HADES — A man was sent straight to Hell after his new phone autocorrected his text message from "Praise God" to "Praise Fod."

PALO ALTO, CA — After experiencing repeated mascara malfunctions, local woman Brittany Gleeson admitted to friends that it took over 600 takes to get the right video of herself crying into her camera.

A few months ago, I was privileged to tour the Michigan steel mill where my daughter works. A month or so later, I toured a brewery in North Carolina. Both establishments would have been recognizable to Adam Smith, whose description of the pin factory perfectly captured the division of labor fundamental to manufacturing. Yet the sage of Kirkcaldy would have been astonished at how much of the manual work of manufacturing is done today by machines — and by the level of capability needed by those who supervise those machines. 

It is a major part of the administration’s policy to “reshore” American manufacturing to bring back jobs, and a tenet of the national conservatism movement that America needs blue-collar jobs that pay a good wage. Yet both of these propositions rest on the idea that manufacturing needs to be labor-intensive. The best chance for reshoring manufacturing capacity is actually based on increased automation. That means more robots, but the administration has recently moved to restrict the import of robots, specifically humanoid robots from China. The more hostile the administration is to automation, the more it will drive manufacturing back offshore. Something has to give. 

This is made apparent when we look at the labor supply. Far from there being a huge supply of untapped manufacturing labor, the Manufacturing Institute estimates it will need 3.8 million workers between 2024 and 2033, and that half of those positions are likely to go unfilled because workers today expect much more flexible working arrangements than their grandfathers did. Automation is the surest way to bridge the gap. 

That’s certainly the current direction. The factory floor is increasingly staffed by robots. Industrial robot installations rose 11 percent in 2025, but America is behind the curve internationally. China installed over half of the world’s new robots last year, ten times the US figure. America lags well behind, ranking only eighth in robot density. This is part of the reason manufacturing productivity growth has lagged recently. 

Yet there are few domestic robot suppliers. We import almost all of our robots from Japan and Europe, but those robots act as an input into American manufacturing, enabling us to produce more of the stuff we are good at producing. 

So if we are going to reshore manufacturing, we must recognize that the factory that comes back to America will not — and indeed should not — look like the sepia-tinted factory that closed down 20 or 30 years ago. Milton Friedman reportedly visited China in an earlier age and was shocked to see people digging a canal with shovels rather than modern earthmovers. He was told that this was for a jobs program, to which he responded, “Oh, I thought you were trying to build a canal. If it’s jobs you want, you should give these workers spoons, not shovels!” 

China learned its lesson, yet to hear American politicians speak today, many of them would seemingly rather ban forklifts and have American workers carry boxes around a warehouse floor. Much of the current anti-China rhetoric is actually based on that earlier Chinese model, assuming that the reason China has gotten ahead is that it relies on cheap, even forced, labor. This fundamentally misdiagnoses the problem. China is automating rapidly.

To respond, America needs to up its automation game – and this will be good for the factory worker too. The Bureau of Labor Statistics expects production occupations (workers operating manufacturing machinery and equipment) to drop over the next ten years, but it expects the technical occupations associated with installing, calibrating, maintaining, and repairing machines to grow. These blue-collar, non-degree jobs pay a median annual wage of $63,500, more than most production occupations. 

Jobs like robotic technicians, which generally require an associate’s degree, and industrial and mechanical engineers are equally likely to grow in importance, and these jobs pay well. The National Institute of Standards and Technology (NIST) agrees. Its analysis of the manufacturing workforce sees a host of different careers with transferable skills opening up in manufacturing, all based around high-technology systems. 

So American reindustrialization requires two things — a highly educated, well-rewarded, flexible workforce and advanced automation and robotics. America does not need to staff its factories with PhDs, but the factory jobs of reshored production won’t be the repetitive jobs that Adam Smith feared would numb workers’ creativity, minds, and capacities. These jobs will require technically sophisticated, non-college career paths, perhaps involving apprenticeships and privately certified credentials rather than expensive college degrees. 

So the romantic picture of the plant gates reopening and thousands of workers starting and ending their shifts at the same time is not going to be painted throughout the Rust Belt. The picture should be far rosier, with higher wages, more flexible working conditions, and a much safer environment. 

Yet that all depends on the second part of the equation – the widespread deployment of robotic technology, which is the fundamental requirement of reshored industrialization. 

