Kris De Decker · blog Low←Tech Magazine (founded 2007 and first hosted on TypePad; since 2018 the main site has been the self-hosted, solar-powered solar.lowtechmagazine.com, which goes offline after long stretches of cloudy weather) · Belgian; about ten years as a science and technology journalist for Belgian media, then moved to Spain in 2007 and started the magazine; lives in Barcelona · Still active: latest post 24 September 2026 · Corpus: 183 entries in the English archive (April 2007 to September 2026), about 170 of them written by him; most articles run three to five thousand words; the busiest year was 2008 (30 posts), then 19 in 2009 and 4 to 12 a year after that · Two turning points: in September 2018 the solar-powered site went live and became part of the argument; from 2022 he shifted toward thematic book collections and hands-on build guides (bike generators, a small solar system, a heated table, a solar oven) · How posts were chosen: there is no "best of" page, and the closest thing to the author's own selection is the thematic book series; this page fetched all 182 article bodies and counted internal links, and once book announcements are set aside the most-linked article is the 2020 piece on biomass (28 links), followed by the 2024 public bathhouse, the 2016 office essay and the 2020 old-laptop piece · This page draws on 16 posts (four of them read in their main parts), from December 2007 to September 2024: four from the earliest fifth of the blog's span (April 2007 to March 2011), and posts from both sides of the 2018 turning point
In January 2020 a small website published its annual check-up. It runs on a balcony near Barcelona: one 50-watt solar panel, one lead-acid battery and a tiny computer that draws a watt or two. After a few cloudy days in a row the battery runs flat and the site goes offline. That year it was online 95.26% of the time.
The report had a stranger finding too. More than half of the server's traffic for the year (6.63 of 11.16 terabytes) came from one broken RSS reader that fetched the feed every few minutes. Once that was fixed, the server's draw fell from 1.14 watts to 0.95. A saving of 0.19 watts sounds pathetic, but over a day it adds up to two and a half extra hours online when the sky is grey.
Then the author did the carbon accounting. Spread the energy it took to make the panel, battery, charge controller and server across each year of use, and the site emits about 9 kilograms of CO2 a year. Had he done nothing clever and plugged the server into the Spanish grid, it would have emitted 1.54 kilograms. He printed that line exactly as it came out:
This seems to indicate that our solar powered server is a bad idea, because even the smallest solar panel with the smallest battery generates more carbon emissions than grid power.How Sustainable is a Solar Powered Website?, 2020-01-28
Then he pointed out that this compares apples with oranges. How is the grid's "carbon intensity" calculated? The emissions from building its wind turbines, solar panels and power lines are counted as zero. By the same rule, his server would score zero as well.
That is what Kris De Decker has spent the better part of two decades doing: finding the columns in the ledger that have quietly been set to zero, and filling them in. He was a Belgian science and technology journalist who moved to Spain in 2007 and started an online magazine, Low←Tech Magazine, which "refuses to assume that every problem has a high-tech solution."
The first column he filled in was efficiency. In 2008, Madrid's Christmas lights grew far more lavish than the year before. A reporter asked the shopkeepers' association why, and was told that LEDs use much less energy, so they could put up many more lights without using more energy. De Decker saw a rule in that: the energy that efficiency saves gets eaten by new uses. When a building is wrapped in 200,000 LEDs, they don't replace incandescent bulbs; they replace a wall that used to be dark. Later, drawing on the British sociologist Elizabeth Shove, he pushed the rule further. Efficiency policy can only compare like with like: tumble dryer against tumble dryer, never against a clothesline; jet against the jet of 1985, never against the propeller planes of the 1950s, which burned about as much fuel per passenger. From that came the central sentence of his whole site:
To focus on energy efficiency is to make present ways of life non-negotiable.Bedazzled by Energy Efficiency, 2018-01-09
The second column was reliability. Today's grid aims to meet demand every second of the year, but wind and sun don't take orders. Making them do so means building several times more generators, power lines and storage, and building all that burns fossil fuel too. He dug up the pre-industrial answer. Windmills and sailing ships never tried to run around the clock: they worked when the wind blew and rested when it didn't. When an English miller was scolded for working on a Sunday, he said that if the Lord was good enough to send him wind on a Sunday, he was going to use it.
