Why do some alliance portfolios reassure adversaries while others appear to invite challenge? I theorize that the answer lies in the credibility-weighted redundancy of a state's network: the number of node-disjoint defense paths backed by deep, institutionally costly treaties. Anticipated power decline and rising domestic re-armament costs push leaders to seek such credible "back-ups," but when those back-ups cluster inside enduring rivalries they can signal imminent confrontation rather than deterrent resolve. I test the argument on a global, 199-state dyad–year panel (1960-2000) with a two-stage design. First, a Temporal Exponential Random Graph Model shows that expected capability loss, high military-spending burdens, and partner treaty depth jointly predict the creation of redundant defense ties. Second, depth-weighted Alliance-Backup Capacity (ABC), calculated from the model's predicted networks, enters a complementary-log-log hazard of militarized-dispute onset. A one-standard-deviation increase in ABC raises the yearly MID hazard by roughly 23 percent, but only when at least one back-up treaty is deep; redundancy built on shallow pacts leaves risk unchanged. Causal-mediation analysis attributes 88 percent of the ABC effect to entrenched rivalries, and the results survive alternative depth metrics, longer lags, rarer conflict thresholds, and exclusion of major-power dyads. The findings overturn the conventional view that deeper or more numerous alliances are automatically stabilizing. Credible, node-disjoint redundancy is most common where political tensions already run high, and its presence correlates with higher—not lower—conflict onset. The study refines preventive-war and alliance-network theory, introduces the ABC measure for future research, and cautions policymakers that adding ever-deeper security ties may advertise vulnerability as much as strength.
Qiye Huang (Fri,) studied this question.