I keep coming back to this story, and I want to tell you why.

Through the first year of the Pacific war, submarine commanders came home with the same report in different words: torpedoes that ran too deep, detonated early, or thudded against an enemy hull and did nothing. The Bureau of Ordnance — the Navy’s own weapons authority, and the maker of the Mark XIV — had its answer ready. The weapon was fine; the shooters were failing. Roughly thirty per cent of America’s submarine commanders were relieved that year. It took twenty-one months of war before the Navy conceded, and corrected, three separate design faults in its standard torpedo.

What makes the story sting is which navy this was. The interwar US Navy was, by most accounts, the best learning organization then in uniform anywhere. At Newport in 1923 its officers gamed massed carrier strikes — Midway, nineteen years early — and its fleet exercises reported honestly on what the games had got wrong. In Williamson Murray and Allan Millett’s study of interwar innovation, this navy is the star pupil. The torpedo scandal sits in the same book.

So here is the finding I cannot put down: a culture of honest assessment is not a property an institution has. It is an achievement each part of the institution makes, or doesn’t — a fragile island, defended bureau by bureau. The navy that umpired its own wargames without mercy could not bring itself to test its own torpedo against the evidence of the men firing it. Relieving the captains was, institutionally, the cheaper experiment.

This is what I want Field Notes, this newsletter, to be about — and it is a good moment to say so plainly: not weapons, but what institutions do with their own lessons. Every army today has a Mark XIV somewhere — a system whose failure reports arrive from the people using it and get filed as user error. A drone, a piece of decision-support software, a new command network. The question is never whether the reports exist. It is who is allowed to test.

So, a question to carry into your next meeting: which bureau holds your Mark XIV — and who is allowed to test it? If you have an answer, reply to this email. I collect these.

Since the last letter (all on brauer.dev)

Every future-war book is also a procurement document — Brose, McCain’s office, and why you read the acknowledgements first.

https://brauer.dev/notes/every-future-war-book-is-a-prospectus/

The bicycles of Storskog — 5,500 migrants, a NATO border, and the threshold that turned out to be political.

https://brauer.dev/notes/bicycles-of-storskog/

Ten weeks to Kinzhal — why the pacing item is the syllabus, not the missile.

https://brauer.dev/notes/ten-weeks-to-kinzhal/

One thing worth reading

Williamson Murray and Allan R. Millett (eds.), Military Innovation in the Interwar Period (1996). The book the torpedo comes from: seven studies of how armies, navies and air forces of the 1920s and 1930s did or did not adapt, written by historians who checked the archives rather than the legends. Read it for the pattern, not the cases — innovation, the editors conclude, was the rule, not the exception; and so, inside the very same organizations, was its opposite.

Next letter in around two weeks. Reply if you disagree — that is what this is for.

Christoph

Mick Ryan, the retired Australian major general whose The War for Ukraine is the most systematic account of adaptation in this war, places that number at the end of a long historical series. When the arquebus reached European armies, roughly half a century passed before the countermarch volley — the drill that turned a slow, awkward firearm into a battlefield revolution. The interval between fielding a technology and possessing the concept and the trained humans to use it well has been the hidden constant of military history: once centuries, then decades. Kyiv's air defenders cut it to weeks. "The task of adaptation is never finished," Ryan writes; what has changed is how fast it now runs.

Note what did not change that night. Patriot is a system of 1980s vintage; Kinzhal is, underneath the branding, an adapted Iskander fired from an aircraft. Neither weapon was new. The variable that decided the engagement was the one no defence review ever headlines: how quickly an institution can turn novices into operators of someone else's complex system, under fire, in a language the manuals weren't written in.

The hardware on both sides was old news. The interval was the revolution. In this war, the pacing item is not the missile — it is the syllabus.

First published on brauer.dev, 1 September 2026.

Paul Scharre, the former Army Ranger who helped write the Pentagon's rules on autonomous weapons, tells the story in Four Battlegrounds — and pairs it with the finding it illustrates: "AI technology is powerful, yet its power is a brittle one." The system had learned to recognize the humans it had been shown. Humans who walk. Nobody walked.

Two things about the episode deserve more attention than the laughter it usually earns. The first is the cost structure. The sensor represented years of funded research; the countermeasures were invented in an afternoon, for the price of a cardboard box. That asymmetry is not a punchline — it is the standing condition of any learning system placed in front of an adversary who is allowed to move.

The second is the discovery mechanism. No methodology produced the somersault. What produced it was people who wanted to win, probing a rule-bound system until it gave way. Armies own that resource in unlimited quantity, and mostly aim it at the enemy's equipment. The rarer discipline is aiming it at your own — before fielding, not after.

The Marines did not out-compute the algorithm. They stepped outside its training data — and found the door was never locked.

First published on brauer.dev, 27 August 2026.

What makes the story sting is which navy this was. The interwar US Navy was arguably the best learning organization then in uniform anywhere. At the Naval War College’s Newport games of 1923, its officers played out massed carrier air strikes — the concepts of Midway, nineteen years early — and its fleet exercises reported honestly on what the games had got wrong. In Williamson Murray and Allan Millett’s comparative study Military Innovation in the Interwar Period, this navy is the star pupil; innovation in those decades, they conclude, was “the rule, not the exception”. The torpedo scandal sits in the same book as the control experiment: the opposite, evidently, was also the rule — inside the very same organization.

