# War on the Rocks: Every Time We Say Revolution, the Change We Want Arrives a Little Later Source: Realpha Blog (blog.getrealpha.com) Original article and charts: https://blog.getrealpha.com/en/blog/waronrocks-2026-10-29-what-s-a-revolution-in-military-affairs-good-for/ > Notes on the 29 October 2026 episode of War on the Rocks, where host Ryan Evans sits down with Mike Horowitz, Ben Jensen and Stacie Pettyjohn to ask whether the phrase revolution in military affairs is worth anything. Horowitz calls the endless definitional fight an empty suit; Jensen prices the cost — another $50 billion thrown at the technology while the doctrine and force-design conversation is skipped buys you conex containers of drones with half-dead batteries. Horowitz also notes the money shortfall for these capabilities has been real. Educational listening notes only — no investment advice, no price targets, and the judgements here may be wrong. Published: 2026-10-30 Locale: en Tags: military innovation, podcast-notes, drones, defence budgets, supply chains ![A dim warehouse with one shipping container standing open, a work lamp picking out boxed small drones in the foreground and a technician holding a swollen battery, identical containers receding in a line towards a far roll-up door with daylight leaking under it](/covers/waronrocks-2026-10-29-what-s-a-revolution-in-military-affairs-good-for-cover.png) > Marquis Wu asked: "What makes an army win?" Wu Qi answered: "Good governance makes it win." The Marquis asked: "Does the number of men not matter?" Wu Qi answered: "If the laws and orders are unclear, if rewards and punishments are not trusted, if the gong sounds and they do not halt, if the drum sounds and they do not advance, then even a million men are of no use." > > —— Wu Qi, *Wuzi*, "Governing the Army" (Warring States period; my own translation) In the 29 October 2026 episode of War on the Rocks, host Ryan Evans puts three scholars who have spent careers on military innovation — Mike Horowitz, Ben Jensen and Stacie Pettyjohn — in one room and asks a question that sounds academic: are we in a military revolution, and does it matter? Horowitz's answer is that the fight over what counts as a revolution is, by definition, an empty suit. Jensen then prices it: the US defence secretary calls for another $50 billion into these technologies, the force-design and doctrine conversation doesn't keep up, and what the money buys is conex containers of drones with half-dead batteries. One boundary first — the target here is "revolution" used as a budget slogan, and Horowitz himself concedes these emerging capabilities have long struggled to get funded at all. ## An empty suit Horowitz starts with why anyone should care. Naval warfare went from Pirates of the Caribbean wooden ships pulling alongside each other and firing cannon, to all-steel battleships, to carriers. There are no battleships today because naval warfare happens beyond line of sight — by the time you are close enough to see the other side, you have probably already been hit. Being able to describe that kind of change is what lets you grasp shifts in power. ![Four horizontal bars show engagement distance across four eras, growing from very short to very long, with the last bar turning dashed and running off the edge of the frame while the enemy dot sits outside the picture entirely.](/figures/naval-engagement-distance-en.svg) Then he turns: every time we talk about revolutions and use that phrase, he thinks it delays the change we want to see. The endless debate over how exactly to define a revolution is, by definition, an empty suit. Pettyjohn's cut on the definition is the sharpest moment in the episode. Evans asks whether a revolution equals technological change plus organisational and operational change plus strategic consequences. She says strip out the strategic consequences, because folding outcomes into the definition conflates whether something is an innovation with how effective it turned out to be. The Maginot Line was an innovation that didn't succeed on the battlefield (Jensen cuts in to say it could have). Mix the two questions and you end up judging the original choice by its result. Her own position: this isn't a revolution yet. To earn the word, the efficacy of fighting and battlefield outcomes have to improve by some magnitude. She isn't sure you must go as far as Krepinevich and say the old ways are completely obsolete, but a small incremental change is just a small innovation. ![A four-quadrant chart with "did it work" on the horizontal axis and "was it innovative" on the vertical axis, the whole right half lit up, and the Maginot Line plotted in the upper-left cell.](/figures/innovation-vs-effect-quadrant-en.svg) ## That archer trained for years The variable Jensen adds is time. He uses Agincourt, the 1415 battle between England and France. The simple story is that the archer won and French knights suffocated in the mud. What it misses is that a longbow is not a crossbow that anyone can pick up — it takes years to train someone with the strength and the aim. The gap between an innovation and a revolution is exactly that stretch of time. ![The top row shows two equally sized blocks for the longbow and the crossbow, while the bottom row's time-to-proficiency bars differ wildly in length, the longbow's bar stretching almost the full width of the chart.](/figures/longbow-training-time-en.svg) People who study military innovation, he says, get more sceptical the deeper into their careers they get, because they realise it is never a function of the technology. It's people and bureaucracy. Achieving anything you'd call a revolution requires wholesale institutional change and even cultural change in how people see, describe, resource and organise war. Clifford Rogers' 1993 punctuated-equilibrium argument still holds up. From which