sources (OSINT), reports from international bodies — UN agencies, ICRC —, NGO and humanitarian agency reporting, reputable media, and event datasets.
No single source determines a score. Information is cross-checked where possible, weighing each origin’s reliability, consistency over time and independence.
IAW does not conduct primary field research. All evidence comes from publicly accessible or institutionally published material.
02
Where a person comes in
The CSI is not an automated index. Indicators are not always comparable across contexts, reporting quality varies, and some dimensions resist simple quantification.
Bias is mitigated through structured criteria, explicit documentation of the rationale and internal review. When a score changes, the change is documented with its justification.
03
What this index cannot do
Information gaps are the rule in conflict zones: events go unreported, or get reported late. Coverage bias affects some regions more than others.
Fast-moving situations outrun verified evidence, so a score can lag the facts. And monitoring infrastructure is not evenly distributed: some contexts are measured better simply because more people are watching.
Scores are revised when new information appears. The CSI is a comparison tool, not a verdict, and it does not replace context-specific analysis.
The methodology may change as new sources appear or outside criticism is taken on. Changes are documented; substantive revisions are announced and explained. If you find an error, write to us.
How to Cite the CSI
If you use the IAW Conflict Severity Index in your work, you can cite it using the following formats:
Standard format
International Affairs Watch (IAW), Conflict Severity Index (CSI), [Conflict Name], [Date], [Score].
Example
International Affairs Watch (IAW), Conflict Severity Index (CSI), <conflict name>, <date accessed>, <score>.
APA 7th
International Affairs Watch. (2026). Conflict Severity Index (CSI): [Conflict]. https://iawatch.org/csi
Institutional or academic use: If you need to cite the CSI in academic publications, institutional reports, or media, you can contact us for additional information about the methodology.