From there he reached a conclusion that sounds absurd: to improve energy security, make infrastructure less reliable. A society that never loses power hands its laundry, cold storage, doors and navigation over to electricity and forgets how to do those things without it, so it is the most fragile when the power does fail. People living off the grid have less power and panic less.
His balcony is the laboratory for that conclusion. The website goes offline, and he says openly that this is by design: past a certain level of uptime, the fossil energy spent on extra batteries outweighs what the solar panel saves.
You may not agree with him, and he admits that asking people to settle for "enough" is controversial and could slide into coercion. But after reading him it is hard to read a phrase like "30% energy savings" the old way. You'll ask first: compared with what, and which column was set to zero?
In October 2008, in Low←Tech Magazine's second year, De Decker wrote a post with a flippant title: "Viva Las Vegas." At the time incandescent bulbs were under attack because nine-tenths of their electricity turns into heat, and everyone was waiting for LEDs to take over. Instead he collected new buildings whose facades were covered in coloured LEDs, some with more than 200,000 of them. Beijing had a 2,200-square-metre LED screen that ran on solar panels and was marketed as "a radical example of sustainable technology."
Every project's publicity stressed how efficient the LEDs were. He asked the question nobody was asking: efficient compared with what?
The LED lights attached to facades (sometimes more than 200,000 of them) are not substituting incandescent light bulbs, they substitute a non-lit façade.Viva Las Vegas: LEDs and the energy efficiency paradox, 2008-10-14
The best evidence in the piece wasn't his. The Spanish newspaper El País found it for him. The previous winter, Madrid's streets carried far more Christmas lights than in earlier years. When the paper asked the shopkeepers' association why, the answer was: "LEDs consume much less energy, so we can use much more lights without consuming more energy."
The sentence is said with confidence, and the arithmetic is correct. A large part of De Decker's writing over the next two decades asks why that sentence is correct and still gets things badly wrong.
Kris De Decker is Belgian. He worked for about ten years as a science and technology journalist for Belgian media, moved to Spain in 2007 and founded Low←Tech Magazine that year. A small team now shares the design, translation and book production, but about 170 of the 183 archive entries are his. In one post he describes his working kit: a second-hand 2006 ThinkPad X60s bought online for 50 euros, running Linux Lite, a system made for old computers. Another detail fits him even better. The website's server and his living-room lights originally ran on separate solar setups. While the server had the 50-watt panel, the lights ran on a 10-watt panel and he was "often left sitting in the dark"; when the panels were swapped, the room got light and the website went down more often. The blog is still updated. Since 2022 he has written less and built more, and most recent posts are illustrated guides to things he made himself.
He ran into his root question in his first year: why does total energy use keep rising when technology keeps getting more efficient?
The first clue was the LEDs. The second came in 2009, when he went looking for the embodied energy of electronics and found a result that surprised even him. Making one 2-gram memory chip burned 1.6 kilograms of fuel. By his estimate, the memory chips in a laptop alone held more embodied energy than the laptop drew from the wall over its three-year life. In other words, the most energy-hungry stage of an electronic product happens before you buy it, and every report on "appliance energy use" counts only what comes after you plug it in.
The embodied energy of the memory chip alone already exceeds the energy consumption of a laptop during its life expectancy of 3 years.The Monster Footprint of Digital Technology, 2009-06-16
Put those two findings together and the shape of the problem appears. Energy is treated as a supply problem: where the power comes from, whether it can be made clean, whether it can be delivered more efficiently. Demand is treated as a given that only grows, like the weather. De Decker has spent the years since arguing that demand is not weather. It is made. Every "more efficient" technology, every grid that never goes down, every new "normal" standard of living quietly pushes demand up, and the standard accounting happens to score that push as zero.
His method is to look for a point of comparison in the past. Not because the past was better, but because people then had no choice: no fossil fuels, no grid, no batteries. How they heated homes, moved goods and ran workshops was design forced on them by constraints. Set those designs beside today's and the "why" becomes visible.
