One would think that under such attacks on infrastructure any society must necessarily collapse. Or at least that’s what Putin hopes for. But the last time I’ve checked, Ukraine was still very much alive and kicking. The question is: how is that possible?
Partly because it's gotten total aid of ~2x its annual GDP in ~4 years. (Which has led to eg a Europe-wide transformer shortage from replacing Ukrainian transformers destroyed by Shaheds.)
I haven't gotten very far into it yet, but Kevin Carson's The Homebrew Industrial Revolution explores this question in some detail (though in a fairly opinionated way).
Decentralize or Die.
In April 2024, a salvo of cruise missiles destroyed the Trypilska thermal power plant, the largest in the Kyiv region, in under an hour. In June 2023, the destruction of the Kakhovka dam left a million people without drinking water and wiped out an entire irrigation system downstream. Throughout three winters, strikes on combined heat and power plants have left apartment buildings in Kyiv at indoor temperatures barely above freezing. In December 2023, a single cyberattack on Kyivstar, Ukraine's largest mobile operator, cut phone and internet service for millions.
One would think that under such attacks on infrastructure any society must necessarily collapse. Or at least that’s what Putin hopes for. But the last time I’ve checked, Ukraine was still very much alive and kicking. The question is: how is that possible?
***
In winter 2022, when the blackout in Kyiv happened for the first time, people had to cope for themselves. Here’s Tymofiy Mylovanov, professor at Kyiv School of Economics, tweeting in real-time:
Later that day he dryly comments: “In the morning I said I was not worried about heating. Instead, I was concerned about water and sanitation. Boy, was I wrong.”
It’s worth reading the tweets from the next few days: Getting a generator, setting it up, placing it on balcony so that fumes stay outside, getting the wires in without letting the cold in as well. Go check it out for yourself.
Anyway, what followed was a series of adaptations, a kind of military vs. civilian arms race. Through the first winter, the strategy was simply to repair what Russia destroyed. Substations and transformers that could be replaced within weeks with donated European spares.
In the meantime, for the millions of affected people, the government created stopgaps. Over 10,000 heated public spaces in schools, government buildings, and railway stations offered electricity, water, internet, and phone charging. Kyiv deployed mobile boiler houses that could run for days without refueling. Hospitals installed Tesla Powerwalls. Cafes ran diesel generators and became de facto community centers.
Mobile boiler house in a shipping container. You truck one in, connect it to a building's existing heating pipes, and it starts working.
I’ve donated to some of those efforts, maybe you did too. And taken all together, it worked. Kind of. But by 2024 Russia adapted. Strikes shifted from repairable transmission equipment to the power plants themselves, assets that take years to rebuild. The Trypilska plant was partially restored after its destruction, then it was struck again by drones months later. And again after that. With two-thirds of generation capacity gone and every thermal plant in the country damaged, it became clear that restoring the old centralized system was not a viable strategy.
Ukraine's response shifted. It was not to rebuild what was destroyed but to replace it with something less centralized. Something too dispersed to target. Instead of restoring the Trypilska plant's 1,800 megawatts, hundreds of small cogeneration units were scattered across the region, compact gas turbines producing 5 to 40 megawatts each, generating heat alongside the electricity. By late 2025, Ukraine's heating sector alone ran over 180 such units as well as hundreds of modular boilers. Hospitals, water utilities, and apartment blocks are organized into autonomous energy islands, microgrids that keep functioning even if the national grid goes dark. No single unit is worth a cruise missile. And a destroyed module can be replaced with a phone call and a truck from Poland.
The same logic extends to water. Ukraine's centralized water systems are inherited from the Soviet era. A single pumping station serves hundreds of thousands of people. They are just as vulnerable as the power plants. Strikes on the grid cut electricity to pumps. Without pumps, water stops flowing. In winter, standing water in pipes freezes and bursts them, cascading damage across entire districts.
In Mykolaiv, a damaged pipeline to the Dnipro River left 300,000 residents relying on salty, barely drinkable water from a local estuary for over a year. The response mirrors the energy transformation: water utilities are installing their own solar panels and battery storage to decouple from the grid entirely.