comes a counterintuitive conclusion: you never get the moment of change right. Military change is slow, and battlefield defeat makes it look sudden. What matters is building a profession that can describe change well enough to start asking the next questions — how am I going to fight, what formations do I build, what do I buy. ## The 1990s debate, where everything came true Horowitz recalls the RMA debate of the 1990s: it started with research out of the Office of Net Assessment and similar places, asking how the then-emerging information-age technologies would shape future warfare. That was when everyone first understood what precision warfare would mean, and the beginning of what we now call cyber operations. The famous question was whether satellites and all those surveillance capabilities would lift the fog of war. Here's the part that matters. All of those capabilities actually came to pass and became useful in modern warfare. But the question of whether this was revolutionary distorted the whole conversation and produced odd things — his example is the Marine Corps arguing the boots they were going to issue were revolutionary and should therefore be funded. That is what the label really does: it's a budget lever. Whether a capability arrives and whether it gets called a revolution are two different things. So when I read "X is a revolution" now, my first move is to ask who said it and what they want from whom. Jensen puts it more bluntly: a tech bro packaging a mid-TRL capability as revolutionary is self-defeating for the department and the republic; military professionals asking the second-order question — fine, you want a drone, how do I resource it, what does that formation look like — are already talking about implementation cost and diffusion. ## Half-dead batteries in a container Horowitz has a piece forthcoming with Lauren Kahn and Shira Pindyck that breaks down exactly the "are drones revolutionary" question. They point out that what people call drones is at least four different capabilities: collaborative combat aircraft, the MALE/HALE big brothers like Reaper, tactical surveillance kit, and one-way attack munitions both short and long range. We call all four drones, wrap them, and ask whether drones are revolutionary. For a military trying to work out what to buy, he says he can think of nothing less helpful. ![A scatter plot with range on one axis and unit cost and size on the other, showing four classes of drone scattered far apart, with one large dashed circle enclosing all four points under a single word.](/figures/four-kinds-of-drones-en.svg) Jensen attaches the cost. The defence secretary calls for another $50 billion into these technologies. Throw money at a technology while skipping the doctrine, organisation, training and personnel conversation, with revolutionary rhetoric playing in the background, and it justifies very bad policy decisions. What you get is conex containers full of drones that would have been effective two years ago, now with half-dead batteries, and maybe an operating manual, maybe not. They describe the version that works too. Pettyjohn likes transformation in contact, which General George pursued before he left: get the capabilities into soldiers' hands and let them experiment, because that is how the organisational and conceptual problems surface. Jensen adds a case that has already paid back — a brigade of the 101st on rotation at the Joint Readiness Training Center used drone-dominance initiatives plus AI agents standing in for staff functions and kicked the stuffing out of the opposing force. His summary: push the technology down to people at the lower level, let them iterate fast, pick the winners and scale. Follow that to the end and it can become a revolution. Calling it a revolution before it happens is foolhardy. Horowitz is blunter about the test: funding something is not winning. Winning is when the capability is so scaled and useful that future me says it's old and busted. ![The upper path runs straight into a thick wall with a stack of grey blocks piled up behind it, while the lower path coils into a loop and then fans out to the right as a widening wedge.](/figures/two-paths-for-the-money-en.svg) ## Where this breaks: the money shortfall was real When is this argument wrong? Horowitz produces the counterexample himself, and he's the one who came out of the Pentagon. In the last administration, scrounging for couch-cushion money to fund ten small one-way attack systems was the job. For the universe of things we call drones, plus artificial intelligence, there traditionally has not been enough money. Throwing money at the issue doesn't solve capability development — and money isn't bad. Jensen accepts that and turns on the services. Look at their own budgets for these capabilities against what flows to the new Defense Autonomous Warfare Group (DAWG), and the bureaucratic imbalance is something we'll be digging out of. Pettyjohn feels conflicted about DAWG as well: buying too much of a bunch of stuff that will be obsolete soon and probably doesn't work on a contested battlefield is waste, yet unless people start using it the organisational and conceptual work never happens, and small drones are potentially the most transformative thing for ground forces. On top of which the services' organisational culture sits there — Carl Builder wrote about this — favouring their big weapon systems, and new technologies challenge those preferences. So the boundary is this: if the real binding constraint is money and bureaucratic inertia, loud rhetoric is the crowbar, and attacking it misses the target. To tell which case you're in, look at whether the money connects to user iteration — is anyone using the thing, and is anyone folding what they learn back into formations and doctrine. ## Why not the other explanation If the word is useless, how