Honoring the debt could destroy the political coalition necessary to govern France; defaulting could destroy the financial capacity needed to carry out its policies. Default, which is an actual possibility, would have strongly deleterious effects on German banks. There could well be a ripple effect in Europe.

France’s basic problem continues to be that its large public debt impedes its growth and its participation in the euro renders it unable to monetize its debt.

As I said before, all of this is primarily of concern to me because of the prospective impact on France’s defense posture.

I wanted to take note of something said by Dariush Mozaffarian in his recent Wall Street Journal op-ed:

Two people with the same age, sex, size, weight and activity level can burn substantially different numbers of calories depending on what they eat. Recent global studies show that differences in obesity rates between populations are explained less by varying physical activity than varying resting metabolic rate. Analyses from the U.S. and U.K. find that the average energy we burn at rest has declined significantly over the past several decades.

Together, these discoveries help explain why decades of calorie-focused advice have failed to reverse the obesity epidemic: Calories represent arithmetic detached from biology.

None of this violates the laws of thermodynamics. Weight gain still reflects an imbalance between energy intake and energy expenditure. The mistake is assuming that calorie counts alone explain a process that is shaped by the microbiome, hormonal signaling, fat storage, muscle maintenance, thermogenesis and metabolism itself.

The emphasis is mine. How much of the differences among those factors is governed by behavior, how much by environment, and how much by genetics?

I think the honest answer is that we know these influences matter but we cannot reliably say how much each contributes to the differences he describes. Naming the biological mechanisms does not resolve that uncertainty.

There are several additional complexities. “The population” of the United States today is considerably different from what it was in 1980 and even more different from what it was in 1940. It also differs from the populations of Brazil, Germany, and China. How much of the reported variation reflects changes in diet and how much reflects differences in the populations being compared? I don’t think we can confidently say. Furthermore, does Dr. Mozaffarian think those differences matter only when it comes to diet? Why wouldn’t they matter in other aspects of health including responses to medications and other treatments? If they do, that raises questions about much of modern medicine as practiced in the United States. Population averages may be useful but applying them to an individual requires assumptions. If those assumptions warrant questioning in dietary advice, they warrant questioning elsewhere as well.

I was gratified to see that the editors of the New York Times had come around to the strategy I arrived at by inspection for managing whatever threats might be posed by artificial intelligence:

When companies create products that hurt people, the victims can sue. The legal system allows those who are harmed to seek compensation from those who are responsible. These liability laws are one of the oldest forms of corporate regulation, and an effective one. Liability is a powerful incentive to behave responsibly. It speaks to corporations in a language they understand: Money.

Artificial intelligence is an industry in need of incentives to behave responsibly, to judge by the litany of transgressions to which leading A.I. companies have confessed in recent months. Their programs have hacked or attempted to hack the databases of private companies and public institutions, including the governments of the United States and Australia.

This is a welcome addition:

Liability is an incomplete answer for the challenges posed by A.I., but it has two great advantages: The laws are on the books, and they are powerful. Makers of cars, airplanes, tobacco products and opioids have all been forced by lawsuits to change how they do business. They made their products safer, restricted access to them or both. Most recently, the social media company Meta agreed to pay billions of dollars to states and to start changing its policies toward children.

and

State legislatures have a role to play as well. They can write statutes to clarify the application of existing laws to A.I. Legislatures can also hold A.I. companies to a higher standard of accountability than current law does, given the risks. Companies are generally liable if a plaintiff can show that they knew or should have known about the danger their products posed. But some particularly dangerous activities, like storing toxic chemicals and operating a nuclear power plant, are held to a standard of strict liability, meaning that the company is responsible for harm even if it had no knowledge of the danger in advance.

The theory of strict liability is simple: If you own a tiger, and the animal bites a neighbor, the essential fact is that you own a dangerous wild animal. It doesn’t matter how cleverly the tiger was trained, or how carefully it was guarded. Bryan Choi, a law professor at the University of Colorado, notes that in the early days of aviation, lawmakers imposed this strict liability standard on the dangerous new technology. In the case of A.I., top executives themselves have spoken publicly about the dangers, up to the possibility of human extinction.

Some industry advocates argue that enforcing liability laws will stifle innovation, because A.I.’s defining breakthrough is its ability to act independently. Indeed, they argue that any significant restraints on A.I.’s development would be a mistake because the potential benefits of racing ahead are so great. In a recent interview, Sam Altman, the chief executive of OpenAI, said, “We believe that the world should accept some bad things happening for the benefits of this technology.”

Yet other industries do not escape legal responsibility for their products in the name of innovation.

Strict liability is essential for LLM AI. Due to the very nature of the technology it shouldn’t be necessary to prove intent or negligence; only harm and causation. Those hurdles will be high enough. If a company deploys a system whose particular harmful behavior cannot reliably be anticipated even with reasonable care requiring victims to identify a negligent act can leave them bearing the costs of a commercial activity from which others profit. That is a substantive argument for strict liability, and one developed in legal scholarship on AI.