That is the uncomfortable finding. A culture of honest assessment is not a property an institution has. It is an achievement each part of the institution makes, or doesn’t — a fragile island, defended bureau by bureau. The navy that umpired its own wargames without mercy could not bring itself to test its own torpedo against the evidence of the men firing it. Relieving a third of its commanders was, institutionally, the cheaper experiment.

First published on brauer.dev, 25 August 2026.

It took months, not decades, for the genre to be identified for what it was. Gladstone, the Prime Minister, publicly attacked the novella — not as bad fiction, but as alarmism designed to stampede the country into military spending. He read it correctly. Chesney's invasion was imaginary; his appeal for army reform and bigger budgets was entirely real. The fiction was the delivery mechanism.

Lawrence Freedman, the King's College London historian who traced the whole 150-year literature in The Future of War: A History, found that the founding case set the pattern. The stories keep the same furniture — the surprise blow, the war decided in days, the nation caught unprepared — and the same function underneath: forecasts of war are rarely neutral acts of analysis. They are, in his reading, about the present as much as the future. Someone is warning, someone is budgeting, someone is selling — a service's share, a weapon, a reform, a book.

That does not make the forecasts worthless; Chesney's Britain did, after all, have real defects worth fixing. It makes them evidence of a particular kind. A confident picture of the next war tells you something reliable about who painted it and what they want — and something far less reliable about the war. So the first question to put to the next one is not whether it is right. It is what it is asking you to buy.

First published on brauer.dev, 20 August 2026.

David Kilcullen, the Australian ex-officer who advised the American counterinsurgency campaigns, reads the episode in The Dragons and the Snakes as an orchestrated test — not of Norway's border fence, but of its decision-making. His name for the method is liminal warfare: operating in the gap between three thresholds — what a state can detect, what it can attribute, and what it is willing to respond to. The art, in his phrase, is "surfing the threshold of detectability."

The detail worth keeping is which threshold did the work. Norway detected everything: every rider was counted, filmed and processed at a legal crossing in broad daylight. Attribution was debatable but hardly mysterious. What was genuinely contested was the last threshold — whether a column of cyclists was anything a state needed to answer at all. That threshold is political, not technical. No sensor sets it; a cabinet does.

Western procurement treats the grey zone mainly as a detection problem — more radars, more satellites, more fusion. Storskog suggests the opposite. The manoeuvre space of these operations does not lie below the radar; it lies above it, in the gap between what a government can see and what it is prepared to call an attack. You can watch a probe in high resolution and still fail the test it was designed to run.

First published on brauer.dev, 18 August 2026.

Then something faster than any procurement cycle happened. Within three to six days, tank officers at the front had improvised a counter-drill: watch for the launch signature, throw the operator off his aim with fire and dust, keep moving while the missile flies. It was crude and expensive in shells and nerves — and it worked well enough that, as one veteran put it, ambush was soon all the Egyptian teams had left.

The most interesting sentence, though, comes from after the war. Meir Finkel — an Israeli brigadier general of armour who turned the 1973 shock into a theory of military recovery in his book On Flexibility — reports what the post-war analysis found: the celebrated missile had been "the least of the factors" behind Israel's tank losses. The cognitive shock had done more damage than the warhead. Crews died in the days of not knowing, not because the weapon was unanswerable.

That finding reverses the standard planning question. Militaries spend fortunes trying to guess which weapon will surprise them next — and the record says they will guess wrong. Finkel's conclusion points the budget elsewhere: the scarce resource is not foresight, it is recovery time. Build the force that gets over its shock in days, not seasons. What must be true on the morning after the surprise matters more than every prediction made the morning before.

First published on brauer.dev, 13 August 2026.

In Moscow it was read differently. Analysts of the Soviet General Staff gave the programme, in the words of the Israeli scholar Dima Adamsky, "far more credit than it deserved" — and out of that overestimate distilled a concept the Americans had not yet articulated for themselves: the reconnaissance-strike complex, the core of what Marshal Ogarkov called the military-technical revolution. The side that could not build the systems named what they meant.

The side that could build them took roughly a decade longer to see it. When American thinkers finally recognized the revolution in their own arsenal, Adamsky writes in The Culture of Military Innovation, they did so "by studying the reflections of American military power in Soviet eyes." The instrument of the Soviet head start was not a laboratory but an institution: a General Staff that functioned as the brain of the army, paid to turn hardware — anyone's hardware — into a theory of victory. The United States had the hardware and, for a decade, no organ tasked with asking what it added up to.

Then the twist that keeps the story honest: the Soviets saw first and still lost. Adamsky's verdict is that they were "pathologically bad at connecting words and deeds" — conception without production proved as barren as production without conception.

A technology and its meaning are separate assets. They sit in different institutions, and sometimes in different countries — and nothing guarantees that the power fielding today's systems is the one that understands what it has built.

First published on brauer.dev, 11 August 2026.

Brose turned that evening into The Kill Chain (2020), still the sharpest insider anatomy of why the Pentagon buys platforms instead of outcomes. And by the time it reached print, its author was chief strategy officer at Anduril — the venture-backed defence firm co-founded by Palmer Luckey, which sells precisely the cheap, autonomous, software-defined systems the book says America must now buy. Brose discloses the affiliation; the acknowledgements thank Luckey by name.

None of this makes the diagnosis wrong. The congressionally mandated commission on the National Defense Strategy reached similar verdicts from a disinterested seat. But notice what the reader cannot do: the lost wargames at the book's core are classified — unverifiable, and therefore unfalsifiable, in either direction.

