There is a particular kind of historical vertigo that comes from reading about the Thirty Years' War as a young man and being unable to fathom how thirty years could pass while everyone concerned — emperors, electors, generals, peasants — apparently wanted the killing to stop, and yet it did not stop. One assumes, at that age, that wars end when someone wins, or when everyone is too exhausted to continue. The Thirty Years' War disabuses one of this assumption rather thoroughly. It ended not because exhaustion produced victory, but because two separate congresses, in two separate Westphalian towns, spent years constructing a peace from scratch — because no natural mechanism existed to produce one.
It is a mechanism worth revisiting now, not as metaphor but as diagnostic tool.
The Wallenstein Problem, Corrected
The lazy version of the analogy to Ukraine goes roughly as follows: wars, once they run long enough, cease to be instruments of policy and become self-sustaining economies with their own constituencies for continuation. Wallenstein's armies, after all, financed themselves through contribution and requisition, which meant that peace was a redundancy notice for the general as much as a blessing for the countryside he was despoiling. One might construct a tidy parallel — foreign-financed war effort, a population subsisting on external largesse, a leadership with every incentive to keep the cheques arriving.
The trouble is that the parallel, examined seriously, does not survive contact with the evidence. Polling this year shows something close to the reverse of a population content to let the war fund its existence: support for pushing hard for a negotiated end has climbed to roughly two-thirds, against fewer than a quarter still favouring the fight to outright victory — almost a mirror image of sentiment in 2022. Whatever else one wishes to say about the Ukrainian public, "living comfortably off the war" is not a description that survives the polling.
Nor does the case for a population-level moral hazard survive a moment's scrutiny of what "temporary protection" status actually is. It is not refugee status under the 1951 Convention; it is a deliberately looser instrument, designed after the Yugoslav wars precisely to accommodate people who might need to cross a border home for a funeral, a harvest, or simply a fortnight's rest without forfeiting protection abroad. A queue at a border crossing is not, on its own, evidence of fraud. It is evidence that the instrument is working as designed — which is a different complaint, and a much smaller one, than the one usually made of it.
Where the Analogy Actually Bites
But dismiss the crude version and a sharper one appears in its place, and it happens to track the Thirty Years' War rather more precisely than the Wallenstein story does.
The structural failure of the 1618–1648 war was not that anyone particularly wanted it to continue. It was that no single actor possessed the authority to impose a settlement on all the others simultaneously. The Empire was too fragmented; the confessional stakes too high for any single defeat to be treated as final. Peace, when it came, had to be built by committee, across years, in two towns at once, because the war itself had never produced — could never produce — a mechanism for its own resolution.
Substitute distributions for sovereigns and the same failure reappears in miniature. A majority of Ukrainians want negotiations. But more granular polling on the actual terms on offer tells a less comforting story: something on the order of half reject the sort of settlement resembling Russia's stated demands, while a rather larger share is prepared to accept a European-brokered alternative that Moscow has shown no sign of entertaining. Wanting peace and wanting a peace that is currently available to be had are not the same poll, and the gap between them is where wars like this one go to become permanent. One can want the war to end and still reject, one after another, every settlement anyone is capable of offering — and the practical effect of that combination is indistinguishable from not wanting it to end at all.
Layer onto this a second, entirely separate mechanism, this one institutional rather than popular: a constitution that forbids elections under martial law, a presidency four years past its mandate, and a peace process in which the scheduling of a vote is explicitly tied to the receipt of security guarantees rather than the other way round. Whatever one makes of the necessity of the arrangement — and the practical case for it is not trivial, given the described impossibility of organising a wartime ballot for soldiers at the front and refugees scattered across a continent — it does produce an institution with every rational incentive to prefer the current arrangement to an untested one. That is not corruption. It is simply what incumbency without an electoral check tends to look like, in any century.
A Peace Deferred to Committee
Even the diplomacy has taken on a faintly Westphalian shape. Ukraine's proposed date for EU accession, floated for 2027, is bound up in the same peace architecture as the security guarantees and the elections — a package that cannot be disentangled into its parts, and that several capitals regard as impossible on its stated timetable regardless. One is reminded, not for the first time, that "as long as it takes" is precisely the sentence uttered by every guarantor of every protracted war, right up until the year it finally doesn't take any longer.
None of which will surprise anyone who has read Osnabrück and Münster with any care. The peace, when it eventually comes, will not emerge from anybody's exhaustion. It will have to be constructed, term by unwelcome term, by people who spent years insisting they wanted something rather simpler than what they eventually signed.
Composed in dialogue with Claude Sonnet 5 (Anthropic), on the basis of a running exchange concerning Peskov's rhetoric, the Thirty Years' War, and the structural mechanics of Ukraine's peace prospects. Polling and diplomatic detail checked against current reporting as of July 2026.