In the 2008 LED piece his conclusion was still mild: efficiency isn't bad in itself, it just doesn't save energy automatically and needs a carbon tax or higher energy prices to go with it. At that point he put the problem down to the rebound effect: once things use less power, people use more of them.
By 2018, in "Bedazzled by Energy Efficiency," his position had changed at the root. He started with the EU's report card. Between 1998 and 2012, fridges and freezers became 75% more efficient, washing machines 63%, tumble dryers 72% and dishwashers 50%. Yet total energy use in the 28 EU countries in 2015 was only slightly below its 2000 level, and even that small drop could be explained by the 2007 economic crisis; when growth returned, so did energy use.
Then he looked at what the EU's "20% energy savings" actually meant, and found that it wasn't measured against today at all. It was measured against a projected future. The EU said efficiency gains were "roughly equivalent to turning off 400 power stations," but not one power station would be switched off in 2020. The claim meant that without the gains, 400 more would have had to be built. The International Energy Agency went further and called this "avoided energy" the "first fuel," larger than oil, gas or coal. De Decker's verdict was dry: this measures a fuel that doesn't exist. And the higher the projected future use, the more energy gets "avoided"; if demand ever actually fell, the number would turn negative.
What really changed his mind was a paper by Elizabeth Shove. To calculate efficiency, she points out, you first have to declare that two things provide the same service, and once you've defined the service, you have also fixed the starting point of the comparison. De Decker's example is aviation. In 1985 an airliner burned 8 litres of fuel to carry one passenger 100 km; today it burns 3.7, so we say planes have become twice as efficient. Move the starting point back to 1950 and the picture changes. In the 1960s jets replaced propeller planes; they were twice as fast and at first burned twice as much fuel. It took fifty years for the jet to get back to the fuel use of the last propeller planes of the 1950s.
If viewed in a larger historical context, the concept of energy efficiency completely disintegrates.Bedazzled by Energy Efficiency, 2018-01-09
Should propeller planes count? That depends on whether you define the service as "flying" or as "flying at around 1,000 km/h," and the second definition already assumes the more energy-hungry service. In the same way, nobody counts the dumbphones of ten years earlier when they work out the efficiency of smartphones. The EU calculated that energy labels for tumble dryers would save up to 3.3 terawatt-hours by 2020, the annual consumption of Malta, but never calculated what clotheslines would save. Clotheslines and bicycles fall outside the comparison because they don't provide "the same service." So the more successful efficiency policy is, the more firmly cars and tumble dryers become "normal," and the further their alternatives are pushed to the margins.
That is what his central sentence means: to focus on energy efficiency is to make present ways of life non-negotiable. He also quotes Shove's own line: efficiency policy is a problem not because it doesn't work, but precisely because it does. It sustains, perhaps escalates, but never undermines an increasingly energy-intensive way of life.
He followed the logic into a deliberately provocative thought experiment. What about the opposite, an "energy inefficiency policy"? Put 1960s engines into today's SUVs and most people could no longer afford to drive them; they would switch to smaller cars or drive less. Mandate inefficient boilers and big houses would become too expensive to heat; people would heat one room and dress warmly. He even found a German study that backs the idea. Comparing the calculated energy ratings of 3,400 homes with their measured use, it found that the most efficient homes used 30% more energy than predicted, while the least efficient used 60% less.
Making everything less energy efficient would reverse the growth in energy services and reduce energy demand.Bedazzled by Energy Efficiency, 2018-01-09
That isn't his real proposal. What he actually argues for is sufficiency: less light, less travel, cooler rooms, smaller homes, or substitutes such as a bicycle for a car, a clothesline for a tumble dryer, thermal underwear for central heating. Efficiency can stay, but as a means rather than a goal: an efficient boiler heating just one room to 16 degrees, an efficient engine in a far lighter vehicle.
Two years later the idea arrived on his own desk. In a 2020 piece on laptops he cited a study that took apart 11 laptops of similar size made between 1999 and 2008. The weight and materials of the battery, motherboard, hard drive and memory had barely changed in a decade: manufacturing had become more efficient, but every gain was absorbed by more functionality. Newer laptops use less power per unit of computing, but they also do more computing. He wrote that "Jevon's paradox is nowhere as evident as it is in computing." His response was to buy a 2006 ThinkPad for 50 euros, while admitting that this was "a hack, not an economical model."