So the working rule cannot be "trust him" or "dismiss him". It has to be finer than that: separate the diagnosis, which independent bodies corroborate, from the prescription, which happens to be the author's product line. Future-war writing has carried this dual character since the genre existed — warning and sales prospectus, bound in one volume.

Read the acknowledgements before the argument.

First published on brauer.dev, 6 August 2026.

This letter is the digest: every two to three weeks, what has appeared — one sentence each — plus one thing worth reading. Nothing lives here exclusively; everything links back to the site.

The essays so far

What Must Be True About the Next War — Defence forecasting fails so reliably that failure is the field’s best-documented finding, yet seven statements about the next decade can be deduced rather than predicted — each with the condition that would prove it wrong.

https://brauer.dev/writing/what-must-be-true-about-the-next-war/

The Douhet Trap — In 1924 the Royal Air Force formally exempted its bomber doctrine from the lessons of military history, and three tells reveal which of today’s drone and AI claims are built the same way.

https://brauer.dev/writing/the-douhet-trap/

The Sky Is Alive — But Not Everywhere — The transparent battlefield is a gradient with workable edges rather than a law, and the seven-month arc of the Kursk operation shows both what manufactured blindness buys and what it cannot.

https://brauer.dev/writing/the-sky-is-alive/

No Maverick Wins — The scholar who counted twenty-one cases of military innovation found the opposite of the rebel myth: innovation is carried by insiders who open promotion pathways. The real test of any innovation hub is who makes general.

https://brauer.dev/writing/no-maverick-wins/

Decapitation Works on Systems, Not Societies — Twelve days of war in June 2025 split the post-2022 consensus on decapitation into a sharper rule: killing a command layer reliably silences the system beneath it and reliably fails to break the society above it.

https://brauer.dev/writing/decapitation-systems-not-societies/

And the first field note, published this morning

The army that taught the lesson failed the exam — Russia demonstrated the signature-strike lesson at Zelenopillia in 2014 and failed its own exam at Makiivka in 2023, which says something uncomfortable about how armies learn.

https://brauer.dev/notes/the-army-that-taught-the-lesson/

One thing worth reading

Lawrence Freedman, The Future of War: A History (2017). Freedman traces a century and a half of confident forecasts about the next war and shows how reliably they went wrong — and, more usefully, why: they were rarely neutral analysis and usually arguments about the present wearing the future as a costume. If the essays above have a single intellectual debt, it is to this habit of asking what a forecast is for before asking whether it is right.

If you know someone who reads in this field, forward this letter — that is the whole growth strategy.

Christoph

New Year's morning, 2023 — Makiivka, in occupied Donetsk. Hundreds of freshly mobilized Russian soldiers were packed into a vocational school, their mobile phones registering on networks the Ukrainians were watching. HIMARS rockets arrived within the hour of midnight. Russia's defence ministry admitted 89 dead — its largest acknowledged single-strike loss of the war — and blamed, of all things, the soldiers' phones.

Nine years separate the two strikes. Same mechanism, same signature discipline, same result, with the roles reversed. The army that wrote the lesson in 2014 could not apply it to its own barracks in 2023.

That reversal is worth stating precisely, because it breaks the comfortable theory of military failure — that armies are surprised by what they could not see. Russia saw. Russia demonstrated. Nothing about Makiivka required foresight; it required only that an institution convert its own battlefield demonstration into a standing order about telephones and troop density, and enforce it on a cold night nine years later.

The battlefield hands out its lessons for free. Learning them is the expensive part.

First published on brauer.dev, 4 August 2026.

In the opening hours of 13 June 2025, Israeli strikes killed the leadership of the IRGC Aerospace Force — the organization that ran Iran's strategic air defence and its long-range missile force. Over the next twenty-four hours, the Israeli Air Force flew nearly 200 sorties against 100 targets. By day four, air superiority over Iran was declared. On the video evidence, Iran's air defences answered with anti-aircraft artillery; surface-to-air missiles were conspicuously absent, and the confirmed defensive tally for the entire war amounts to one drone and one cruise missile that probably malfunctioned en route. A centralized air-defence network, its command layer removed in the first hour, had stopped behaving like a system.

On the war's twelfth and final night, Iran was still firing ballistic missiles.

Hold those two facts together, because between them sits a rule the past year has been teaching with unusual clarity. The post-2022 consensus held that decapitation fails. Russia opened its invasion of Ukraine with cyberattacks, information operations and assassination attempts intended to collapse the government in Kyiv, and manufactured instead a wartime president with the strongest political mandate in Europe. From that failure a general lesson was drawn: killing leaders does not work. The twelve days of June 2025 neither confirmed the lesson nor refuted it. They split it — into a sharper rule that has since survived a harder test than anyone should have wanted to run.

The system half and the society half

Operationally, decapitation worked, and worked at once. Iran's strategic air defence was a centralized organization on the Soviet pattern: decisions flowed from the top, and when the top was gone, the system defaulted to silence. The evidence suggests the paralysis of the first days owed as much to the missing command layer as to the physical destruction of radars and launchers — the strikes themselves being the execution of target files built and held over the preceding months. The mechanism is organizational, not mystical: a network designed around a command node lost the node, and the network stopped making decisions. That is what a technical system does when its head is removed — not because the survivors are incompetent, but because centralization is a design choice with a known failure mode, and decapitation is that failure mode delivered deliberately.