Factory robots today are not some form of steampunk C-3POs – they are primarily dedicated arms and tools. So the Federal Communications Commission’s ban on imported humanoid robots from China doesn’t look immediately relevant. Deployment of humanoid robots in factories is very much in the demonstration and pilot stage. Yet that’s what makes the FCC’s action so insidious to American reindustrialization. It will preempt the next stage of development. If China is currently deploying older-style robots by the hundreds of thousands, America should be leapfrogging that and experimenting with much more versatile humanoid robots, which can embody the artificial intelligence at which America leads the world and engage in all-purpose mobile manipulation. 

Ah, says the objector worried about national security, but won’t using Chinese-manufactured robots present an inherent security risk? Not if the hardware is run by American software. For example, a couple of months before the FCC action, NVIDIA, the American leader in global chip technology, announced its Isaac GR00T reference robot, giving fans of Real Steel and Guardians of the Galaxy equal thrills. The robot combines a Chinese-manufactured Unitree H2 Plus humanoid robot with NVIDIA onboard compute and software. As NVIDIA CEO Jensen Huang said, this “will bring physical AI to the world’s largest industries.” 

The GR00T platform also supports other Unitree products and was slated to be used by Stanford, UC San Diego, and other research institutions. The relatively inexpensive physical hardware was Chinese, but all the value-added aspects, such as chips, AI models, software, and the research programs were to be American. In a very real sense, the physical AI leap forward is American. 

The FCC’s action puts all this in jeopardy. As Reuters noted, Unitree is expected to be one of the principal casualties of the import ban. A case can certainly be made on national security grounds that the FCC ban was necessary. The hardware includes sensors that can map areas, can connect to networks, and potentially be commandeered. 

Yet those risks are probably better addressed by measures already taken into account by NIST in its cybersecurity guidance, such as network segmentation and tightly controlled remote access for industrial systems. As so often, the question is one of appropriate response to risk. Risk can be mitigated by strict rules for defense and critical infrastructure, cybersecurity certification, network segmentation, and approval for demonstrably secure platforms. This would allow American R&D in the emerging field of humanoid robotics without forcing American companies that are ready to go now to wait until American-based suppliers reproduce every technology already manufactured affordably overseas. 

China has the advantage in affordable humanoid robotics for the same reason it has the advantage in so many other manufacturing areas. Its companies cut costs relentlessly throughout the supply chain. The current speedy robot deployment is one of the contributing factors. America has plenty of innovative robot developers – from Boston Dynamics to Figure to Agility Robotics – but their capabilities are different from the mass Chinese manufacturers like Unitree. 

If America is to remain the global manufacturing powerhouse that, contrary to assertions by so many, it actually is, it needs to come to terms with the fact that China is installing robots at extraordinary speed. Restricting access to robot hardware cannot be the answer. The sensible response is to make America the best place to invent, improve, and deploy robots in new and remarkable ways.

For the moment, that requires access to state-of-the-art technological inputs. The FCC should rethink its actions and allow American companies to jiu-jitsu Chinese hardware into sustainable American competitive advantage. 

American factories, and American workers, will reap the benefits.

How does Paris get fed? Frédéric Bastiat famously explained in Economic Sophisms (1845) how market exchange reliably provisioned the (then) million people of Paris with agricultural produce from the countryside that they were able to enjoy “peaceful slumbers…not disturbed for a single instant….” 

In stark contrast, Bastiat predicted that there would be “much suffering within the walls of Paris — poverty, despair, perhaps starvation…” if a presumptuous minister decided to replace the market with their own decision-making for what “should be produced, transported, exchanged and consumed….”

We can appreciate Bastiat’s observation about the miraculous functioning of the market even more when we look at a time when Paris actually went hungry.

France’s Experiment in Forced Provisioning

Leading up to the French Revolution in 1789, France found itself in a precarious fiscal position. It had accumulated crippling debt from the Seven Years’ War and its support for the American colonies during their War of Independence. This heavy debt burden left the kingdom woefully unprepared to withstand the economic shocks that followed.

Economic shock came in the form of the eruption of the Laki volcano in Iceland in 1783, which contributed to climatic disruptions and poor harvests in France in the years that followed. These problems were compounded by a severe hailstorm in 1788 that devastated crops and livestock, raising prices, especially for bread, which was the main staple at the time. Increased demand for grain to support the military and its draft animals, when France declared war on Austria in 1792 (followed by war with Great Britain), pushed prices even higher. When France implemented a draft that drew agricultural workers into the military and then began requisitioning agricultural horses and wagons, the supply of grain was further reduced.

