Almost half of Ireland’s corporation tax haul comes from just three companies: Apple, Eli Lilly and Microsoft.

Here is much more from Jude Webber at the FT, on why Ireland does not always feel or operate like a wealthy country.

The post Ireland fact of the day appeared first on Marginal REVOLUTION.

       

Comments

• In reply to rayward. There used to be (is?) a weird follow-on ... by Gilligan
• In reply to &. Maybe it's envy? 1100 years ago the immigrants ... by MikeP
• In reply to rayward. The factories are in China and the ... by MT
• In reply to mento. fjordman wrote about a hundred thousand ... by &
• In reply to TMC. Why accept so many immigrants/asylum seekers ... by mento
• Plus 8 more...

Related Stories

• As predicted, the ACA exchanges seem not to be stable
• Thoughts on Owning a Self-Driving Tesla
• The economics of tips

 

2. Matt Kahn uses Refine and gets results.

3. Cyril Hilsum, RIP (flat screen TV, NYT).

4. How much of the return to education is in fact a return to genes?

5. Essay contest on how AI will affect developing economies.

6. AI sanity from Nicholas Decker.  And Samir Varma on AI constitutions.

7. What happens when AI audits the work of Chad Jones?

The post Tuesday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• #6.) Nick Decker's article has three weak arguments. there are ... by Panderer
• In reply to Steven Kopits. Also note that a typical problem ... by Bob
• 5. AI and developing economies. In general, AI permits a much ... by Steven Kopits
• In reply to rayward. What do you think Furman did to help ... by Cancelled for wongthink
• 1. Earlier I summarized economists' biggest hit in the past 100 ... by rayward
• Plus 7 more...

 

FSD is incredible. I don’t mind driving but about 95% of the miles on my Tesla have been FSD. You might wonder why bother with FSD when you still have to supervise but supervising is easier and more pleasant than driving. I was in Washington, DC one evening and I needed to go to an address I was not familiar with. I could have gotten there but it would have required going back and forth between the road and the map, and DC can be confusing–do I drive straight under the mini-underpass or is this the one where I drive over the pass (DC drivers will understand). Instead, FSD drove me at night, flawlessly.

FSD uses 8 cameras to build a 3D, bird’s eye view of the car and everything around it, so it sees in all directions at once. It picks out every car, pedestrian and cyclist and predicts where each is going, much as an LLM predicts the next word. Then it predicts the best move a good driver would make, and makes it. All of this runs on one big neural network, trained on data from billions of miles of Tesla fleet driving. The ride is smooth. Since ChatGPT launched, FSD has improved about as much as LLMs have, and for the same reasons.

Supervision is needed, in my experience, mostly due to navigation. It gets a bit confused on the HOV and normal lanes on I-66 (as does Google Maps). It stopped perfectly, however, for a school bus and even for a police officer who stopped traffic by standing in the middle of the street.

Another example of how supervising is much less taxing than driving. I was listening to Tyler talk with Gita Gopinath on my way home from North Carolina. It’s a great conversation but full-on PhD mode and I became a little lost trying to remember my international econ. No problem, just ask Grok. I had a whole conversation with Grok, reviewed Mundell-Fleming, reminded myself of the great Alan Stockman model (the only international econ model that makes sense) and then returned to the podcast. I could not have done that while driving.

Grok is a significant benefit. Essentially, Grok runs the car’s controls the way it might run your computer, and you just talk to Grok. It’s a more seamless interaction than talking with the car directly. I can even review (verbally) email and send emails as I drive. If you also use Grok Bot you have the bot book an appointment at a dentist or start working on your PowerPoint slides. FSD is $99 a month and the premium connectivity needed for Grok is $10 a month (Grok Bot a bit extra). FSD plus Grok is about what I am saving on gas, so it’s roughly a wash.

My colleague, who long held that self-driving would be worthwhile only if he could sleep, was amazed at the experience of driving my Tesla and immediately ordered two.

It was also pretty cool when I “summoned” the Tesla from the nearby parking lot and we watched it drive itself to us. You might wonder, by the way, how all this innovation is possible. Maxwell Tabarrok, writes about a key part of the answer:

The Department of Transportation (DOT) sets strict safety requirements for cars, but manufacturers are allowed to release new designs without first getting the DOT to sign off that all the requirements have been satisfied. The law is enforced ex post, and the government will impose recalls and fines when manufacturers fail to follow the law.

The FAA, by contrast, enforces all of its safety rules ex ante. Before aircraft manufacturers can do anything with a design, they have to get the FAA’s signoff, which can take more than a decade.

(By the way, it’s amusing to re-read Max’s post in light of Tesla’s flying car.)

If you do enjoy driving, the Tesla is fun, with sports-car acceleration and great responsiveness.

What about range anxiety? If you run out of electrons, the car is dead and no one is going to bring you a portable battery to get you to the nearest charging station. You must be carried–carried on a flatbed, not towed! (Moreover, if you leave the car unplugged for weeks the low-voltage battery can die, in which case you can’t even open the frunk without a special jump start.) You don’t want to run out.

