Helmand ABM — Interactive

Every dot is one person — the model's full population (~5,300, its validated scale), on the real Helmand terrain: the Helmand River, the irrigated Green Zone farmland along the valley, and the towns (Lashkar Gah the largest). Drag the year slider to watch behaviour change; scroll to zoom, drag to pan, click a dot for one person's whole-war history.

Colour by Year — 2014

Selected person

Click any dot to follow one of the model's ~5,300 people — and see the behaviours they moved through across the entire 2006–2014 war.

Helmand in real geography — with the war's real operations. Subtle relief, the Helmand River and Arghandab, the irrigated Green Zone farmland, and towns (Lashkar Gah largest) mark the terrain. The markers are 86 sourced operations the model uses — coalition/US offensives, clearing, Taliban takeovers, and coalition bases arriving (handover) and leaving (withdrawal). Drag the big timeline below (or press play) — events appear and fade as their moment passes. Click any marker to read it.

events network

show all at once

District / village / event

Click a village for its population, control and violence — or an event marker for what happened, when, and the source.

2006

The complexity, measured — not asserted. These four metrics run on the validated province. Together they describe a place whose violence you can't forecast past a week, whose interventions you can't learn from, whose behavioural space keeps expanding, and whose parts are inseparable yet causally untraceable.

M1 · inherent uncertainty

How far ahead can you predict?

~1 week for violence

violence

1 wk

control

31 wk

capability

130 wk

Given perfect knowledge of the present, violence becomes unforecastable almost immediately. Only slow structural state persists — the thing a commander most needs to predict is gone first.

M2 · counter-intuitiveness

Does a coercion surge help?

P = 0.58

coin-flip

0.58

A 12-week surge reduces later violence barely more than a coin-flip — and its effect varies 5.7× more than the average (blue band). Whether it helps is essentially unpredictable: doctrine can't form.

M3 · space of possibilities

How wide is the behavioural repertoire?

24 of 96 behaviours / agent

province

1 village

Behaviours unlocked per person — triple the single-village figure. Coupling and diffusion keep expanding what's thinkable as the war goes on.

M4 · inherent uncertainty

Can any district be understood alone?

85% of district pairs coupled

coupled

85%

Most district pairs move together — none is understandable in isolation. But it's undirected: shared shocks, not traceable contagion. Inseparable, yet you can't say who causes whom.