Henry Bourne, writing a two-part article in the Journal of Political Economy in 1919 about this era, notes that in the fall of 1792, “One of the longest and most important debates [of the National Convention] was upon the best method of insuring a supply of bread at a reasonable price.” This was a problem that especially loomed over the major city of Paris. Bourne argues that the threat of starvation fueled not only the French Revolution, but the mob mentality and interventionism that followed. As Bourne writes, “People, in a panic because they do not know where next week’s bread, meat, and coal are to be found, are not likely to apply the rules of evidence to every rumor.” The French clamored for state intervention on the “fixed idea that dearness and scarcity were the result of speculation” rather than underlying economic conditions.

Transporting grain became a risky enterprise as mobs sprang up to seize it, further decreasing the supply of grain to Paris. To add insult to injury, the transportation of grain to major cities was further suppressed by inflation, which made the issued assignats unappealing to country farmers.

The National Convention and the Paris Commune turned to “a series of ventures in price-fixing and food control” to solve the problem. Bourne notes that “price-fixing became one of the characteristic features of the Reign of Terror.” In 1793, the National Convention imposed a maximum price, or what economists today call a price ceiling, on grain. In a futile attempt to warn of the potential consequences, Pierre Vergniaud, who later that year was executed under the accusation of the radical Jacobin Maximilien Robespierre, urged that “If you destroy commerce, you decree famine.”

French attempts to deny the economic reality reflected by market prices, by attempting to suppress them, resulted in severe shortages and long lines. 

“The scheme not only failed to encourage the farmer, it threatened him with ruin,” Bourne noted. “His expenses for tools, draft animals, and wages were steadily rising, but his profits were cut down, with the prospect of further losses every succeeding month.”

But politically savvy politicians blamed these disappointing outcomes on greed and used them to justify further interventions backed by the threat of imprisonment and death. The National Convention created a Commission of Subsistence and Provisioning to be the “Food Director” of France. Swarms of officials were commissioned to survey farmers’ inventories and fields in an attempt to enable government officials to redirect grain to where it was needed. Rules were issued detailing the precise percentage of bran that millers could extract and even dictated the one type of bread that would be allowed. A bread card rationing system was created but was abused as families failed to report the death of family members to continue receiving the same allotment. Bourne reports that in 1794, rations fell to a single pound of bread for each laborer and three-fourths of a pound for others, and that “it was practically impossible to obtain meat, butter, eggs, oil, and other articles of food commonly regarded as necessary,” as price ceilings were extended to these items as well.

Officials attempted to appeal to the higher motives of the people, telling them that they were “brothers and that they should help” even if it meant turning over the grain needed for their family, for storage for future use, or even the seed necessary to plant the next year’s crop. This proved insufficient, however, so the officials eventually turned to force.

Bourne writes that “An attempt was made to provide for Paris by compelling every farmer to furnish within twenty-four hours sixteen bushels of wheat for each hide of land.” French dragoons were soon released upon the countryside to “scour the country” for food and to arrest any suspected hoarders. As Bourne notes, “merchants were thrown into prison upon the accusation of the first intriguer who shouted out his suspicions at a popular society. The local revolutionary committees acted as judges without appeal. To escape a similar fate the other merchants hastened to dispose of their merchandise and did not restock.”

If a farmer had grain in the field but no laborers to gather it, laborers were drafted by local authorities. Millers and bakers in Paris were drafted and forbidden from abandoning their work without sufficient notice. Eventually, the National Convention even attempted to extend maximum price laws to the wages of laborers as well.

Despite the substantial and systematic efforts of the National Convention and the boards of the separate departments of France, Parisians and much of the rest of France, went hungry under government control. In Cahors, people “were so poorly fed that they were falling in the street from sheer weakness.” In Nord, “grain of every sort disappeared from the markets…” The people of Paris would stand “with famished eyes” for hours in line “only to be told when their turn came that nothing was left.” As Bourne concludes, “If the maximum laws were meant to save the common people from want and wretchedness, they failed.”

Bastiat’s Market-Fed Paris

It is unclear whether Bastiat, when writing in the 1840s about the remarkable way in which free markets coordinated the efforts of countless individuals to feed Paris every day, was implicitly contrasting this outcome with the French Revolution’s earlier rejection of market exchange. He almost certainly knew that revolutionary France had experienced severe food shortages and government price controls, making the contrast between the two episodes striking even if he did not intend it.

As Bastiat stressed, government officials could not replace the information and incentives provided by market prices. Orders, price controls, requisitions, forced sales, and even forced labor failed to feed Paris. When the National Convention tried to do so, it produced exactly the outcome Bastiat had predicted more than half a century later: not peaceful slumbers, but long lines, empty markets, and widespread hunger. Notably, these outcomes began to recede as the Commission was abandoned and markets were restored.