Solar panels are, under these circumstances, close to an ideal solution. They are cheap, manufactured at scale, and can be replaced in a single day. By early 2024, Ukrainian households and businesses had installed nearly 1,500 megawatts of rooftop solar. Not because of climate change, but because of survival. Solar panels are inherently dispersed. There is no single set of coordinates an attacker can hit to disable them all. And destroying them one by one would cost the attacker more in munitions than the panels are worth.
This kind of arithmetic pops up everywhere. In the ongoing Iran war, Ukrainian military observers were flabbergasted by Gulf states and the US burning through hundreds of Patriot missiles, $4 million each, to shoot down cheap Iranian Shaheed drones, $35,000 apiece. If destroying a target costs more than the target itself, the attacker loses even if the strike succeeds.
A different kind of decentralization is happening in the telecommunications domain. The infrastructure was already fairly decentralized to start with, a legacy of makeshift internet adoption that happened in many Ostblock countries, with many small ISPs emerging independently. The war pushed this further. Ukraine has adopted a layered backup approach: if fiber broadband fails, mobile networks fill the gap; if mobile networks are knocked out, Starlink steps in as a last resort.
The logic extends to government services. There’s the Trembita data exchange platform, where government services talk each other directly without centralizing the data. (Trembita is based on Estonian X-Road system — the birth of Estonian e-gov technology is a fascinating story in itself, and there’s a whole book about it!) Built on top of it, there’s the Diia app that allows citizens to file taxes, register vehicles, access medical records, open bank accounts, register births, and start businesses, all from a smartphone. This, of course, means there’s no single office building to target so as to disrupt a particular kind of activity.
Add to that Ukrainian governmental data are now stored in the cloud. A week before the invasion, Ukraine's parliament quietly amended a law that had required government data to be stored physically in Ukraine. On the day the missiles started flying, the Ukrainian ambassador in London met AWS engineers and decided to fly three AWS Snowballs, hardened suitcases that hold 80 terabytes each, from Dublin to Poland and then move them to Ukraine the very next day. Ukrainian technicians copied population registers, land ownership records, and tax databases onto them and shipped them back out.
It was a race. On the day of the invasion, cruise missiles struck government server facilities while Russian cyber operatives simultaneously deployed wiper malware, software designed to permanently destroy data, against hundreds of Ukrainian government systems. Some data was lost, but the most critical registries were already gone, smuggled out of the country in carry-on luggage.
***
On the battlefield, where all these trends are even more severe, concentration has become suicidal. Russian infantry now advances in groups of two or three. Anything larger is an invitation for a drone strike. Warships are floating targets. Russia's Black Sea Fleet retreated from Crimea after losing vessels to cheap unmanned boats. In the Hedgehog 2025 exercise in Estonia, a small team of Ukrainians and Estonians with drones, acting as the opposing force, wiped out two NATO battalions, thousands of soldiers, in half a day, not least because they had moved in columns, parked their vehicles in close formations and failed to scatter under attack.
They made the same mistake as the designers of Soviet-era power grids: they concentrated value and got destroyed for it. Call it the blast radius principle. In a war of attrition, any asset whose destruction is worth more than the cost of the weapon that can reach it will, sooner or later, be destroyed. The only effective strategy is to push the value of each individual target below that threshold, to become, in effect, too small to bomb.
When Rheinmetall’s CEO recently made a condescending comment about Ukrainian housewives 3D-printing drones in their kitchens, much merriment ensued. Because Rheinmetall, of course, builds the very kind of heavy conventional, WWII-style hardware that the developments in Ukraine are rapidly making obsolete.
But mockery aside for a moment: if you’ve spent any time around progress studies, the phrase “housewives building drones in kitchens” makes you prick up your ears. It triggers a specific association: cottage industry, the small-scale, home-based production that preceded and enabled the industrial revolution. It makes you think about how the modes of production change over centuries.
