Technological unemployment, the permanent unemployability of whole industries because of advancing automation, is on the horizon. But technology is a two-edged sword: in replacing labor, it gets rid of jobs (bad), but also makes goods cheap and accessible (pretty good). In 1930s terms, that means it’s not clear if automation will be more like the depression or the new deal.

If near-future manufacturing tech ‘compresses’ supply chains for a set of low- and medium-tech consumer staples, the technological unemployed gain the option to opt out of competitive global economies, seeking instead a slow tech life in hyperlocal networks of production and trade. But these hypothetical non-participants will have to interact with the global supply chain for some Clarkeian magic: dank TSMC silicon maybe, or that 2nd jab.

The leavers will live in a parallel material culture which values adaptable machines, and which prioritizes maintainability, compatibility, and repair – a ‘lego economy’ where the tractors run linux and every charger works in every phone. Manufacturers will hate this; they’ll strive to keep their goods from being reusable, dissectible, compatible, and maintainable (‘no user-serviceable parts within’ from Vernor Vinge).

Planned obsolescence isn’t new, but digital hardware changes the game: it’s too easy for vendors to lock down and degrade. We’ve seen this in our century with the prevalence of hardware DRM and remote bricking, and the rise of the open source and repair movements in response.

The cold war of 20th century labor relations likely won’t survive the end of full employment. But producers will still need to move inventory; the economic conflict of the 21st century will be over who gets to farm consumers. The weapons in this war will be the locking down, and the hacking open, of physical technology.

There’s some history here. Scroll down to read about past crises, current trends, and future conflicts.

(If you read with a soundtrack, try everything is free now from gillian welch. I paid for my copy. The phoebe bridgers cover is also good).

Technological unemployment isn’t new

They’re not always good jobs, but as a society, we’re very used to having enough jobs for everyone. People suffer when this assumption doesn’t hold. The bottom of the Great Depression was the only time in the 20th century that more than 1 percent of the US labor force was unemployed for more than a year1. That time had a lot in common with ours, and can help us understand what we’re facing.

The depression was plausibly a combination of cyclical unemployment and structural unemployment2. This claim surprised me when I found it; I always imagined the GD as a really long, bad business cycle caused by them not knowing anything about monetary policy back then.

(Definitions: cyclical unemployment is temporary and driven by the business cycle, structural unemployment is ‘permanent’ unemployment that eliminates types of firms, roles or even industries).

A big fraction of depression job losses were from firm closures rather than individual firings3. In industries that had begun to mechanize, the firms that survived were more likely to be mechanized4, and therefore more likely to have a high bar for hiring. Well-educated workers with seniority were more likely to keep their jobs5.

Paradoxically, the firms that survived weren’t doing it by underpaying their workers. In the 20s, firms discovered that cutting wages no longer increased margins, and in fact raising them helped by attracting better workers6. This efficiency wage was non-market-clearing, meaning that there were many people who were willing to work for less, but were still un-hireable. Government supported this too: in November, a month after the crash, Hoover brokered a truce between labor and industry that maintained hourly wages at existing levels in order to ‘maintain the social order and industrial peace’7.

Eventually we recovered, maybe because we were the only country to survive the war intact. Some economists8 say that universal high school education brought us out by improving training, making the country ready for white-collar jobs requiring advanced thinking skills.

One of my sources controversially thinks that expanded high school, college, and social security in combination took the labor of the young and old off the market9 – that the solution wasn’t to increase supply of trained workers, but to reduce supply of untrained, that the increases in manufacturing productivity let us work for fewer years of our life.

In a sense they balanced the deflationary trend of technology10 with entitlement programs funded by inflation.

Unemployment in the GD peaked at 25%. As in 2008 and 2020, some people were badly off, but others had stable or increased wages. Factories were hiring a million workers per quarter even in the deep depression11, and 25% deflation meant lower cost of living. If you had income, you were doing great.

