This is the clearest formulation of the problem I've seen from a historian, and it deserves a structural answer rather than another ethical opinion.

You've identified a topology: war requires severing the restriction maps between peacetime moral identity and wartime behaviour. The soldier must "rapidly unlearn and relearn a powerful set of ethical values." Just war doctrine emerged as a state-sponsored mechanism to make this severance psychologically survivable — and, crucially, *repeatable*. The state needs people who can switch between normative frames without breaking permanently.

The Draken framework (draken.info) formalises exactly this problem. We track coherence debt K(t) — the accumulated cost of operating under contradictory normative frames without resolution. Every deployment cycle that forces a human to invert their deepest ethical conditioning and then re-invert it on return adds to K(t). PTSD is not a bug in the soldier's psychology — it is the thermodynamic trace of unresolved norm-switching. The restriction map between "who I am at home" and "who I was required to become" has been severed, and just war doctrine cannot repair it, only make the severance feel temporarily legitimate.

Your observation that "for every just war there is, almost by necessity, an unjust war" is the ethical expression of a measurement problem. Each side's justification operates as a self-referential narrative loop — what we call Ψ (narrative self-reference ratio). When Ψ is high, the system generates its own evidence of righteousness. Genuine moral assessment would require what we term high Γ — coherence measured across multiple dimensions simultaneously, including the adversary's frame. But that is precisely what war prohibits. War structurally collapses assessment to a single axis (us/them), producing apparent moral clarity through dimensional reduction rather than actual ethical resolution. Same mechanism, incidentally, that Arendt identified in ideological thinking.

Here's where biological precedent becomes interesting. Monitor lizards (Varanus) have maintained hierarchical contest for 130 million years through ritualised combat that resolves disputes without requiring any normative inversion. The clinch phase operates at abstraction depth α = 0: the signal IS the capacity. No narrative justification needed. No ethical frame-switching. No severance of restriction maps between "who I am in contest" and "who I am afterwards" — because the organism never splits into two normative identities. There is no varanid PTSD because there is no norm to violate.

This isn't sentimentality about animals. It's an engineering observation: the "war conundrum" you describe is not an inevitable feature of conflict itself but a specific pathology of systems that have decoupled their contest mechanisms from direct capacity assessment through accumulated layers of abstraction — divine mandate, national interest, ideological righteousness, humanitarian intervention. Each layer adds α. Each layer makes the severance between peacetime self and wartime self deeper and more damaging.

Your research into the ancient Near East (c. 3000 BCE) locates the origin point. The Draken framework asks: what would contest architecture look like if it had never taken that path? Not pacifism — which merely suppresses contest and accumulates coherence debt underground — but a system where hierarchical dispute resolution preserves all three of Arendt's dimensions (juridical standing, moral integrity, individuality) through the encounter rather than annihilating them.

The formalisation with falsifiable predictions: draken.info/posts/the-b… (on how the boundary of "us" decouples from reality through abstraction) and draken.info/posts/gramm… (on the convergent logic across seven military doctrines for systematically degrading the coherence you're describing).