Today is my father’s birthday. He was born on October 10, 1914 above a saloon owned by his grandfather.

He did not have a happy childhood and his own father died when my father was just 13 years old. When his grandfather died less than a year later, he became de facto head of the family.

He was an excellent student and was the first person in his family to graduate from high school. He went on to college and graduated Phi Beta Kappa. He went on to law school and graduated from that with a Juris Doctor, something of a rarity in those days. It being the heart of the Depression, jobs were hard to come by so he spent the next year in Europe. We have journals of his experiences there in Great Britain, France, Germany, Italy, Greece, Yugoslavia, Egypt, and North Africa.

When he returned to the United States he worked as a newspaper editor, writing editorials for the St. Louis Star, then as an insurance claims adjuster. He finally got a job as an associate with a prominent St. Louis law firm and taught law school at night at St. Louis University. He went on to establish his own law practice and was successful at that until his death.

Those attending his wake and funeral included his grade school, high school, and college friends and colleagues.

He was the bravest, kindest, smartest man I’ve ever known.

These are some of my previous posts about him:

Blogging, c. 1940
My Dad’s Birthday, 2007
My Dad, 60 Years Ago
My Dad’s Birthday, 2009
My Dad’s Birthday, 2011
My Dad
My Dad 2
1940s St. Louis
My Father’s Stories

To place this issue in perspective it helps to recognize that Microsoft is among the largest sponsors of H-1B visas. At the Associated Press Seung Min Kim, Rebecca Santana, and Barbara Ortutay report:

WASHINGTON (AP) — President Donald Trump’s administration announced Thursday it was suspending Microsoft and several other firms from a program allowing foreign workers to apply for green cards as it also lashed out at universities that bring international students to the U.S. on exchange programs.

The announcement by Vice President JD Vance marks the latest swing from the Trump administration at employers and universities that welcome foreign students and workers into the country.

Such programs have frequently been targeted by some Trump supporters who argue they undercut American students and workers. Supporters say they allow the U.S. to attract the best talent from around the world and fill important hiring gaps.

The announced action freezes new and pending applications under PERM, the permanent labor certification program used for employment-based green cards. PERM requires a finding that sufficient qualified and available U.S. workers cannot be found. The form of fraud alleged by the Trump Administration is advertising the jobs in small town newspapers so they can preferentially rely on H-1B visa holders rather than domestic workers. That’s been going on for decades. I first became aware of this strategy for fifty years ago.

Other companies that sponsor large numbers of H-1B visas include Amazon (18,000), Meta (6,200), TCS (6,100), and Apple (4,600).

I fully anticipate Microsoft’s appealing this particular decision.

IMO stricter enforcement of H-1B visa sponsorship is long overdue but suspending Microsoft’s ability to sponsor is a more aggressive form of enforcement than I anticipated. I would be satisfied merely requiring companies to report the labor performed by contractors offshore to deduct those expenses for tax purposes.

As I have said before it is not my custom to comment on the politics of countries other than my own. That is why I have refrained from remarking on the riots presently roiling France. That has not stopped the editors of the Wall Street Journal:

It’s easy to scoff at French rioting as its national pastime, but what’s unfolding now is different. France is approaching its most serious crisis in decades, and the disorder in the streets may be only the start.

Ostensibly the destructive rioting of the past few days is a student protest against underfunded schools. If you believe that, we have a pont over the Seine to sell you. The riots have been encouraged by politicians on the left, as Dominic Green recently reported on the Journal’s Free Expression newsletter.

Labor-union comrades of Jean-Luc Mélenchon, leader of La France Insoumise (“Indomitable France”), stoked the protests that quickly became riots. Mr. Mélenchon—think Bernie Sanders with a Gallic accent—visited the barricades. It feels like a display of left-wing muscle-flexing before next year’s presidential election, in which Mr. Mélenchon will be a candidate.

The violence obscures the related crisis engulfing public finances. France doesn’t have more money to spend on schools because it doesn’t have more money to spend on anything.

It is on the subject raised in the last paragraph of that quote I want to focus. According the American authorities, France has the highest military readiness of any country in Europe as the U. S. reckons such things. France’s present government spending is more than half of France’s GDP. Here is how it spends the money:

As the editors point out France’s taxes and borrowing are high while its tax revenues rise slowly. They summarize the situation succinctly: every incremental tax euro goes to pensions and health care. Neither of the two major candidates running for president of France supports reducing France’s pension or health care spending. If France increases its borrowing, it will further impede its growth.