But if you didn’t, you weren’t. Let’s look at how 30s households coped, and then think about the decade to come.

Barter-autarky and feed-sack dresses

If technological unemployment brings about broad job loss without goods shortages, we have to figure out how to share our stuff without using cash to keep score. Money was tight in the 30s too. ‘Money is tight’ is a euphemism for not being able to afford something, but I mean literally that some households had to meet some of their material needs in non-cash ways12.

Home economics, a thing that I thought of as a set piece for 50s high school time travel scenes, was a lifeline to depression households. Print media and radio were used to distribute new science-based advice about meeting nutrition needs on a budget. People made feed-sack dresses.

Change and hard times force us to discover new categories of subsistence goods, new staples, famine foods.

Money is nice but it’s not the only way to get by. Barter can be a handy way for local producers to stay warm and fed when their monetary system has exploded, especially if your economy has relatively few goods so you don’t have to deal with the ‘double coincidence of wants’. (Anthropologists will remind me that prehistoric societies were probably based on gifts, not barter).

I read a paper about the Peruvian highlands13 which looked at how inflation affected the participation of subsistence fishers in markets. During inflation, they can ‘expand subsistence’ by opting out of the cash economy. When prices for manufactured goods rose much faster than fish, fishers showed a preference for exchanging their fish with other ‘highland goods’ (grain and potatoes). When things cooled down, they sold fish for cash which could be exchanged for ‘urban commodities’.

Demonetization isn’t restricted to highland fishing villages. When hyperinflation hit the 90s ruble, half of b2b transactions were barters rather than cash14. Wages and taxes were often paid in goods. Monopoly utilities like Gazprom collected only 10% of their revenue in cash. The rest was in “veksels, coal, metal, machinery and even jet fighters”. (A veksel is an IOU I think).

Technological unemployment is an inflationary crisis if you’re in the group that’s left behind and can’t earn cash. But tech advance is in a sense supply chain compression15. Viktor Shvets, my favorite thinker on economic change, calls this ‘physical disintermediation’. Small communities will be able to produce more categories of stuff and trade it locally rather than compete globally.

There’s an observation I associate with MLK’s poor people’s campaign that urbanization made poverty worse because in the city, with no land, people couldn’t grow their own food. Supply chain compression could in principle reverse some of the 21st century’s trend to urbanization.

But near-future hyperlocal supply chains won’t be able to make everything. Like the highland fishers, they’ll transact in the broader economy for cutting-edge goods and services. Used stuff may bridge the gap; I’m probably as happy to get an MRI in a 90s machine as a new model (health was only 12% of GDP back then, vs 18 now). But vendors hate it when you resist their mandatory upgrade cycles and will resist.

If you’re not the farmer you’re the cow

In commercial broadcasting the viewer pays for the privilege of having himself sold. It is the consumer who is consumed. You are the product of t.v. – R. Serra, C.F. Schoolman

There’s a war brewing over repair. Makers of digital hardware are using their whole technical and legal arsenal to control how people use their stuff. This is leading to a conflict between companies and ordinary people that we’ve only seen before in labor disputes16.

When Hoover negotiated the deal that traded high wages for low hours in the depression, his goal was to preserve the ‘industrial peace’.

For an example of what it’s like when the industrial peace isn’t preserved, look to the life of Henry Ford’s top henchman, Harry Bennett, the head of Ford’s shadowy ‘service department’. When not firing live rounds at marching UAWs, he lived in a rotation of comic-book villain hideouts in the woods of michigan, weird postwar fortresses with armed guards, escape planes, and gilded age gaudy accents. One had a moat; the drawbridge was rigged with TNT.

In theory the purpose of a state is to sublimate physical conflict into judicial conflict. In the case of labor relations, we didn’t always succeed. But we’re realistically approaching a post-labor economy, or one that is so high-productivity that compensation isn’t a major line item for big business.