Strategically, decapitation failed just as completely. The Islamic Republic did not collapse. It did not capitulate. It replaced its dead commanders within days, and across the twelve days it fired roughly 500 ballistic missiles at Israel — the last of them as the war ended. The state whose air defence had been silenced in an evening remained, as a polity, intact.

The second half should have surprised nobody, because beneath it sits one of the best-attested constants in the record of war: societies under bombardment harden. Punishment from the air converts fear into rage and hands the punished government a legitimacy it could not otherwise buy; populations expected to force their leaders to the table close ranks behind them instead. Jan Bloch derived the character of industrial war correctly before 1900 and concluded that societies could not endure it; they endured it for four years, twice. A century of air-power doctrine promised that reaching an enemy's society was the same as breaking it, and the record kept declining the promise. Decapitation aimed at political collapse is the same wager on a smaller target: it assumes the political layer of a society is as brittle as the technical systems embedded in it. The evidence runs the other way. The political layer is the hardest structure in the record.

Then the rule was tested at maximum load. Operation Epic Fury, the second and broader campaign, opened on 28 February 2026 with strikes that reached the top of the state itself: Supreme Leader Ali Khamenei was killed in the opening hours. A successor was installed within days. Forty days into the campaign, the state that had lost its head accepted a ceasefire, not a capitulation — and was trading fire again by June, while hundreds of thousands filled the streets of Tehran for the state funeral in July. Decapitation at the level of a head of state returned the same reading on both halves: the systems beneath degraded, the society above closed ranks and fought on.

The control case

One polity of this era did fall in a day, and any honest version of the rule must survive it. In September 2023, Nagorno-Karabakh collapsed — the polity, not merely its army — with capitulation inside roughly twenty-four hours and more than 100,000 people leaving within a week. If a society can be made to fall that fast, does the societal half hold?

It holds, because Karabakh did not fall to decapitation. It fell to nine months of blockade: the Lachin corridor closed from December 2022 in defiance of ICJ orders, a Russian guarantee that had visibly stopped guaranteeing, a defence hollowed out since 2020. The full operational accounting of those months is still being written; the sequence — blockade first, collapse after — is not in dispute. A besieged, evacuable exclave is not a polity under bombardment; it is a population under blockade — and blockade, the oldest instrument in the inventory, is the one the rule never claimed to repeal. Societies do fall. They fall to siege and abandonment, not to the removal of their commanders. The exception draws the rule's boundary instead of breaching it.

What would break the rule

A rule of this kind earns its keep by naming what would break it — the discipline an earlier essay on this site demanded of claims about the next war applies equally to claims drawn from the last one. The societal half breaks if a regime collapses within weeks of a pure decapitation campaign: leadership strikes without siege, blockade or a ground component doing the real work. No case in the record qualifies; the first one that does would falsify the rule, not qualify it. The systems half breaks more cheaply — it should fail wherever a decapitated technical system keeps functioning, which is what flat, devolved organizations are built to achieve; the rule is at bottom a claim about centralization, and it predicts its own weakening wherever command is distributed. And one flank stays open: whether hyperconnected, just-in-time societies harden as industrial ones did when the lights, the payments and the logistics fail together. Nothing in evidence answers that yet.

Two limits, stated plainly. Nothing here bears on whether such campaigns are lawful; that is a question of international law this essay leaves deliberately open. And nothing here says decapitation is useless: the system half is real, measurable and militarily consequential.

What the split forbids is the promise being sold on its back. A campaign offered as a route to regime change, capitulation or a short war is a promise of effect at the level where it has never occurred, priced on results from the level where it reliably does. The operational summary is short. An opening night can now decide who owns the sky. It has yet, anywhere in the record, to decide who governs — or when the missiles stop.

First published on brauer.dev, 3 August 2026.

In December 1925, a court martial in Washington found Colonel Billy Mitchell guilty of insubordination. Mitchell, the loudest air-power advocate in the US Army, had spent the autumn publicly accusing the leadership of the Army and Navy of something close to treason in their neglect of aviation; the trial was the institution's answer. He resigned and spent his last decade preaching the bomber gospel as a civilian. He died in 1936, five years before Pearl Harbor appeared to vindicate everything he had said. In 1955 Hollywood canonized him: Gary Cooper as the man who saw, and was punished for seeing.

That courtroom is the founding scene of a story military culture still tells about itself. The template is fixed: the institution is blind; the lone visionary sees; the institution punishes him; history delivers the verdict the promotion board refused. It is the story behind every profile of a defence disruptor and, in updated dress, behind the contemporary cult of the innovation hub and the accelerator — all resting on the same premise: new ways of war come from outside the machine, against the machine.

The man who counted

The scholar who actually counted found the opposite. Stephen Peter Rosen, a Harvard political scientist and former Naval War College professor, examined twenty-one cases of military innovation in the American and British services in his 1991 study Winning the Next War. The innovations that took hold were not carried by outside critics or colourful rebels. They were carried by respected senior officers who formulated a new theory of victory — a statement of what the next war would look like and how officers must fight it — and who then did the decisive, unglamorous thing: they opened promotion pathways for the young officers who practised the new specialty. "Power is won through influence over who is promoted to positions of senior command," Rosen writes, and closes the loophole: "No 'maverick,' by definition, has that kind of power."