The second idea is the flip side of the first. Part of the reason efficiency looks effective is that some things never enter the ledger at all.
The 2009 chip calculation was the first time he showed this. He cited figures from Timothy Gutowski at MIT: century-old manufacturing processes such as injection moulding, casting and machining use 1 to 10 megajoules of electricity per kilogram of material, while semiconductor and nanomaterial processes use up to six orders of magnitude more. That led to a counter-intuitive conclusion: recycling can't rescue chips. For conventional products most of the energy goes into the raw material, so recycled material saves a lot. For chips nearly all of it goes into the processing itself, so swapping in recycled material barely helps.
Recycling is not a solution for energy consumption if all your energy use is concentrated in the process itself.The Monster Footprint of Digital Technology, 2009-06-16
In 2017 he applied the same ruler to the whole power grid. People like to say Europe's wind potential is ten times its annual electricity demand and America's solar potential a hundred times. Those, he said, are annual averages, and a grid has to match supply and demand at every moment. In London a solar panel produces 65 times less at 10 a.m. on a heavily overcast December day than at noon on a sunny June day; when wind speed halves, a turbine's output falls to an eighth. There are only four ways to make such sources meet demand around the clock: keep fossil plants as backup, overbuild for the worst case, link distant regions with long power lines, or store energy. He worked through them one by one. According to the research he cites, a European grid running on wind and sun alone, with no balancing plants, would need 400 terawatt-hours of storage. Converting every passenger car in Europe to an electric car with a 30 kWh battery would give 7.5 terawatt-hours in total. The batteries would need to be sixty times larger.
His conclusion: counting only the energy payback time of each panel or turbine greatly overstates how sustainable a renewable grid is. Everything built to keep supply and demand matched at all times also takes fossil energy to make, and that column is usually entered as zero.
In 2020 he turned the ruler on himself. The whole second half of the website check-up is an embodied-energy audit of the balcony setup: 1.03 megajoules per watt-hour of lead-acid battery capacity, 3,514 megajoules per square metre of solar panel, with batteries spread over 5 years and panels over 25. He also tabulated the trade-off between battery size and uptime. To survive four cloudy days without going offline would take a 440 watt-hour battery, the size of a car battery, and over 70% more embodied energy than the configuration he actually ran. He arrived at a number that looked bad for him, then showed that it exposed a flaw in how the grid is measured, not a flaw in his site:
When the carbon intensity of the power grid is measured, the embodied energy of the renewable power infrastructure is taken to be zero.How Sustainable is a Solar Powered Website?, 2020-01-28
For a coal-fired grid, ignoring construction emissions does little harm, because the fuel burned dwarfs what went into building the plants. With renewables it's the other way round: almost no emissions in operation, nearly all of them in construction. The same accounting convention, applied to a different energy mix, turns from harmless into a systematic bias.
If supply can't keep up with demand, let demand follow supply. In September 2017 he published two articles in a row, the second called "How to Run the Economy on the Weather." He started by asking how pre-industrial people coped with wind and water that came and went. On the eve of the Industrial Revolution, Europe had an estimated 200,000 windmills and 500,000 watermills sawing wood, making paper, polishing glass, pressing oil and making gunpowder. With no batteries and no grid, they relied mainly on a strategy we have almost forgotten: work when the wind blows, rest when it doesn't.
From a 1957 book on windmills he copied out a miller's schedule. When the wind blew in autumn, the miller might work from Sunday midnight to Tuesday evening, from Wednesday morning to Thursday night and from Friday morning to Saturday midnight, with only snatches of sleep. A good miller woke up in bed when the wind rose and got up in the middle of the night to start the mill, "because the wind was his taskmaster." When a Mr. Wade of Wicklewood in Norfolk was criticised for working on a Sunday, he replied:
If the Lord is good enough to send me wind on a Sunday, I’m going to use itQuoted by De Decker from a 1957 book on windmills, in How to Run the Economy on the Weather, 2017-09-21
And when there was no wind? Another miller would "sit on the fan stage of a fine evening and play his fiddle." Sailing ships worked the same way: no wind, stay ashore and repair the ship; at sea, plan voyages around the seasonal winds and currents.
