When the Protector Becomes the Threat

Changing Face of War

When we think of war, we imagine armies fighting armies. But the data tells a different story.

Over the last two decades, the dominant forms of conflict-related deaths are increasingly internal. These represent fatalities from Intrastate conflicts (civil wars like those in Syria and Afghanistan). While interstate wars remain present, the largest shares often come from intrastate conflicts. But to understand the true shadow of modern conflict, look at the red line.

That continuous red layer is One-sided violence—instances where armed forces or governments target unarmed civilians. While civil wars eventually peak and fall, this red line remains a steady, continuous source of casualties year after year.

How to read

Each layer shows the number of recorded deaths by conflict type per year. The total height is the overall burden; the composition shows what kind of violence dominates.

Where Civilians Face Highest Risk

Mapping one-sided violence shifts attention from battlefields to civilians. Instead of interstate wars, this map highlights where people are directly targeted within their own states.

Violence is not evenly distributed. It clusters geographically, particularly across parts of Sub-Saharan Africa and the Middle East, pointing to structural instability rather than isolated events.

How to read

Click the buttons to reveal two different dimension: Darker colors indicate higher levels of one-sided violence against civilians.

• Absolute totals shows where the largest number of civilian deaths occurred.
• Per-capita rates (per 100,000 people) shows intensity, where violence affects a larger share of the population.

Notably, several countries appear in dark gray represent no recorded data in the dataset.

Concentration of Civilian Targeting

When we isolate incidents where civilians are the explicit targets, a stark ranking emerges. The chart shows the top 10 countries by the total number of recorded civilian-targeting events in recent years.

This is not a scoreboard but a way to reveal structural concentration. High values can reflect both persistent violence and stronger reporting capacity.

Mexico and Brazil sit at the top, shaped by sustained criminal violence across very large populations. But the pattern shifts when smaller countries rise: Syria and Palestine appear near the top despite far smaller size, signaling an intense concentration of civilian targeting.

How to read

• Each row is a country; farther right means more recorded events.
• Each circle shows total recorded events from 2017 to 2025.

Protest → Response on the Map

Mapping protest activity alongside violence against civilians reveals how unrest and repression occupy shared spaces. Rather than viewing these events in isolation, the map traces their geographic proximity and concentration.

Conflict does not unfold evenly across territory. Instead, protest events and targeted violence cluster within specific regions, particularly across parts of the Middle East, suggesting localized patterns of escalation rather than random distribution.

How to read

Use the layer selector to isolate different event types. Dot size represents fatalities per event, while color distinguishes between protests, battles, and violence against civilians. Hover over each dot to inspect country, administrative region, and fatality details.

Security Shifts: 2017 vs 2025

This slope chart compares fatalities between 2017 and 2025 to highlight structural change over time. Rather than focusing on annual volatility, it reveals how conflict intensity evolves across a longer horizon.

Fatalities decline sharply in Syria, Iraq, and Yemen compared to their peak conflict years. However, other countries such as Iran and Lebanon show rising or persistent levels of violence. The region does not move in a single direction — intensity contracts in some areas while emerging elsewhere.

How to read

Each line represents a country. The left point marks fatalities in 2017, and the right point shows 2025. Upward slopes indicate increasing fatalities; downward slopes indicate decline. The steeper the line, the larger the magnitude of change.

Nature of Violence Changes Across Contexts

This chart compares how violence is structured across countries. Instead of measuring how much violence occurs, it reveals how different types of events compose the overall pattern — military confrontation, civil unrest, and state repression.

The composition varies significantly. In some countries, violence is dominated by military conflict, while in others, civil unrest accounts for the majority of recorded events. State repression appears as a smaller but distinct component in most contexts. These differences suggest that conflict is not uniform in nature — it reflects different political and social dynamics.

How to read

Each bar represents 100% of recorded events in a country. The segments show the percentage share of each event type. Switching between “Events” and “Fatalities” reveals whether the structure of violence (event mix) aligns with its severity (fatal impact).

Protests and Civilian Violence Over Time — Iran

This dual-axis chart compares the monthly number of protest events with recorded civilian fatalities over the same period from 2017 to 2025 in Iran. Rather than measuring scale alone, it allows us to examine timing, how these two dynamics evolve side by side.

In several moments, spikes in protest activity coincide with increases in civilian fatalities. These overlaps suggest potential escalation windows, where unrest and lethal repression appear closely aligned.

An escalation window does not imply a simple chain of cause and effect. Rather, it highlights compressed periods of instability, when institutional control, public mobilization, and coercive force converge.

How to read

• Blue line (left axis): monthly protest events in Iran.
• Red line (right axis): recorded civilian fatalities.
• Alignment in peaks indicates temporal co-occurrence.

Waves of Violence

This ridgeline compresses Iran’s weekly record of civilian fatalities from 2017 to 2025 into a layered landscape. Each ridge represents one year, and its vertical height encodes the number of recorded fatalities in a given week.

Violence does not unfold evenly across time. Extended periods remain relatively subdued, interrupted by sudden, sharply rising peaks. In certain years — most visibly 2019 and 2022 — these peaks rise distinctly above surrounding periods, forming what can be described as “bloody weeks” in the recorded data.

How to read

Read horizontally to follow the progression of weeks within a year. The horizontal axis approximates months of the year (derived from the week index), while vertical height reflects weekly fatality counts. Comparing ridges vertically reveals differences in intensity across years.

From Protest to the Void

This Sankey diagram traces documented protest-related outcomes in Iran during the “Woman, Life, Freedom” uprising, which began in September 2022. The figures reflect cumulative cases recorded through February 21, 2023.

A large flow moves into arrest, a narrow red branch ends in deaths, and a substantial share fades into Status Unknown (The Void) — cases whose identities or legal status could not be independently verified amid censorship and reporting constraints.

How to Read

Hover a link to highlight its path and see the reported count. Node values are cumulative totals from the compiled dataset. The leftmost “Protesters” node is a constructed baseline: it equals Arrested + Killed within documented outcomes, not the true number of protesters.

Where Conflict Exists but Data Does Not

The Illusion of Calm

This calendar shows monthly civilian deaths recorded by ACLED in Iran from 2017 to 2025. Each cell reflects documented events reported through international monitoring systems.

November 2019 stands out as one of the darkest months in the entire series. Even within international datasets, it registers as a moment of severe violence.

But what appears here is only what could be observed and verified under conditions of restricted visibility.

Scroll ↓ forward for more context

Isolating the Uprisings

When we isolate the months of the 2019 and 2022 uprisings, a clearer pattern emerges. The heatmap reveals spikes in recorded fatalities during these periods, contrasting with the surrounding months.

In both cases, the state responded to protest with force. But the informational environments were different.

In November 2019, Iran experienced a near-total nationwide internet shutdown. When connectivity collapses, international event-based datasets become structurally dependent on delayed or indirect reporting.

The Internet Blackout Gap

When the internet was completely severed in 2019 (NetBlocks), ACLED recorded several hundred fatalities. Reuters, drawing on security sources and subsequent investigation, estimated approximately 1,500. The difference is not a contradiction between datasets. It reflects different reporting channels under different levels of visibility.

In Fall 2022, internet disruptions were extensive but not absolute. This time, citizen journalists bypassed the firewalls. VPNs worked. Videos leaked. The world watched. As a result, international databases captured the violence in real-time, drastically shrinking the gap between formal datasets and reality.