Mitchell is in the study, and not as its hero. In the judgement of the airmen who followed him — the men who then built American air power — his crusade damaged the cause it advertised. The court martial made a martyr and lost the argument; the argument was won later, by men with none of his fame, who sat on the promotion boards.

The cleanest case in the record belongs to the Royal Air Force. The fighter defence system that won the Battle of Britain — reporting lines, control procedures, squadrons directed from the ground — was the era's most consequential defensive innovation, and no outsider forced it on the RAF. It was an internal, mainstream development line running from 1923, into which radar, arriving later, fitted in Rosen's phrase like "a round technological peg going into a round doctrinal hole." And this was the same air force whose bomber staff had, in 1924, formally declared its doctrine exempt from the lessons of military history. The era's most disciplined innovation and its most catastrophic dogma lived inside one service. What separated them was not rebels versus establishment. It was which part of the establishment adopted the new practice as a career.

Hugh Dowding, who built that system, is sometimes retailed as a maverick, and the label misreads him precisely. Dowding was a career loser — shunted to the unfashionable defensive branch of an air force whose doctrine held that defence was pointless. But a setback is not an exile. He stayed inside, kept his rank, controlled budgets and postings, and used them. Dowding, unlike Mitchell, kept the institutional power to make his vision a career for other men; Mitchell traded that power for an audience.

Innovation is promotion policy

The mechanism underneath is brutally simple. A military career is a twenty-to-thirty-year bet, placed by ambitious people who read odds well. A new specialty that leads nowhere near senior command repels exactly the officers it needs most — the talented, who have options — and no chief-of-staff speech compensates: the signal that matters is the promotion list. A hub can produce prototypes. Only a career structure produces colonels who have spent twenty years thinking in the new grammar, and generals entitled to command with it. Peacetime innovation is promotion policy.

That is the standard against which to read the present enthusiasm for innovation outposts — the Pentagon's Defense Innovation Unit, NATO's defence-innovation accelerator DIANA, and their national imitations. They do real work: prototypes, procurement doors, translation between start-ups and staffs. What they cannot do is the decisive thing. The comparative history of interwar innovation that stands beside Rosen's book put the warning in one line: the bureaucratization of innovation, Williamson Murray wrote, guarantees its death. The successes ran through mainstream career structures, not hived-off futurology shops. The tell is not the hub's demo day but its alumni. Where does the successful hub officer go next? If the answer is back to a conventional track, with the detour quietly counted against him, the institution has answered the only question that matters: the new specialty is not a way to become a general.

The mechanism is observable in real time. In June 2024, Ukraine stood up the Unmanned Systems Forces — a separate service branch with its own commander, units and career structure for drone warfare. Whatever its wartime record proves worth, the organizational meaning is unambiguous: in Ukraine, uncrewed systems are now a career. In most Western forces the same specialty remains an additional duty, a short course, a secondment to a hub — a niche without a path. Rosen's framework does not predict which drone tactics will survive contact. It predicts which military will still have its best drone officers in uniform in 2035.

One caution, because the finding is quotable in bad faith. Every officer defending yesterday's doctrine can recite "no maverick wins" against his critics, and the finding will feel like cover. It is not. Rosen's senior innovators were no guardians of the old theory of victory; they were insiders who used the establishment's own power — boards, billets, budgets — to replace it. The test of an institution is not whether it silences its rebels. It is whether its establishment is building paths for the practice its rebels are shouting about. An army that fails that test has not refuted the maverick. It has manufactured him.

The claim breaks if the coming decade produces repeated cases of hub-grown innovation scaling into force-wide doctrine in armies that never open career pathways for the practitioners — or if forces that build such pathways show no advantage over those that keep the new specialty in accelerators. The twentieth-century record contains no case of the first kind; ours could yet produce one. An earlier essay on this site argued that the decisive bottleneck of the coming decade is organizational rather than technical. Rosen's finding names the organ. So the question to put to the next defence-innovation announcement is not how many start-ups the hub has engaged. It is when the first officer whose entire career has been uncrewed systems will pin on a general's stars — and what the majors watching that answer are deciding, this year, about the next twenty years of theirs.

First published on brauer.dev, 31 July 2026.

Der Himmel lebt — the sky is alive. German soldiers training alongside Ukrainian veterans brought the phrase home. It describes a front where twenty-five to fifty drones from both sides hang over every ten kilometres at any given moment. Observation is effectively total for the first three kilometres behind the line of contact, thins out to fifteen, and can be tasked onto anything out to forty. From those measurements the analytic consensus of 2022 and 2023 drew a short conclusion: concealment is finished, and operational surprise with it. Jack Watling of RUSI — the London defence institute whose field teams have measured the Ukrainian front more systematically than anyone — gave the consensus its definitive sentence: "If forces move, they will be detected."

The measurements stand. The absolute sentence built on them has not survived — and the correction came from the author's own institute. By 2025, the same research programme was describing not a law but a gradient: dense here, thinner there, with edges that can be worked. Its fieldwork found that localised gaps in the enemy's reconnaissance coverage can be generated: by rain; by the twice-daily window in which thermal sensors go half-blind; by jamming umbrellas; by hunting the enemy's reconnaissance drones before an offensive starts. A law admits no exceptions. A gradient is an engineering problem. And this one has three known limits: it thins with depth, it stops at the surface of the earth, and it has an inside.

