Study Shows How Escalation Can Send Wars Spiraling Out Of Control

What Decades of War Data Reveal About How Conflicts Get Out of Hand

Most people assume the deadliest wars in history arrived as catastrophes from day one. A sweeping new analysis of armed conflicts says otherwise. Wars become massive not because they begin at a massive scale, and not because they drag on for years. According to researchers, the real engine of destruction is escalation, a step-by-step intensification of violence that, once set in motion, can spiral into some of the most devastating events in human history.

Published in Science Advances, the study examined civil wars using records from 1946 onward and wars between nations using data reaching back to 1823. Researchers found that the single most important factor separating a contained conflict from a catastrophic one is whether it escalates, meaning whether fighting intensity ramps up over time. That finding carries weight now, given that, as of 2026, interstate wars are at their highest level since the late 1980s, with conflicts like the 2022 Russian invasion of Ukraine reshaping global security.

Two Long-Running Databases Track War, Year by Year

To dig into this, researchers from the University of Colorado and three other universities pulled records on nearly 300 civil wars going back to 1946 and close to 100 wars between nations going back to 1823. Instead of just adding up how many people died in a war by the time it ended, they broke each conflict down year by year, checking whether the death toll went up, went down, or held steady from one year to the next. That single shift, looking at a war’s ups and downs instead of just its final body count, is what let them see something earlier researchers had missed.

Escalation Is Common, and Fighting Doesn’t Reliably Cool Wars Down

Escalation turned out to be common, not a rare feature of a handful of extreme wars. Most years within a conflict show little change in intensity, but occasionally a conflict undergoes a massive spike, many times more deadly than the year before, and the chance of intensification closely mirrors the chance of cooldown. Wars did not wind down as both sides learned more about each other; violence tended to hold steady, then spike unpredictably.

That challenges a popular theory in conflict studies holding that war is fundamentally an information problem, in which fighting teaches both sides about each other’s real strength and makes peace easier to reach. If that were true, wars should de-escalate as they go on. Instead, fighting does not reliably reduce uncertainty or make de-escalation more common over time.

Civil Wars Tend to Pull Back at Extreme Scale; Interstate Wars Don’t

Among civil wars, the pattern that mattered showed up once fighting got extremely deadly. When a civil war’s yearly death toll climbed past a few hundred, that war became more likely to calm down the next year than to keep escalating. It is not a hard limit, more like a brake that tends to kick in once things get bad enough, and the researchers suspect it might come down to running out of fighters to recruit, war fatigue setting in, or outside pressure from other countries once the violence gets extreme enough to draw attention. They are careful to say the real cause is still an open question.

Wars between nations show no version of that brake. When the researchers built a model of interstate war using the same kind of year-to-year ups and downs, but without that civil-war-style pullback, it reproduced a pattern scientists have documented for decades but never fully explained: most wars between nations stay fairly limited, but every so often one grows into an outlier so much larger than the rest that it skews the whole picture. That pattern has a name, Richardson’s law, after the scientist who first identified it in the 1940s. This study suggests the reason it holds up so well may be simple: escalation among nations just doesn’t have a built-in brake the way it does inside civil wars.

Escalation Makes War’s Final Toll Nearly Impossible to Predict

One sobering takeaway concerns forecasting. Because escalation is so unpredictable, no one can reliably estimate how bad a war will get once it starts. In scenarios the researchers modeled, projected death tolls varied by more than a factor of ten in either direction, and each additional year of fighting carries roughly a 10% chance the conflict’s severity doubles and about a 1% chance it multiplies tenfold. Those figures are model-based estimates, not guarantees, but they echo Carl von Clausewitz’s classical warnings about the fog of war. Decision-makers who assume they can anticipate a conflict’s eventual scale are, according to this data, operating under a dangerous illusion.

Most conflict-risk tools focus on whether wars will start and how they end, with far less attention paid to how large they might grow once underway. This study makes the case that escalation deserves a seat at that table.

---

Disclaimer: This article is based on peer-reviewed research and is intended for general informational purposes. It does not offer geopolitical, military, or policy advice, and readers should consult the original study and other expert sources for decisions involving specific conflicts or regions.

---