Then he carried the strategy into the present. Sawing, drilling, grinding and pressing can run on an intermittent supply; it changes their speed, not their quality. Freight trains already run at night, which is exactly when wind is strongest and demand lowest. He spelled out the cost too:
If I want to buy new shoes, I might have to wait for the right season to get them manufactured and delivered.How to Run the Economy on the Weather, 2017-09-21
He also spotted a neat reversal: instead of storing energy to make products later, make products whenever energy is available and let the products themselves be the storage, kept in warehouses and sold in the lean season.
A year later, in December 2018, he pushed the idea to its most jarring conclusion. He looked up definitions of "energy security" and found that more than forty of them share one premise: supply must meet demand without interruption. By that standard Germany's grid, with power available 99.996% of the time, is extremely secure.
But he noticed a counter-example. People who design off-grid solar systems usually don't size them for zero outages; they accept a set number of days a year without power. A calculation he cites for an off-grid house in Spain shows that lowering reliability from 99.75% to 99.00% cuts costs by 60%, while outages rise from 22 to 87.6 hours a year. By the official definition that is a failure. Yet researchers who visited about a hundred voluntary off-grid households in Canada found that they didn't feel insecure: on cloudy days they used less power, they ran the washing machine only when the batteries were full, and appliances they couldn't power they simply didn't buy.
And a society that never loses power? Drawing on the sociologists Olivier Coutard and Elizabeth Shove, he argued that an unlimited, uninterrupted supply turns washing machines, air conditioners, refrigerators, automatic doors and round-the-clock mobile internet into "normal," while washing by hand, storing food without electricity, keeping cool without air conditioning and finding your way without a phone slowly disappear. The more reliable the infrastructure, the more society depends on it, and the more fragile it is when the power fails.
To improve energy security, we need to make infrastructures less reliable.Keeping Some of the Lights On: Redefining Energy Security, 2018-12-09
His alternative measures security in services rather than kilowatt-hours. People don't need electricity as such; they need to store food, wash clothes, open and close doors, reach each other and see in the dark. Fridges could be insulated far better so they ride out an outage; cold cellars, fermentation and daily shopping nearby all work without a cold chain. The slogan should change too, from "keeping the lights on" to "keeping some of the lights on," or "which lights should we turn off next?"
Nine months earlier he had already turned the argument into a working object: the solar-powered website launched in September 2018. At launch he wrote that it would go offline, and that this was deliberate:
Less than 100% reliability is essential for the sustainability of an off-the-grid solar system, because above a certain threshold the fossil fuel energy used for producing and replacing the batteries is higher than the fossil fuel energy saved by the solar panels.How to Build a Low-tech Website?, 2018-09-24
The site's background doubles as a battery meter; the lower it sits, the darker or cloudier it is in Barcelona. A dashboard shows the local forecast so readers can "plan" their visits. A website arguing that demand should chase supply makes its readers chase its own supply.
By this point it is easy to file him as a nostalgist. He does write a great deal about old technology, but the more of him you read, the clearer it becomes that he uses the past in an unusual way, and that he arrived at it by correcting himself.
His 2011 article "Medieval Smokestacks" opens by taking apart a myth he had enjoyed himself: that the Middle Ages and the Renaissance were a paradise of renewable energy. Dutch industrial windmills really were remarkable, powering paper-making, sawmills and glass polishing. But windmills deliver motion, not heat: you can polish glass with wind, but you can't make glass with it. The Dutch Golden Age made glass, bricks and tiles, refined sugar, brewed beer and baked bread, all of which needed heat, and the region's forests had been gone since before 1600. The answer was peat, a fossil fuel that renews so slowly that a three-metre layer takes at least three thousand years to grow back. Peat digging turned more than 60,000 hectares of Holland and Utrecht into open water, close to a tenth of the two provinces, and whole villages were swallowed by man-made lakes.