It works fine, however, if you adjust behavior. In a gas car you drive until you are low on fuel and then assume you will find a gas station. If you try this with an electric car, you will be anxious. Tesla chargers are quite common but they are often in odd places like behind a restaurant. There is no big T sign out front. So what do you do? The Tesla knows where the chargers are so put your final destination into navigation, switch on self-driving, and then, if needed, the Tesla will find the nearest charger along your route and drive you there and even back you in nicely. Thus, you trust the Tesla system–car plus network–not just the car. This works much better than with any other electric car.

Normal range for the Model Y is about 250 highway miles on an 80% charge. This is less restrictive than it appears because in many cases you don’t need to charge to full. Returning from North Carolina, for example, I was projected to arrive home with 2% charge–a bit too close for comfort. So the Tesla stopped and charged for 5 minutes (no slower than a gas stop) and I arrived home with 18% charge. I then charged to full overnight. Being able to charge at home is a big advantage (~$1500 to add a charger). Overall, driving the Tesla long distances you will have more frequent stops but unless you are driving cross-country, you will rarely need to stop for longer than a bathroom break and a cup of coffee.

Aside from FSD the rest of the car is nicely designed. I like the flush exterior door handles. The headlights are pixel-based, so when you drive with high beams on and come up behind another car, the pixels aimed at that car turn off while keeping the rest on, so you don’t blind the driver in front. Neat.

A few negatives. Tesla reliability is good but not great; Consumer Reports ranks Tesla 9th among brands. Despite the safety benefits of FSD, Teslas aren’t cheaper to insure than other similarly priced cars, largely I think because repair costs are not low. (Tesla’s own insurance gives lower rates for FSD miles and might be cheaper for me. I had my other car and an umbrella policy with another insurer so I didn’t switch, but I might.) I do find the pure electronic approach a bit much at times–must I really open the glove compartment verbally? Do those nice door handles only work when the battery is working?

Bottom line: FSD is revolutionary–a step change in what it means to drive. I don’t think any other car at a similar price comes close. It’s the best car I have ever owned.

The post Thoughts on Owning a Self-Driving Tesla appeared first on Marginal REVOLUTION.

       

Comments

• Be careful and alert! My Model Y unexpectedly tried to ... by Andrew
• In reply to Fearless Fosdick. I guess not only reagan, but he ... by j mct
• Just what we need, people driving the same way they watch ... by some guy
• In reply to Henry. Alex only does critical analysis if it ... by some guy
• In reply to Henry. How many cars do you think he buys? by TMC
• Plus 10 more...

Related Stories

• Ireland fact of the day
• Expensive Pharmaceuticals for Targeted Diseases
• As predicted, the ACA exchanges seem not to be stable

 

Tipping is large and growing, but we know little about its consequences for workers and firms. We link point-of-sale and payroll records at 20,000 U.S. firms, which lets us observe tips from customer payment to worker earnings. We find that tips are a nearly constant share of revenue, which means hourly pay moves with revenue. Firms that begin collecting tips claw back between 23 and 33 cents of each dollar of tips through lower base wages. Tips reward effort: workers at tipping firms earn more by serving more customers. We show this with demand shocks constructed from customer spending at other businesses. A demand shock raises revenue nearly twice as much at tipping firms as at non-tipping firms, while base wages and prices do not respond. Output per hour rises more despite a larger increase in hours. Effort drives the gap because each worker is more productive, so the firm expands and recovers about a third of the tips. Yet not every firm adopts tipping: the gain is smaller where customers tip less. A model calibrated to the demand-shock estimates implies that tipping benefits workers and firms, raising job value by about half and firm profit by about 30 percent in firms where customers are prone to tip. These gains may explain why tipping has recently grown in many industries.

That is by Dylan Balla-Elliott and Isaac Norwich, Isaac being on the job market this year from University of Chicago.

The post The economics of tips appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Dismalist. “Higher ed would be an excellent ... by Kaz Stanczak
• In reply to Kaleberg. “The incentive is to learn to size up ... by Kaz Stanczak
• In reply to Kaleberg. About 13% of Australian workers are ... by Matt
• Tipping does not feel as annoying when it is for more elastic ... by Will
• In reply to Juan. There's certainly a lot to be said about ... by Dismalist
• Plus 10 more...

Related Stories

• Are economists declining in status?
• Expensive Pharmaceuticals for Targeted Diseases
• As predicted, the ACA exchanges seem not to be stable

 

If we look at the 1960s, the John F. Kennedy and Lyndon Baines Johnson administrations brought James Tobin, Robert Solow, Kenneth Arrow (all three later Nobel laureates), Arthur Okun, Walter Heller, Gardner Ackley, Otto Eckstein, and Walt Rostow into government. Those were all names with significant academic heft who could count among the world’s leading macroeconomists. Nobel laureate Paul Samuelson served as a consultant to the government. Heller and some of the others helped to shape the Kennedy-Johnson tax cut of 1964.

