A new RAND dissertation by Krystyna Marcinek identifies three mechanisms by which autonomous weapons raise conflict odds; the third, and least studied, is the increasing political benefits of fighting.
Robotic weapons on the battlefield mean fewer body bags at home, and that asymmetry is reshaping the political calculus of going to war. A new RAND dissertation from Krystyna Marcinek argues the political payoff of fighting, not the dollar cost or the prospect of victory, is the channel through which autonomous weapons most directly raise conflict odds.
Marcinek's framework was built to test that claim directly. She developed a game-theoretic model of great-power confrontation over a third country, populated it with three channels by which autonomous systems could shift a leader's incentive to use force, and asked which channel moved the probability of war the most. Two of the channels are familiar. Cheaper wars in dollar terms and higher prospects of battlefield success have dominated the academic and policy conversation about robotic weapons for years. The third channel, the political payoff of fighting, has largely gone unmodeled.
That third channel works through asymmetry. When a leader can deploy unmanned systems instead of troops, the visible domestic cost of intervention shrinks: no flag-draped coffins, no draft, no clear footprint at home. The benefits of action still arrive in full: a crisis resolved, an adversary deterred, a regional order reshaped. Marcinek's model captures this as a tilt in the political utility function of going to war, where the expected payoff of aggression rises and the expected domestic punishment falls. The decision to use force becomes easier to defend in a televised address.
"Overall, the research suggests that innovations in autonomous weapons systems may increase the likelihood of war through three mechanisms — namely, by (1) improving prospects of success, (2) decreasing economic costs of war, and (3) increasing political benefits of fighting," Marcinek writes. The other two channels reinforce, rather than compete with, the political one. Cheaper wars in dollars translate, in democracies especially, into cheaper wars at the ballot box, with less fiscal pain, fewer second-order economic effects for voters, and less organized opposition to a deployment. Higher prospects of success make the same political move more defensible after the fact.
There is a countervailing signal in the dissertation that the policy debate has not yet metabolized. Marcinek tested whether the prospect of using unmanned systems changes Russian public support for military intervention, and found no statistically significant effect. The implication cuts two ways. If publics do not reward leaders for sending robots instead of soldiers, the political payoff of going to war does not rise automatically with autonomy. It rises only when leaders successfully frame the intervention in the terms their domestic audience cares about. The political benefit is contingent, not mechanical.
That contingency is what makes the dissertation most useful to defense planners and arms-control researchers. The question is no longer whether autonomous weapons change escalation incentives in the abstract, but under what domestic-political conditions the political-payoff channel activates, and when it does not. For the United States, China, and Russia, the answer will depend on casualty aversion, media environments, and the structure of public accountability for the use of force, all of which differ sharply across regimes.
Marcinek frames the work as a starting framework rather than a final answer. The model abstracts from specific weapon systems, doctrines, and rules of engagement, and does not yet distinguish between remotely piloted drones and fully autonomous targeting. For two decades, the policy debate has asked how to make autonomous weapons safer once a conflict begins. The dissertation turns that question around, asking instead whether they make conflict more likely to begin at all.