Our romantic image of the Middle Ages and Renaissance as a paradise of renewable technologies results largely because of our failure to distinguish between thermal and kinetic energy.Medieval Smokestacks: Fossil Fuels in Pre-industrial Times, 2011-09-29
So what does he take from the past? Not a particular old object, but a principle, which he then improves with modern technology. His 2015 article on heating is the clearest case. He noticed that modern heating nearly always heats air, while how warm we feel depends heavily on radiation and conduction. Stand barefoot on a cold floor and you'll be cold in a 21-degree room; sit in spring sunshine in a T-shirt and a metre away, in the shade, you need a jacket. With only one fire, people in the past heated people, not places: hooded chairs blocked the draughts behind them, foot stoves held embers shovelled from the hearth, the Japanese had the kotatsu and northern China the kang.
In an air-heated room, it doesn’t matter much where you are. In a room heated by a radiant heating source, location is everything.Restoring the Old Way of Warming: Heating People, not Places, 2015-02-11
He adds immediately that the old concept of heating was more efficient, but the old devices were not. An open fire sends most of its heat up the chimney and can even leave the room colder. What he wants are electric radiant panels, heated tables and heated cushions. In 2024 and 2025 he built a heated table and a foot stove powered by a small solar panel, turning the principle into objects.
His 2020 article on biomass, the most-linked piece on the site, shows the method most fully. It began with reader anger. He had mentioned firewood in a previous post, and the comments filled up with "How is cutting down trees sustainable?" He looked into how pre-industrial people got their wood and found that until the early twentieth century at least half of Europe's forests were coppiced: trees cut to the ground, new shoots harvested every two to four years, the tree itself never killed, some stools more than a thousand years old. The shoots served many purposes: willow for baskets, chestnut for barrels, hazel for wattle walls, leaves to feed sheep in winter, and worn-out wooden objects burned at the end of their lives.
The key move is that he asked why the system existed:
It was the limitations of human and animal power that created and shaped coppice management all over the worldHow to Make Biomass Energy Sustainable Again, 2020-09-20
Cutting by hand gets harder as the trunk gets thicker, so many thin branches were cheaper to harvest than one big tree. A cart hauling firewood over bad roads used up more energy in animal feed than the wood contained once the trip went beyond 15 to 30 km, so the woodland had to sit right next to the village. With a limited supply close at hand, a community had to keep harvest and regrowth in balance, or it ran out of fuel, fodder and materials and had to move. Fossil fuels, with chainsaws, harvesting machines and long-distance transport, cut that tie: wood could be grown on one side of the planet and burned on the other, and felling big trees became the most profitable thing to do.
Our forebears relied on coppice because they had no access to fossil fuels, and we don’t rely on coppice because we haveHow to Make Biomass Energy Sustainable Again, 2020-09-20
His real conclusion isn't about wood. About 40% of people today still burn wood for cooking and heating. A 2017 study he cites estimated that this accounts for 55% of the global wood harvest but only 2% to 8% of human climate impact, because roughly two-thirds of it is harvested sustainably: dead wood collected, trees grown outside forests, coppicing and pollarding, multipurpose trees too valuable to fell.
If the whole of humanity would live as the 40% that still burns biomass regularly, climate change would not be an issue. What is really unsustainable is a high energy lifestyle.How to Make Biomass Energy Sustainable Again, 2020-09-20
That sentence pulls the four ideas together. Efficiency, embodied energy, reliability, old technology: he takes the long way round each time and lands in the same place. The problem is not where energy comes from, but how much of it we want. His 2024 article on public bathhouses is the most sellable version of that conclusion. Ancient Rome had about one public bath for every thousand people; the Forum Baths at Ostia burned about as much fuel in a day as 557 modern showers, while holding up to 500 bathers at once for hours. He isn't proposing cold showers. He proposes pooling resources for "sustainable extravagance," which "may be an easier sell than cold showers."
Move the baseline. His favourite move is to push the starting point of a comparison further back: planes from 1985 to 1950, smartphones to dumbphones, tumble dryers to clotheslines, the grid to windmills. Shift the baseline and a lot of "progress" shows its real shape. It looks simple, but he uses it with discipline, always saying which "service" you are quietly assuming once the baseline is set.