You know that kings and generals don’t make history. One empire falls, another rises, nothing fundamentally changes. What does matter is new technology. Even more so new technology which fundamentally changes how things are done. Technology that reshapes the economics of entire production chains. Agriculture. Road system. Bill of exchange. Putting-out manufacture. Joint-stock company. Assembly line. The humble shipping container…
Does decentralization, as seen in Ukraine, fit the bill? We don’t know. FirePoint, the Ukrainian company producing the much-spoken-about FP drones, is distributed across more than 50 manufacturing sites throughout the country. But that’s nothing new. The allied bombing campaign during WWII failed to halt German aircraft manufacture precisely because Germany had decentralized its industries. Albert Speer, then the minister of armaments, dispersed production into hundreds of small workshops, caves, tunnels, and forest sites across the Reich. German aircraft production actually increased in 1944, the year of the heaviest bombing. But then, after the war, German industry did concentrate again.
What seems different this time, though, is the spillover into the civilian sector. Speer dispersed munitions factories, but German civilians kept heating their homes the same way throughout the war. In Ukraine, the dispersal extends to utilities, water systems, telecommunications, government services. Russians bomb a heating plant, the heating network disperses into dozens of autonomous microgrids.
The obvious objection is that this is a wartime hack, not a permanent transformation. Distributed systems sacrifice economies of scale. A hundred small gas turbines are less efficient than one large power plant. Once the war ends and the skies are safe, the economic logic will reassert itself and everything will concentrate again.
And indeed, in some cases, that's exactly what will happen. Ukraine is currently bombing Russian oil refineries and fertilizer plants, and although cracking crude oil in plastic bottles in a kitchen is exactly the sort of thing you might expect Eastern Europeans to do, it's unlikely to match the efficiency of a proper refinery. Some industries have genuinely irreducible physical economies of scale. The chemistry demands large reaction vessels, the thermodynamics reward concentration. Similarly, some infrastructure simply cannot be distributed. It's hard to imagine a decentralized railway system or a dispersed deep-water port — at least short of giving up on it and transporting everything by drone.
But not all economies of scale require spatial proximity. Sometimes, it’s just sheer scale that matters, not necessarily the co-location. Case in point: solar panels. Other times the crucial element is the organizational structure, not the physical location of the employees. Basically any service offered over internet is like that.
But all that being said, there’s a specific reason to think some of these changes may stick.
Over the past fifty years we’ve accumulated an entire arsenal of distributed technologies. Packet-switched networks. Drones. Solar panels. Distributed databases. 3D printing. Even nerdy cypherpunk inventions like public key cryptography, zero-knowledge proofs and cryptographic ledgers. And it’s not just technical stuff. We’ve developed distributed social technologies too: open-source-style cooperation (who would have predicted that military intelligence, of all things, would be the next domain to go open-source?), market design, remote work, video conferencing. Even prediction markets as a tool for aggregating dispersed knowledge.
Some of these are already ubiquitous. Around 70% of the world’s population already has access to the Internet, a network famously designed to route around damage during a nuclear war. But others feel like we’re barely scratching the surface. 3D printing has existed for decades, yet it still feels like a technology that we are only playing with. We may be like pre-Columbian Americans, whose children played with wheeled toys, but the adults carried loads on their backs.
Mesoamerican wheeled toy.
Based on historical examples, we know that inventing a technology is often not the bottleneck. The aeolipile was invented in the first century AD, but we still had to wait another seventeen centuries to get an actual steam engine. Gutenberg went bankrupt. Adopting a technology is dependent on complex interplay of socio-economic forces that, at a certain moment, make the technology so desirable that people start using it despite all the drawbacks and overcoming all the vested interests. Then the learning curves kick in.
Two questions remain. Are those distributed technologies already adequately exploited, or are they like dead wood lying around in a forest, waiting for a spark? And if the latter is true, are the incentives created by the war in Ukraine — or for that matter, by similar future war elsewhere — sufficient to ignite it? They may be. Because once the enemy starts bombing companies, the incentives change. Working from home ceases to be a nice perk. Suddenly, it’s either work from home or die.