🔎 The puzzle: Why do some governments commit genocide or politicide against civilians during civil war? Prior work treats such brutality mainly as a tool to secure power through outright military victory. This article offers a different take by returning to the logic of conflict bargaining and reconceptualizing genocide and politicide as methods for extracting information about an opponent’s strength and intentions.

📌 Core argument: Atrocities can function as information-gathering strategies. A government facing uncertainty about its probability of victory is more likely to employ genocide or politicide to force revealing behavior on the battlefield and learn faster whether victory is attainable.

📊 How the claim is tested: All civil wars since 1945

• Empirical tests use a global dataset of civil wars covering the post-1945 period.
• Statistical analyses examine the relationship between government uncertainty about victory and the onset of genocidal or politicidal campaigns.

📈 Key findings:

• Greater uncertainty about the probability of victory predicts a higher likelihood that governments will initiate genocide or politicide.
• These dynamics are notably stronger when rebel movements depend on the civilian population to mobilize fighters, making civilians a critical information channel.
• Results are consistent across the universe of civil wars since 1945 and robust to alternative specifications.

⚖️ Why it matters: This argument connects and advances two literatures—genocide studies and conflict bargaining—by showing atrocities can be driven by informational imperatives, not solely by a desire for territorial control or punishment. Understanding atrocities as attempts to reduce strategic uncertainty clarifies when and why governments escalate violence against civilians and points to different intervention and monitoring strategies during contested conflicts.