Firms’ apparatus for controlling labor will be retargeted at the neo-luddite localized economies who dissect and repurpose their toys. The first front of this war is the repair movement, a group of farmers and repair shops who have been fighting companies like apple and john deere to modify and maintain their hardware. 1920s firms saw strikes as a weapon17. 2020s firms see repair in the same way.

Legal battles over used stuff and spare parts aren’t new. In 1953, US v United Shoe stopped a decades-old ‘lease only’ strategy. In 1994, Allen-Myland v IBM put an end to IBM “recapturing old parts that could otherwise have been used to extend the useful services lives of existing used mainframes”18. Restricting the secondhand market is an old trick for turning durable goods into a revenue stream.

Disposable things are physical products, but repairable things are in a sense information products. You need to have them assessed by an expert before you can buy. You need to source spare parts (themselves often secondhand). You need to track down rare expertise to repair them. It’s difficult to make them compatible with new stuff, but their lives can be infinite given the right information.

The point here is that disposable things get consumed or wear out, but information products have long lives. If you want to monetize them, they have to be controlled with hostile measures, a.k.a. legislation and DRM.

The dream of vendors is to price their goods based on what you can afford rather than having a set market price. A Xerox exec, quoted in a 70s business week article, says “we’ve set prices from an economic value to the user – how much a copy is worth, not according to the value of the machine”19. Note that ‘economic value’ is more or less a synonym for ‘tax the client’. These vendors are acting like predators when they should be mutualists. Separatist-consumers won’t put up with it.

‘Consumer technology’ has always been an oxymoron; if you’re not operating the technology, it’s consuming you.

An automation new deal

Alfred Sloan, depression-era inventor of car loans and planned obsolescence (he denies it), is the reason you want a new car every year. He made his money selling to a stagnant, saturated market that didn’t benefit from his product; to me he’s the slick face of waste and exploitation. He said in the 40s:

Some have an idea that the reason we in this country discard things so readily is because we have so much. The facts are exactly opposite – the reason we have so much is simply because we discard things so readily. We replace the old in return for something that will serve us better.

Compare with Henry Ford, who made his fortune in Detroit’s growth phase:

Our principle of business is precisely to the contrary. We cannot conceive how to serve the consumer unless we make for him something that, as far as we can provide, will last forever. … It does not please us to have a buyer’s car wear out or become obsolete.

We never make an improvement that renders any previous model obsolete. The parts of a specific model are … interchangeable with similar parts on all the cars that we have turned out. You can take a car of ten years ago and, buying to-day’s parts, make it with very little expense into a car of to-day. Having these objectives the costs always come down under pressure.

Yeah he was a mitlaufer and had protesters shot, but boy could he turn a techno-utopian phrase. For me, ford + sloan are the two possible futures ahead: mandatory upgrade cycles imposed by centralized producers, or infinite compatibility and repair.

A sufficiently large wealth gap starts to look like a trade barrier. This is a bad thing because the great things society makes for rich people end up unaffordable to everyone else. But it’s a good thing if you’re producing goods inside the barrier. Tariffs are moats. Tariffs cultivate infant industries (except for the smoot-hawley tariffs which just hastened the apocalypse).

I’m not a marxist and this isn’t a marxist paper, but wouldn’t it be funny if neo-luddites seized not the means of production, but the role. Not the physical plant, but the plant’s place in the supply chain. Who pockets productivity increases? So far, per piketty, it’s been more capital than labor. A decentralized, bottom-heavy supply chain changes that.

I don’t know whether post-scarcity leads to utopia, famine, or war. In 3000 BC, after 400 years of growth driven by massive productivity increases in agriculture, neolithic populations in europe suddenly collapsed. We don’t know why – maybe plague, maybe the grain stopped growing, maybe a boom/bust cycle. Maybe it was technological unemployment.

Maybe post-scarcity is a pipe dream and thomas malthus, like my man harold camping20, was only wrong about the date.

In any case, look out for:

  • Tech that compresses the supply chain by moving capabilities down the stack
  • Legislation that impacts rent-seeking by producers, one way or the other

These are the pivot points for the hyperlocal future.

