How to track global conflicts

The source types, and what each is good for

Why one source is never enough

Every source type above fails in a characteristic way. Wires are slow and cautious; local outlets are fast and partial; OSINT is precise and selective; maps are broad and unverified; sensors are honest and blind. Triangulation is the habit of reading a claim across types so their failures do not overlap. A strike reported by a local channel, placed on a map, matched by a fire detection at the same hour and picked up by a wire the next morning is well supported. The same strike reported only by the channel is a lead.

The principle is independence, not count. Ten Telegram channels reposting one video are one source. A video and a satellite hotspot are two.

Verification basics

1. 01

   Provenance

   Find the earliest version of the claim. Who posted it first, and what is their record? A repost has the credibility of its origin, not of the reposter.
2. 02

   Timestamp

   Does the publication time fit the claimed event? Is the footage's lighting, weather and season consistent with the date? Old footage resurfaces after every major incident.
3. 03

   Location

   Can the place be confirmed from landmarks, road layouts, terrain or shadows against reference imagery? Geolocation communities publish their reasoning; use it.
4. 04

   Corroboration

   Does an independent source of a different kind agree — a wire, a fire detection, a transponder track, later imagery? One is good; two of different kinds is strong.
5. 05

   Reverse search

   Where a still or clip is doing the work, search for earlier appearances of the same image. A match to an earlier date ends the claim.
6. 06

   Label it

   Confirmed, reported, observed, inferred. Keep the label attached to the claim as it travels, because it will be stripped by whoever repeats it.

Common mistakes

• Treating virality as confirmation. Reach measures interest, not truth.
• Ignoring timestamps. The single most common error: last year’s footage as today’s event.
• Relying on one account. However good its record, one account is one point of failure.
• Reading map markers as evidence. A marker is a report, placed by a source, with the source’s precision.
• Reading silence as calm. No reports from a region may mean no access, no coverage or no connectivity.
• Reading intent from activity. Flights, movements and statements may indicate something; they do not confirm plans.

Tool categories for the work

The tools follow the source types: an aggregator or live map to find what is being reported; unfiltered ADS-B and AIS trackers for transponder signals; Copernicus Sentinel and NASA FIRMS for imagery and fire detection; geolocation communities and OSINT publishers for verified footage; the registers — IMF, World Bank, sanctions lists — for the country background. Two category pages on this site go through them individually: live conflict maps and OSINT tools for conflict monitoring.

How WarWatch simplifies the workflow

WarWatch does the collecting and the first sort. It reads 45 news and OSINT sources continuously, keeps the link on every item, classifies each for type and severity, places the high-severity subset on a globe, scores every country on its volume of conflict reporting with a daily trend, and attaches a dossier per country with figures fetched from the registers. Alert rules watch for change so you do not have to; a daily brief summarises the week’s reporting for the most active countries.

What it does not do is verify. Classification decides what is shown, not whether it is true; the source link is kept so the verification steps above are one click away. Read it as an index and a context layer, and keep the judgement.

Limitations

• Coverage follows the sources; under-reported theatres are under-represented.
• Automated classification errs in both directions.
• Severity measures reporting volume, not intensity, and the map says so.
• Model-written briefs are summaries of the platform’s own reporting, labelled as such.

Frequently asked questions

What is the best way to track global conflicts?

Combine source types and compare them. Read at least one wire service and one regional outlet per theatre, follow a few OSINT researchers with a record of showing their working, use a live map as an index rather than as evidence, and check important claims against a physical signal such as fire detections, transponder tracks or satellite imagery. A monitoring platform can do the collecting and ranking; the judgement stays with you.

How do analysts verify conflict reports?

By provenance, time, place and corroboration: who first published it and what is their record; when was it published and does the timestamp fit the claimed event; can the location be confirmed from landmarks, shadows or terrain; and does an independent source, ideally of a different kind, say the same thing. A report that passes all four is well supported; one that passes none is a lead.

Are live conflict maps reliable?

They are reliable as indexes of reporting and unreliable as evidence. A marker means a source reported something at roughly that place; it does not mean the event was confirmed, and the position is often a town rather than a coordinate. Use the map to find the report, then verify the report.

Which sources should I avoid?

None outright, but weight them. Accounts that never show their working, channels aligned with a party to the conflict, and anything that cannot be traced to an original post deserve the least weight. State media is a source for what a state says, not for what happened.

How much time does this take?

Collecting takes most of it, which is why aggregators exist. With a platform reading the feeds, ranking countries and keeping source links, the remaining work is reading the reports that matter and checking the ones you will act on.

Related resources