We hear enough of how appalling capitalism is in this modern world. We should, perhaps, pay more attention to the disasters that befall those in the absence of capitalism:

More than 100 people have died after an artisanal goldmine at a small village in western Central African Republic (CAR) caved in due to a landslide, local officials and aid workers said on Wednesday.

“Artisanal” in mining terms means men with shovels and little else but a shovel and their labour.

A second prosecutor from Baboua told the press that the incident is believed to have been caused by “the collapse of several underground tunnels in which miners were operating”. “The ground completely collapsed and buried people,” Cedric Mbango, a miner at the site, told the AFP news agency, describing the incident as “catastrophic”.

As we’ve noted before this is non-capitalist mining. The safer way to mine is to perhaps run with an open pit. Or, if tunnelling, to shore up said tunnels. But both of these options require much more money to be spent before there is any revenue from output. They require capital that is. Requiring capital does indeed usually mean requiring capitalists. You need bulldozers and diggers and scoopers and something to shore the tunnels up with. Capital goods which are afforded from capital provided by capitalists.

True, capitalist mining might mean that the workers will be exploited, the surplus of their labour being expropriated. They might have a greater chance of being alive too.

As Deirdre McCloskey, echoing Joan Robinson, has been known to point out, the only thing worse than being exploited by a capitalist is not being exploited by a capitalist.

Tim Worstall

Mike Masnick, in a thread on Bluesky:

Multiple people I know have told me that they love the idea of Bluesky, and want it to succeed, but have abandoned it for X because they use agentic tools in their work and find them incredibly useful, and feel that any mention of their usage here leads to hate and ridicule.

I wouldn’t bring this up if it was just one person, but it was multiple people since the beginning of August who mentioned in passing something to the effect of either “I’d spend more time on Bluesky if people weren’t so anti-AI” or “As much as I hate to be there all the useful AI talk is on X.” [...]

I’m not telling anyone what to do or what tech to use or what to like or not like. I’m just relaying that the reputation this place now has is one that a large segment of the population who are NOT fascists, saying they don’t feel comfortable here because a main area of their interest is considered worthy of being mocked and attacked. And even as I’ve explained that I’ve been able to have positive conversations about productive AI usage here with many of you, the reputation of the community here as insisting their own lived experience can’t possibly be real has driven them away.

It’s not just “the reputation”. That really is the culture of Bluesky. I’ve obviously written more about AI this month than usual — in particular, my two long pieces regarding watermarking LLM-generated text through word-choice diddling. I don’t really write much about AI generally.

Broadly, I’m making two arguments against text watermarking. First, that the embedding of a detectable signal (even if subtle) must necessarily degrade the quality of the generated prose (even if only slightly), and that this is an unacceptable trade-off for a feature that has no benefit to the user and in fact can only be used against the user. Second, that even if I’m wrong about the first part, and a clever enough watermarking scheme can embed a detectable signature without any degradation of prose quality, this is a bad idea and should be rejected.

Reaction to my takes on text watermarking have largely been about these two arguments. Some people agree with me that what watermarking proponents are claiming is not possible — the text quality must degrade. Others agree that while it must have an effect on text quality, it in fact is so subtle that it truly does not matter and even a careful reader will not notice an overall degradation. Others disagree, tell me I don’t understand the cleverness of the systems, and are convinced that somehow this watermarking truly embeds a detectable signal in prose without any degradation of text quality whatsoever. There’s a range of disagreement, but it’s all reasonable. Likewise on the broader question of whether it’s a good idea or not to even try to do this.

The reaction on Bluesky to my recent work is like nothing else. It’s largely in the vein that watermarking can’t possibly make AI-generated text worse because it’s already random gibberish, and that the only explanation for my being opposed to this is that I must be using AI to produce my own writing and I’m upset that my scheme will soon be detectable. Read the replies to this one; they speak for themselves.

I continue to think my friend Daniel Jalkut has espoused the one true take: “My take on AI is, essentially, everybody who’s against it is too against it and everybody who’s for it is too for it.” It is way less inflammatory to espouse an AI-skeptical stance on Twitter/X than it is to simply state that AI can be truly useful on Bluesky.

Anyway, Masnick, at the end of the day yesterday:

The people bitching about me turning off replies on my thread earlier today are doing an excellent job reinforcing why I was right to turn off replies.