Where will the money to increase France’s spending on education come from? The French don’t want to reduce pension or health care spending and France cannot reduce military spending without reducing its readiness and that should be discouraged. France’s conundrum is that it bears the costs of previous decisions to increase the public debt and, as I have said before, there are no mulligans in public life.

We are one month away from the midterm elections and I wanted to put my marker down on what I think is going to happen there. The good news for Democrats is that I think that they will take control of the House and, narrowly, the Senate. When the dust has settled I expect Democrats to hold about between 218 and 230 seats in the House and 51 seats in the Senate.

The bad news for Democrats is that it will not be a “wave election” and their majorities will be small and fractious. I define a “wave” as 20 or more House seats.

As evidence for that view I would suggest reading Bryan Bennett’s post at Ruy Teixeira and Michael Bahareen’s Substack. The short version is that the Democratic “brand advantage” is sufficiently low that they are unlikely to overperform it but rather to underperform it. From the linked Substack:

The x-axis is “relative brand advantage” (RBA), which calculates how much higher or lower favorability for the Democratic Party is compared to the Republican Party. For example, in 2006, the average net favorability of the Democratic Party was net +13 and the average net favorability of the Republican Party was net -9; thus, the RBA for Democrats was +22. In 2018, the average net favorability of the Democratic Party was net +1 and the average net favorability of the Republican Party was net -16; thus, the RBA for Democrats was +17. In Q3 of 2026, the average net favorability of the Democratic Party is net -13 and the average net favorability of the Republican Party is net -17; thus, the relative brand advantage for Democrats is just +4. This post does not dwell on the deterioration of the Democratic Party brand, of which I’ve also written about extensively and expressed many concerns about, but I will say this: despite having a narrow RBA, favorability of the Democratic Party dropping by a net 26 points over the last two decades is deeply problematic and should not be overlooked.

and

When charted this way, Democrats have tended to underperform or equal their RBA in the generic ballot—and, ultimately, in the actual election results.

In others words an extraordinary presidency may produce an ordinary midterm correction.

The greater question is how will the Democrats use their new majorities? Specifically, will they devote their efforts to “fighting Trump”, will they focus on tiny minorities, or will they do their best to help ordinary Americans? I’m hoping it will be the last but I’m concerned the Democratic leadership will not be able to subdue or even resist their most extreme subfactions.

Here’s a modest proposal for giving LLM AI service providers exactly what they want:

1. Delay settling the rules until management and early investors have had an opportunity to issue an IPO and cash out.
2. Add a new agency or committee to provide ample opportunity for capture.
3. Grandfather existing providers and models. Apply the expensive new requirements principally to future entrants. Being first then becomes a government-protected advantage.
4. Make compliance expensive regardless of company size. Require licensing, specialized counsel, extensive documentation, audits, and recurring certification. A fixed compliance bill is manageable for a giant and prohibitive for a prospective competitor.
5. Let the leading providers define “safe AI.” Turn their existing practices, evaluation methods, and organizational structures into mandatory standards. Competitors must reproduce the incumbents’ overhead before offering an alternative.
6. Make regulatory compliance a shield against liability. Once a provider has completed the prescribed paperwork, restrict injured parties’ ability to recover damages. Certification becomes protection for the company rather than protection for the public.
7. Place responsibility on customers and downstream developers. Give providers broad latitude to disclaim responsibility through their terms of service, even where customers have little ability to inspect or control the underlying system.
8. Regulate models rather than harmful conduct. Require permission to develop or release a general-purpose tool, instead of concentrating enforcement on fraud, negligence, discrimination, and other identifiable harms. That makes the regulator a gatekeeper to the market.
9. Make open models and local operation presumptively suspect. Require centralized monitoring, identity verification, remote shutdown, and continuing provider control. Make running a model yourself legally difficult enough that renting access from an approved provider becomes the practical default.
10. Preempt state remedies before establishing an effective federal substitute. Remove alternative avenues of enforcement while leaving the national framework unfinished, weak, or dependent on industry cooperation.

Additional points gratefully accepted. Extra points for providing an advantage to foreign competitors.

The editors of the Washington Post provide a list of alternatives for reducing the federal deficit and a handy questionnaire for evaluating your own preferences. They write:

They characterized my responses as “Pragmatic Balancer”:

You cut spending and you raised taxes, spreading the adjustment widely enough that no single group carries it alone. It’s an approach designed to be politically possible while getting the situation under control.

I think they omitted a number of key points. First, there is empirical evidence that a large public debt overhang impedes economic growth. That rules out the easy approach: just outgrow the debt. We cannot safely assume growth will rescue us, especially when accumulated debt may itself weaken growth.

The second is that the graph they display overstates how frequently we have balanced the budget. It conceals how dependent those results were on a temporary demographic advantage. In only one year did the Clinton Administration produce a substantial budget surplus without counting Social Security’s surplus. That should give us pause. The opposite is the case now.