"Offense/Defense Dominance" is a reference entry by Keith Dowding in the Sage Reference Encyclopedia of Power, published April 7, 2011. It defines the offense/defense balance in international relations — whether prevailing military technology, geography, and doctrine make conquest relatively easy (offense dominant) or relatively difficult (defense dominant) — and surveys the factors that shape that balance. It is held in this collection as background for AI autonomous-weapons and cybersecurity debates.
Summary of the framework
The entry distinguishes two states of the offense/defense balance. Under offense dominance, conquest of territory is unusually easy, even if absolute defense may still be easier than offense. Under defense dominance, conquest is unusually difficult.
It further separates two levels at which the balance operates. The tactical offense/defense balance is the relative ease of offense versus defense between military forces at a point of attack. The strategic offense/defense balance is the relative ease of conquering or defending states. Tactical advantage tends to create strategic advantage, but not always: Dowding notes that Napoleon defeated Russian armies yet could not conquer Russia.
Factors shaping the balance
The entry identifies several factors that move the balance toward offense or defense:
- Military technology and doctrine. Nuclear weapons are strongly defense-dominant, because a secure deterrent is easier to maintain than to threaten. Guerrilla warfare is defensive. The combination of machine guns, barbed wire, and railroads made WWI defense-dominant, while blitzkrieg made WWII offense-dominant.
- Geography. Mountains, rivers, oceans, and forests favor defense by impeding attackers, so island states and landlocked plains differ in their exposure.
- Domestic factors. Mobilization capacity, political cohesion, and economic depth shape a state's ability to sustain offense or defense.
- International institutions. Alliances can shift the effective balance.
Relevance to AI governance debates
The offense/defense framework supplies vocabulary for several cross-cutting AI policy debates.
In the autonomous-weapons debate, the AI arms race is generating offense-dominant technology at scale, including swarm drones and autonomous strike systems. RAND exercises found that "the speed of autonomous systems led to inadvertent escalation" (NYT 2026). The framework is used to analyze why AI weapons might be destabilizing, on the argument that fast autonomous offense could erode the deterrence structures that nuclear weapons established.
In the cybersecurity debate (AI and Cybersecurity), the central question is framed as an offense/defense balance one. Bruce Schneier argues that AI is better at finding vulnerabilities than patching them, an offense-dominant dynamic in cyber (Source: washingtonpost.com). Dimitri Alperovitch argues that this advantage is temporary, lasting roughly 12–18 months.
The nuclear analogy recurs across AI arms-race discussion (NYT 2026, Superintelligence Strategy (Hendrycks, Schmidt, Wang), We Need a Science of Scheming). Nuclear weapons shifted the offense/defense balance toward defense via assured second-strike capability; one line of argument holds that AI autonomous weapons may move it in the opposite direction.
Provenance
Author: Keith Dowding. Publication: Sage Reference, Encyclopedia of Power. Date: April 7, 2011.
Relationships
- supports: Autonomous Weapons — analytical vocabulary for why AI weapons might be destabilizing
- supports: AI and Cybersecurity — offense-dominant framing of AI vulnerability finding vs. patching
- related: Mutually Automated Destruction: The Escalating Global A.I. Arms Race (NYT) — the AI arms race article uses implicit OD reasoning
- related: ICRC Position on Autonomous Weapon Systems (source summary) — prohibitionist argument implicitly relies on defense-dominance reasoning (autonomous weapons remove defender human judgment advantages)
- related: Superintelligence Strategy (Hendrycks, Schmidt, Wang) — MAIM deterrence logic draws on nuclear-age OD analysis