Fill in the column set to zero, including his own. The manufacturing energy of chips, the overbuilding of a wind-and-solar grid, the battery behind his own website, the software that makes old computers obsolete. He doesn't just accuse others of leaving things out; he publishes his own site's worst number, along with admissions that a software upgrade once dropped uptime to 80% and that the site quickly went down on a 10-watt panel. Much of his credibility comes from these self-incriminating paragraphs.
Treat history as an experiment run under different constraints, not as a museum. He never asks whether the past was better. He asks what constraints produced a design. Once the constraint is clear, you can tell whether the design still works: coppicing was shaped by animal power, so today cargo bikes and aerial ropeways can extend its radius; fireside warmth was shaped by fire, so today it becomes an electric radiant panel. That only works if you first get the basic distinctions right, such as heat versus motion. Otherwise you end up where he was before 2011, mistaking the Dutch Golden Age for a wind-powered utopia.
Be his own test subject. The server on the balcony, the 50-euro laptop, swapping panels between the server and the living-room lights, and lately the heated table, the solar oven and the foot stove. His arguments often end with a photograph of something he built. That makes his articles more than opinions; they come with instructions you can follow.
Stand on other people's research, and say so. He isn't an academic. His core arguments often rest on the work of Shove, Coutard and Gutowski, footnoted in full. His strength is connecting research scattered across sociology, engineering and history into a single line, then pinning it down with one concrete object: an LED wall, a windmill, a ThinkPad.
First, who does the adapting? "Demand chases supply" sounds smooth, but who pays for the adapting? Wind-powered factories would put workers on night shifts, which he concedes "is considered unhealthy," before moving on to fully automated "dark factories." On sufficiency, he writes that deciding how much is enough requires normative judgments that are "sure to be controversial," and that it "risks being authoritarian, at least as long as there is a cheap supply of fossil fuels." On bathhouses he sees that a shower bathhouse where the government sets water temperature and duration could become "a technology to enforce frugality upon the whole population." He names all these risks but rarely takes the next step of saying who should decide what is "enough," and by what process. What he actually does is mostly personal example, and he himself calls that a hack.
Second, the age of the numbers. His 2009 chip calculation rests on a 2002 life-cycle analysis of a 32 MB memory chip. His 2020 laptop calculation cites an analysis of a 2001 machine, defended with a 2015 study arguing that such figures don't change over time. His 2017 storage calculation uses the battery energy and lifetime data of the day. All of these should be recalculated today; the direction may hold, but the size may not. In the comments under the 2009 article, a reader accused him of misreading the Smart 2020 report he cited, arguing that its appendix already accounted separately for the embodied carbon of manufacturing ICT equipment, so scaling 1.4 gigatonnes up to 7 doesn't stand. This page did not check that appendix and reports only that the accusation was made; in the comments read for this page, there is no direct reply from him to it.
Third, he sits at one end of the rebound debate. In "Bedazzled by Energy Efficiency" he fairly sets out the other side: some researchers find that behavioural responses shave only 5% to 30% off intended savings, no more than 60% with macroeconomic effects, and that "energy efficiency does save energy." He then pivots to Shove's argument that the problem lies not in the size of the rebound but in how efficiency locks in a definition of service. It's an elegant move, but it makes his position hard to refute with data: if rebound is small, the problem is the definition of service; if it's large, efficiency doesn't work. On the empirical question of whether efficiency saves energy, researchers have not come over to his side.
Fourth, no substantial named critic turned up. A search for systematic criticism of him and the magazine found only French-language pieces on whether the wider "low-tech movement" risks being co-opted by business (not read closely here, so their arguments aren't repeated), and no long, named piece taking on his core arguments directly. That could mean his arguments are sturdy, or only that his readers and his critics rarely meet.
The sharpest criticism is one he wrote himself: a website built to cut carbon turned out, by the most direct measure, to be dirtier than plugging in. He didn't delete that line; he used it to interrogate how the grid is measured. That the line was published at all shows that, at least on this point, he was willing to let the data come before the position.