should the change be described? Evans offers his version: what's happening is the industrialisation of precision capabilities, and their spread to people who previously couldn't have them. Horowitz says this is the matured version of the 1990s revolution. Satellite imagery, satellite communications — things that used to need a rich country with a sophisticated military — are now cheap, largely commercial and available to anyone. His term is precise mass: commercial manufacturing, advances in AI and autonomy, plus the diffusion of precision guidance, a fifty-year-old technology at this point, which is how the Houthis fire what he calls flying lawnmowers at American ships in the Red Sea. Pettyjohn prefers affordable mass, because the weight falls on quantity and producibility — the key move goes beyond buying lots of drones that are obsolete within twelve months, to being able to produce lots of them very quickly, on timelines the other weapons the US military uses simply can't meet. Horowitz pulls the thread back through history. The American military buys capabilities it plans to field for fifty years, which only works if you are so far ahead that routine upgrades are good enough. Using a $4 million Patriot to shoot down a $50,000 Shahed is not a sustainable cost exchange ratio. Historically militaries did not buy capabilities to last fifty years; that's an artifact of the post-Cold War period. Your drone going obsolete in a year or two only matters if you planned to keep it for a decade. If the model is twelve months, you pay as if it's twelve months and your factories are set up for the next generation, that's simply a different business model. Which answers the question readers usually have about anything with a twelve-month shelf life: start from how many years you intend to hold it, then check how many years the seller priced it for. When those two don't match, the arithmetic is wrong. ![Two cost bars sit side by side: the interceptor's bar spans the whole width of the chart while the incoming weapon's bar is just one thin sliver, a ratio of eighty to one.](/figures/cost-exchange-ratio-en.svg) Jensen points the bottleneck further upstream: the change he wants is an acceleration in the conversion capacity of the US economy. He's been on several government and DARPA projects asking whether there's a widget you can insert that suddenly explodes American industrial capacity. His conclusion is that this isn't a technology problem; what has to change first is policy incentives and regulation. When money goes in and capacity doesn't come out, that's usually the layer the answer is hiding in. ![On one shared year axis, the top row is a single bar spanning fifty years, the middle row is a small block only twelve months wide, and the bottom row is a chain of twelve-month blocks running end to end.](/figures/fifty-years-versus-twelve-months-en.svg) ## A few other things worth keeping - Punishment campaigns are back. Jensen says their firepower strike tracker shows drones aren't the problem — even with the jet-powered Shahed, the intercept rate hasn't moved significantly. The problem is ballistic missiles, an old technology whose production and rocket-engine costs have fallen, making cheap long-range precision strike on infrastructure worth doing. Historically punishment campaigns don't achieve the political aims of a war, and countries run them anyway. - Horowitz's PhD student Natalya Henry has a paper on drone use by rebels in Burma: it transformed their effectiveness against the Burmese military, until China shut the border, parts dried up and the advantage collapsed. Commercial manufacturing lets more actors produce useful military capability, and it also generates more proxy dynamics. - AI only comes up near the end. Horowitz cites the Maven smart system raising the number of targets that can be identified and struck per day, and says open-weight models will diffuse that to everyone. Jensen plays sceptic: watching the US use AI looks to him like Wile E. Coyote and Acme rockets, and the opportunity he sees is in poisoning adversary models. - Jensen read Horowitz's book on a cot in Helmand, and tells him to his face that it's really good for getting some rest at night. ## Worth a look - The 29 October 2026 episode of War on the Rocks, the source for this piece - Mike Horowitz's book on the diffusion of military innovation, the one Ben Jensen mentions on air - The forthcoming piece by Horowitz, Lauren Kahn and Shira Pindyck breaking drones into four capability types - Natalya Henry's paper on drone use by rebels in Burma - Carl Builder's classic work on the organisational cultures of the US services ## The one thing to take with you **The part that takes time is never the new thing itself.** The longbow didn't win on the bow; it won on arms trained for years. Everything argued about in the 1990s did arrive, and what the arriving cost was twenty years of formations, doctrine and people. The moment something gets called a revolution, that stretch is what gets skipped — and that stretch is what decides whether it ends up useful. ![A small block on the left marks new technology and a small block on the right marks usefulness, while a thick band between them — far wider than either end — is split into five cells for organization, doctrine, training, people and culture, and a dashed arc leaps straight from left to right over the whole band.](/figures/the-skipped-middle-en.svg) Here's something I've tried. Pick something you've acquired recently that you think will change your life — an app, a piece of equipment, a course. Write down three things: which part of my routine has to change for this to earn its place, who else has to change with me, and how long from now I'll be able to tell whether it's gathering dust or actually plugged in. Put that date in the calendar and come back to it.