"But war is a Darwinian process"

No, it's not. If "Darwinian" means "pertaining to or resembling biological evolution", which it generally is understood as, then there are certain characteristics of biological evolution that one can reasonably expect to be present for the term to be proper: a genotype governing the development of organisms and consisting of genes (units of heritability), a phenotype physically expressing a particular genotype, a copying process that is largely faithful in transmitting genes but occasionally mutates them, a method of combining genes from two individuals, and a selection process based on phenotype. What is the "organism" in the case of war? The genotype? The phenotype? The genes? The copying process? Etc. Ask Billy Beane whether the mere existence of competition means perfect efficiency. Warfare involves signaling, brinkmanship, collusion, negotiation, bluffing, path dependence, local equilibriums, information asymmetry, superstition, stochastic processes with samples sizes completely inadequate to estimate underlying parameters, and more. There's no reason to think it's efficient.