Editors: Cameron Zargar and David Burt read an early draft of this post and helped get it where it is.

Appendix: UBI isn’t new either

(Hey 👋! If you made it this far you should subscribe, I’m full of odd ideas).

‘Work relief’, programs like the WPA that employed the unemployed, was a sticking point for organized labor. The head of the AFL forced roosevelt to take him on a tour of a CCC camp (‘civilian conservatoin corps’, I think a dorm-based park ranger org for young men) to prove they weren’t learning marketable skills that would take union jobs. WPA / CCC et al failed as retraining programs. Employers may have also dropped the ball here, reducing on-the-job training21.

Also: training didn’t make people better than machines at heavy lifting. Who’s the john henry of the digital age? We don’t know what re-training looks like when the mechanized substitution good is replacing human brains rather than human muscles.

The idea that people can be left behind by advances isn’t new. Underrated 19th-century economist Sismondi, early theorist of the business cycle, advocated slowing the pace of technological advance to let society catch up. In practice I don’t see how one country can do this unless every other country also does it.

The impact of technology on employment can vary by case. Adding robots to German factories costs fewer jobs than in the US, possibly because of less sticky wages22. Conversely, a Foxconn factory in Jiangsu installed a bunch of robots in 2016 and cut employment from 110k to 50k23 (possibly also due to falling demand).

Depression-era people preferred work relief to ‘home relief’ (a.k.a the dole) because working felt productive and the dole was stigmatized. I’m not sure how our society will process these issues if we move to UBI.

In Rome the ‘cura annonae’ was a subsidy of grain that went to 200k residents of the city. It was made possible by a vast global supply chain. Satirical poet Juvenal coined a phrase ‘bread and games’ to dis the practice, and call out what he saw as decadence in a culture that once conquered a third of three continents.

There was context here he was maybe missing – many small farmers had been bankrupted in the previous generation by consolidation and by increasing intensity of slave labor on large estates.

Juvenal’s take that a government based on handouts was at the end of its accomplishments, and was in some way unethical, is a view that our culture has in it too. A load of 20th century wonks, especially the 70s inflation hawks, thought that keynesian economics was a violation of the protestant work ethic. But this is fading; the younger generation has weeeird ideas about work and big questions about the legitimacy of the system they’re inheriting24.

I’ll leave you with this thought, not mine:

The lack of a positive vision for what comes after the Industrial Age has created a narrative vacuum exploited by nihilist forces such as Trump and ISIS. The failure to enact radical changes is based on vastly underestimating the importance of digital technology, which is not simply another set of Industrial Age machines. – Albert Wenger


  1. Causes and Cures of Unemployment in the Great Depression / Jensen 1989, p556 

  2. Jensen p573 

  3. Employment and Unemployment in the 1930s / Margo 1993, p50. Citing Lebergott 1989. “much of the initial decline in output and employment occurred among firms that exited” 

  4. Margo p50 “Plants that exited between 1929 and 1933 had lower wages and lower labor productivity than plants that survived”. 

  5. Margo, p53. Race also played a role. “unemployed were more often nonwhite, less educated and had fewer skills than employed persons. Such differences tended to be starkest for the long-term unemployed” 

  6. Jensen, p560 

  7. What or Who Started the Great Depression / Ohanian 2009, p2. This deal also encouraged factories to reduce hours rather than reduce wages; Hoover thought he was inventing an economic rescue package but was actually inventing gig work. 

  8. I think I have this from Robert Gordon, Decline and Fall of American Growth. I’ve either read the book jacket or the kindle preview but not both. 