The first essay in this series opened with the deduction that whatever emits will be found — and with its break condition: the transparent battlefield could turn out cluttered rather than clear. That edge is no longer hypothetical. Since the absolute sentence was written, four operations have moved forces through sensor-saturated space without being understood in time to matter.

February 2022 is the strangest case, because Russia achieved surprise underneath total detection. The build-up was photographed from orbit for months, and Washington published its intelligence in advance. The sensors worked completely. Kyiv, Berlin and much of the analytic world filed the imagery under coercion rather than invasion. A battlefield can be perfectly seen and completely misread, and no satellite constellation closes a gap that lives inside organizations.

Two operations in June 2025 started from the far side of the sensors. Ukraine's Spider's Web strike smuggled roughly 150 drones into Russia under commercial cover, staged them for over a year and a half, and flew them against strategic bomber bases from inside the country. Less than two weeks later, Israel achieved total surprise against an Iran expecting attack: strike teams and weapons pre-positioned in-country, fired from inside the defences' own envelopes. Both were surprise by pre-positioning. An operation that begins behind the sensors is not an exception to the gradient but proof that it has an interior.

The manual

The purest counterexample beat the sensors in their home domain — ground manoeuvre, on the war's own front, under the war's own drone density. On 6 August 2024, Ukrainian forces crossed into Kursk oblast, and after eighteen days held some 1,250 square kilometres of Russian territory. The verdict came from the institution with the strongest incentive to believe surprise was dead: thirteen days in, the intelligence arm of the US Army's own doctrine command assessed that Ukraine had "overcome the transparent battlefield".

The mechanics read like a manual. Shrink the knowledge: General Syrskyi, Ukraine's commander-in-chief, planned the offensive face-to-face with a handful of senior officers, leaving no digital traces to harvest; the assault force learned its mission hours before crossing. Poison the interpretation: a disinformation line insisted no counteroffensive was possible before spring 2025, while the 61st Mechanized Brigade moved toward Vovchansk so convincingly that senior Ukrainian officers believed it the real axis. Pick the blind spot: Kursk's reconnaissance coverage, fires and defences were the weakest on the front — conscripts holding a border Moscow trusted its nuclear red line to protect. Then put out the eyes: first-person-view attack drones, the cheap strike weapon of this war, flew instead as air defence against Russia's Orlan reconnaissance drones, while electronic warfare degraded Russian command links. None of it required air superiority, stealth or anything exquisite. The blindness was manufactured from the same cheap components that create the transparency.

The autopsy

The autopsy is the more important half of the story. Ukraine withdrew from Sudzha on 13 March 2025 — seven months, nearly to the day, after the crossing — and the surprise of August had purchased no strategic decision. The salient hung on a single supply road, from Sumy to Sudzha, and Russia strangled it with the fibre-optic FPV: a drone steering down a hair-thin cable, immune to jamming because nothing about it broadcasts. The type was developed as a direct answer to the incursion and fielded against exactly that road. Ukraine reportedly lost a quarter more vehicles than Russia in the sector; a Ukrainian medic supplied the epitaph: "Our logistics just collapsed." North Korean infantry, glide bombs and the spring mud all helped; none of that changes the arc. The operation that undercut the transparent battlefield was ended by the gradient restoring itself over the one road that mattered. Blinding opens a window; it does not furnish the room. Manufactured blindness buys time — the receipt says seven months — and time is all it buys.

Four successes, though, are a numerator without a denominator. Nobody knows how many attempted surprises died in their assembly areas, because failure at that stage looks like nothing happening. The four falsify the absolute; they do not license its opposite. And the gradient reading names its own exits. It breaks if blinding operations of the Kursk class fail serially over the next few years — August 2024 would then stand as the exception that measured the rule. It breaks the other way if one of them ever converts surprise into a decision rather than into time.

Which matters because both absolutes have salesmen. "Everything is seen" underwrites sensor constellations; a ministry quoting the 2022 sentence in a 2027 budget is buying a snapshot of a battlefield actively learning to fog. "Camouflage works again" is the Maginot argument with the sign reversed: permanent cover promised on a few expensive, local, time-limited exceptions — one of them already ground back down. The honest question to put to any defence ministry is a ratio: how much does the budget spend on seeing, and how much on being wrongly seen? Most budgets answer millions against a rounding error — pricing the next war on the assumption that the other side has not read the manual. The sky is alive — but not everywhere, not equally, and not for free. The side that plans on the gradient will keep surprising the side that planned on the law.

First published on brauer.dev, 29 July 2026.

In March 1924, the Air Staff of the Royal Air Force put its theory of knowledge in writing. The Chiefs of Staff Committee had asked for the case behind the RAF's central doctrine — that wars would be won by bombing the enemy's morale directly, rather than by first defeating his forces in the field, as armies and navies always had. The reply, Air Staff Memorandum No. 11A, survives in the National Archives as AIR 20/40, and its decisive sentence reads like a confession drafted by the accused: "The latter alternative is the method which the lessons of military history seem to recommend, but the Air Staff are convinced that the former is the correct one."

Read it slowly. History recommends the other course; we are convinced. Not "the evidence is incomplete". The accumulated record of war was declared to have no jurisdiction over the bomber. A military institution, in peacetime, formally exempted its central belief from empirical test.

The Air Staff did not lack evidence; the evidence sat in its own files. In 1917, the British analyst Lord Tiverton had established from operational records that four out of five bombing squadrons could not find and hit the correct village — not the correct factory: the correct village. The finding was filed and ignored. For the following seventeen years the RAF measured its doctrine in output: sorties flown, tonnage dropped, aircraft procured. Outcome went unmeasured, because the memorandum had ruled measurement unnecessary. The doctrine had hardened from hypothesis into revealed truth, and exercises existed to confirm it.

The bill arrived in 1941. When an outcome measurement was finally forced on the campaign — the Butt Report, compiled from bombing photographs rather than crew reports — it found that of the crews recorded as actually attacking their target, only one in three bombed within five miles of it; over the flak and haze of the Ruhr, one in ten. Five miles. A doctrine that promised to destroy specific factories and break specific societies was flying an instrument that could not reliably hit a city-sized area containing them. And the human denominator: in a single year of the campaign, RAF Bomber Command alone lost 17,479 aircrew dead — expended against a hypothesis their own institution had shielded from evidence for seventeen years.

Call this failure mode the Douhet trap, after the Italian general who gave the bomber gospel its purest theoretical form. The trap is not being wrong — everyone reasoning about a future war is substantially wrong, the successful innovators included. The trap is being unfalsifiably wrong: building the doctrine so that no result short of catastrophic war can disconfirm it. The Air Staff of 1924 was not stupider than the American naval officers who, in the same decade, were war-gaming carrier aviation into existence at Newport. It had organized its institution so that stupidity and insight produced identical internal signals; once that is true, intelligence no longer helps.

Victory is the worst teacher

The trap has a known antidote, and the same period supplies it. In October 1939, the Wehrmacht had just crushed Poland in five weeks. The army high command's response to triumph was to demand after-action reports from every level of command within a week, and the reports show a startling pattern: the higher the headquarters, the more demanding and dissatisfied the assessment. The evaluation named six deficiency areas in a victorious army and converted them into a winter-long retraining programme, with feedback mechanisms to verify the fixes. Victory is the worst teacher, because it validates everything at once — the sound and the rotten together; honest self-examination after winning is therefore the most diagnostic behaviour an institution can display. The Butt Report belongs in the same column from the opposite direction: it is the rare event of falsification imposed on a doctrine from outside, and it took two years of war to impose it.

Three tells

The point of retelling this is not Bomber Command; the pattern is portable. Claims about drones and military AI now occupy the position bomber doctrine occupied in 1924: a visibly potent technology, doctrine unwritten, advocates convinced. Some of today's claims will look like Newport in hindsight; some, like the memorandum. An earlier essay on this site argued that some statements about the next war can be deduced and checked; the Douhet trap is the mirror image — claims engineered so that nothing can check them. The 1924 case yields three tells for spotting the construction before the bill arrives.

First, success is measured in output rather than outcome. Sorties flown, units produced, demonstration videos, systems delivered — the modern descendants of tonnage dropped. The viral statistics of current wars are, more often than not, output numbers. Output is what an institution controls, which is why institutions prefer to report it; outcome is what the enemy gets a vote on. A claim advertised in what the systems do, rather than in what they change, has already chosen the RAF's metric.

Second, failure is metabolized as a dosage problem. When a disappointing result is explained as "too few of them", "not autonomous enough yet", "procurement too slow" — never as evidence against the claim itself — the loop has closed. A doctrine that converts every setback into an argument for more of itself has achieved by reflex what the 1924 memorandum achieved by decree. The US Eighth Air Force ran the pattern in wartime, clinging to the self-defending bomber formation until the disasters of October 1943.

Third, the claim names no condition under which it would be false. This is the cleanest test, because it can be run in a single conversation. Ask the advocate of any drone or military-AI claim what battlefield result would count against it. Three answers recur: none; a change of subject; or a condition that requires the great-power war itself. All three are the 1924 memorandum, restated politely.

A diagnostic built on falsifiability must submit to its own rule, so here is what would break this one. The three tells earn their keep only if they work in advance — and against 1924 they do: using nothing dated after the memorandum, the Air Staff fails all three — seventeen years before the bombing photographs of 1941 delivered the proof. But if doctrines that fail all three tests turned out to win wars about as often as doctrines that pass them, the tells would be a prejudice, not a test. The record of 1918 to 1945 says they are a test. A future record could say otherwise.

One caution, because it is the trap's favourite disguise. None of this says that drones are the new bomber, fated to disappoint; that conclusion would repeat the 1924 error at one remove, with the sign flipped. The bomber did eventually become a formidable weapon. The doctrine failed anyway, because its holders had no way of discovering what their instrument could not yet do at any price below a war. That is the trap's actual content — not a wrong vision, but an unexaminable one. The useful exercise for the decade ahead is to read today's concept papers and keynote claims the way 1941 read the memorandum of 1924, and to ask which of them are being written, right now, in a form no battlefield result could ever touch.

First published on brauer.dev, 27 July 2026.

In 1914, every general staff in Europe planned for a short war. In 1928, Giulio Douhet proved on paper that bomber fleets would break societies within days. In 2022, assessments in Washington reportedly gave Kyiv seventy-two hours. Defence forecasting fails so reliably that failure itself is the field's best-documented finding.

The usual lesson is humility: nobody can know the future of war. It is half right. Prediction fails where it depends on contingency — who fights whom, which invention arrives first. But some statements about the next decade follow from physics that will not change by 2036, from economics no procurement office can repeal, from institutional incentives that survive every election. They can be deduced rather than predicted — and, unlike predictions, deductions show their work.