  9. Jensen p581. “on the supply side, the growth of high schools and colleges, the postwar draft, and social security retirements removed young and old from the labor force”. I’m darkly fascinated by turn of the century images of teenagers with real jobs. Yes, they were crappy jobs, but not always. Rudyard Kipling was a journalist at 17. The high school and college system holds back self-starters, and we’re seeing the youtube / tiktok generation resist this. Labor force participation peaked in 2000; a 3 point decline since then is mostly down to teenagers and 55+

  10. mini book review of jeff booth’s ‘deflation and abundance’, which I haven’t read. Another good writer on inflation issues, more focused on the macroeconomics than the technology, is lyn alden 

  11. Jensen p564 

  12. The town of Tenino WA printed custom currency in the 30s to get through the depression, and rebooted the press last year to get through covid. 

  13. Barter and Cash Sale on Lake Titicaca / Orlove 1986. Interestingly, trout fishing was more capital intensive in that it required an outboard motor and fuel; trout fishers generally did sell their fish on markets because they needed the cash to run their small biz. 

  14. Barter in Russia, Guriev + Ickes 1999. Uhhh there were other reasons for this too, not just inflation. Receiving goods instead of cash let businesses in tax arrears avoid instant deductions from their bank accounts. Goods also let utilities avoid having set prices, which let them charge more to clients that could afford more. 

  15. Is 3D Printing a Threat to Global Trade / Freund, Mulabdic, Ruta, 2019. They expected 3D printing of hearing aids would reduce trade, paradoxically it increased it. They also talk about dental crowns, making the point that a specialized alloy process in the UK and a specialized coating in europe require the part to be shipped around a lot; even though it’s printed, the materials they use are specialized. 

  16. And in Erin Brockovich, I guess; injuries and pollution. 

  17. Ohanian, p22. Also from Ohanian: “Bernstein describes firm violence during several union organization attempts and strikes during the 20s. Tactics included kidnapping organizers, firing workers who met with organizers, evicting strikers from company-owned homes, denying medical care to striker families from company health providers, and beating and shooting strikers. Firms were sometimes able to bribe local police. … Firm actions during strikes were rarely prosecuted, but union actions were often prosecuted.”. This tracks Hofstadter’s ‘american violence’, which claims that the pattern of violence in this country is typically the middle enforcing the desires of the top against the bottom. 

  18. The Durapolist Puzzle / Orbach 2007. For planned obsolescence strategies specific to software, check out The neo-Luddite’s lament, which can tell you about forward and backward compatibility, network effects, setup costs of forced upgrades, and oligopolies. 

  19. I think my source for this is Orbach again 

  20. Camping ran a radio show about the bible which I would listen to in high school while driving to karate class. He predicted the apocalypse from biblical cues, several times, was wrong (several times), died, but previous to dying collected large donations from his followers. Like, people sold their possessions. I don’t know what he told them he was doing with the money. He was an ineffective radio host and one guy would always call back and make him read random bible verses until camping realized it was the same guy and yelled at him. Mostly he signed off every call with ‘thaaank you for calling and sharing, shall we have our next caaall please’. I’m currently reading ‘when prophecy fails’ by festinger, about why people stick with cults that have been wrong, and may understand camping better when I’ve finished. 

  21. Jensen p575. “A silent, deadly effect of the great depression was that firms sharply reduced on-the-job training. By the end of the decade, widespread shortages of skilled labor indicated that the nation had been living off its accumulated stock of human capital.” 

  22. German robots: The impact of industrial robots onworkers / Dauth, Findeisen, Südekum, Wößner, 2017. p42-43. “German unions have a strong preference for maintaining high employment levels, and are willing to accept flexible wage setting arrangements, such as opening clauses, in the presence of negative shocks in order to keep jobs.” 

  23. 60,000 workers culled from just one factory as China’s struggling electronics hub turns to artificial intelligence , Mandy Zuo in SCMP, 2016 

  24. There are a lot of ways in which americans are questioning the legitimacy of our system (as implemented and as designed), including the criticisms of police practices that motivated the George Floyd protests. An economic example: gallup found favorable views of socialism significantly up for young adults. They also believe in climate change and feel like they’re being handed a bunch of student debt and melting icecaps.