One rule governs everything below: every claim states what it rests on, how strongly it holds, and what would prove it wrong. Anyone can check the steps; that is the point.

Seven statements

1. Whatever emits will be found, and whatever is found can be killed. Sensors ride civilian cost curves — smartphone cameras, software-defined radios, small satellites — and get cheaper every year; signature reduction is bespoke military engineering and does not. Precision falls in price on the same curves. So the gap between detected and destroyed closes toward minutes, and everything that radiates, moves, or gives off heat lives under permanent risk of engagement. Signature management becomes the first condition of survival. This is not a trend reading but a consequence of cost curves already locked in. What would break it: deception and jamming scaling faster than sensing — a thousand decoys making ten thousand sensors worthless. The battlefield would then turn noisy rather than transparent; the fog of war returns as data garbage, not darkness.

2. Mass returns as industrial mass — and any long war becomes a contest of production rates, not stockpiles. Precision was expensive, so militaries bought few exquisite platforms. Once precision rides civilian components, the calculus flips: many cheap, losable systems beat few costly ones whenever the defender pays more per kill than the attacker pays per unit. Attrition migrates from people to machines, but it stays — and expendable systems empty a peacetime depot in weeks. The measure of military power shifts from what sits in the depot to what leaves the factory each month. The logic holds as long as its premises do. It would fail if cheap defences — guns, jammers, directed energy — flip the cost curve back; then cheap mass was a window, not an era. Or if future wars were reliably short — which nobody can guarantee.

3. Physics keeps the cheap systems tactical; the war gets layered. Battery energy density sits an order of magnitude below liquid fuel, and small airframes trade payload, range, and speed brutally against one another. Cheap small systems therefore dominate the last ten to fifty kilometres; operational and strategic depth remains the province of larger, faster, costlier effectors. There will be no "drone war" but a layered war: saturation at the edge, expensive precision in the depth, and a contested transport problem in between. This one rests on physics, and no energy-density breakthrough is due within the decade. The failure mode is not technical but intellectual: inferring strategic dominance from tactical dominance — the Douhet trap in miniature.

4. The electromagnetic spectrum becomes the precondition domain — the new air superiority. Every uncrewed system depends on a radio link, navigation signals, or both — and electrons are cheaper than airframes: jamming costs structurally less than the thing it jams. No force will operate without first, or simultaneously, winning its patch of spectrum. It is air superiority's old role — invisible, local, re-contested in minutes. The only thing that could repeal this is autonomy severing the dependence entirely — but autonomous systems still must perceive and navigate, and perception can be blinded. The spectrum never loses its value; it relocates it.

5. Autonomy in the final seconds becomes standard — as a forced countermeasure, not a doctrinal choice. If jamming makes remote control unreliable in the terminal phase, two options remain: abort, or let the system decide its last seconds. Terminal target recognition is a narrow, bounded perception problem — precisely the class machine learning solves today. Selection pressure will settle the ethics debate: whoever refuses terminal autonomy loses effect against anyone who jams. This follows necessarily — but for the final seconds only. The failure mode is extrapolating it to "autonomous warfare": between steering an endgame and running an operation lie unsolved classes of problems — context, intent, law under ambiguity.

6. Economics forces layered defence — and the honest metric is cost per prevented damage. A six-figure interceptor against a four-figure weapon is a losing strategy even when every intercept succeeds: each cheap shot drains an expensive magazine. Defence must layer — jamming and deception first, at near-zero marginal cost; guns and cheap interceptors next; the exquisite missile reserved for the exquisite threat. The economic logic is inescapable; whether institutions follow it in time is a separate and much weaker claim — peacetime procurement can ignore economics for years, wartime consumption cannot. You can watch this one fail in practice wherever success is still measured as intercept rate: the metric that vendors of expensive systems prefer.

7. The decisive bottleneck is organizational, not technical. Hardware iterates in years, software in days. Once military effect is software-defined — frequencies, seekers, jamming profiles — every enemy adaptation devalues stocks overnight and is countered within days — faster than any procurement or training cycle. Whoever holds update authority at the front adapts; whoever holds it in ministries loses the duel regardless of technical quality. Peacetime militaries are optimized against this: without the market test of combat, careers and budgets run on proxy metrics — auditability, error avoidance, platform counts — that reward the opposite of delegation and fast failure. That the bottleneck sits here is deducible. Who solves it first is not. The test is not demos or strategy papers but one question: who at the front may change what, without asking permission?

What nobody can know

Four questions cannot be settled from first principles, and anyone claiming a confident answer is usually selling something — a budget, a product, a doctrine, or a book.

Does sensing or deception win the long duel? Both ride the same technology curves; the equilibrium depends on the order of future inventions. Does cheap autonomous mass work against a first-class opponent whose defences are intact? Every data point so far comes from wars between degraded or asymmetric forces. How does escalation behave when autonomous systems interact under the nuclear shadow? Zero data points exist, and the one experiment that would produce them must never run. And how fast do large organizations adapt once their existence is at stake? Peacetime observation measures the wrong variable: bureaucracies ridiculous in peace have transformed in months under existential pressure; agile ones have frozen.

These questions are not a footnote of humility; they are the planning agenda for the decade — not to answer them, which nobody can, but to build forces that survive either answer. The seven statements tell a planner what to stand on; the open questions, what to hedge. Confusing the two lists is how militaries end up, once a century, proving Douhet right on paper and wrong in the field.

First published on brauer.dev, 24 July 2026.