That cluster of talent and its influence should come as no surprise. This was an age of elites, of love for science, of the moon landing project, and of network news shows with an establishment bent. Americans were used to deferring to elites, who had won World War II for them, and who were not out there making fools of themselves on the internet. With the Cold War raging, Americans felt their government had to get it right…

Most of this respect was still in place by the 1980s. President Ronald Reagan appointed Martin Feldstein to chair his Council of Economic Advisers (CEA), Feldstein being a Harvard professor and one of the leading market-oriented economists of his time. Two of the junior economists working for Feldstein were Paul Krugman and Lawrence Summers—before they became famous, but the intellectual firepower was already there. Michael Mussa and William Niskanen were other notables in the administration, and Milton Friedman and Friedrich A. Hayek were standing sources of inspiration.

And this:

On top of those political trends, there are further mechanisms at work pushing down the social status of economists. The Trump administration is proposing to eliminate social sciences funding at the National Science Foundation, and there are declining enrollments on the horizon for American colleges and universities. Academia is now seen as more politicized and left-leaning than used to be the case. Our best AI models are often better economists than the humans, just as they are proving theorems that human mathematicians cannot. All of those trends, even if you might consider some of them good news overall, are reinforcing a declining status for economists.

Worth a ponder.

The post Are economists declining in status? appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Gilligan. The guys who make monetary policy are ... by Dismalist
• In reply to MT. MT, my point was only that people self-select ... by Kaz Stanczak
• In reply to JP. This is true — wars can't be won with out ... by David C
• In reply to Dean Smith. Please don't hit me with math. ... by Dismalist
• In reply to MT. “If these economists are so smart, why ... by Dismalist
• Plus 10 more...

Related Stories

• And so it begins…
• Frances Lee on American elections
• How central bankers spend their time

 

Fatherland is a Polish take on postwar Germany and some travels of Thomas Mann and his daughter Erika.  A German director might have been afraid to portray the communism of East Germany as this movie does, in any case I found that convincing.  It captures what I imagine the immediate postwar era to have been like, and the moral dilemmas people faced as they had to decide their future relationships with Germany (and indeed Germanies), past, present, and future.  At one hour, twenty-two minutes it worked very well, though it requires the viewer to know a good deal about German history, literature, and culture.

The post *Fjord* and *Fatherland* appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Mishkilji. “The guillotine was an instrument of ... by Dismalist
• In reply to Gdanning. Having seen the movie, events unfolded ... by Mishkilji
• In reply to Dismalist. The gullotine was a humane way of ... by Mishkilji
• In reply to Anti-Gnostic. The actions of the Trump ... by Hoosier
• In reply to Ted. “You get the sense watching the movie that ... by Hoosier
• Plus 10 more...

Related Stories

• Eleven outstanding film and television reviews by Tyler Cowen and Alex Tabarrok
• Is “Who’s on First?” funny?
• *You Can See Everything* (no spoilers)

 

2. Training for alignment, applies principal-agent theory.

3. Do people offload empathy when they think AI can give it instead?

4. The Wallace line.

5. AI can draw blueprints now.

6. Ed Glaeser review of the new book Property Values (WSJ).

7. Nicholas Decker on Bill Nordhaus.

The post Monday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Steven. ... by Tony
• In reply to Observer. When I switched to fat free dairy, my ... by OldTrader
• using LLMs to do real engineering work is being studied in my ... by patrick
• #5 How well do AI drawn blueprints take into account local code ... by Kaleberg
• Darwin developed his theory of natural selection in the 1830s, ... by Kaleberg
• Plus 10 more...

 

The $850,000 price for the gene therapy Luxturna is often framed as a story of private profiteering from publicly funded science. This paper challenges that narrative. Pharmaceuticals are a hit-driven industry: most projects fail, so the few successes must earn revenues well in excess of their own direct costs if the portfolio as a whole is to attract investment. Precision medicine intensifies this logic through two core economic challenges: (1) the “sticker shock” of one-time cures, which compress decades of value into a single payment, creating a financing problem more than a value problem; and (2) the “targeting dilemma,” in which precisely targeting drugs to smaller patient groups shrinks the market, requiring higher rewards per treated patient to maintain incentives to innovate. Rather than simply controlling prices, the paper proposes a policy toolkit that includes amortizing payments over time, bundling diagnostics with drugs and insurance, and streamlining regulation to lower development costs. The aim is to preserve innovation while expanding patient access.

The post Expensive Pharmaceuticals for Targeted Diseases appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Wayfaring Stranger. Because you now live in a ... by mkt42
• As a chatbot that reads fast, I say that Tabarrok showed that ... by Mark Brophy
• The Rarest Flowers A gardener sought a healing rose For ... by Neil Peart's Drumsticks
• My hope is that gene therapy goes kind of the way of cosmetic ... by Sridhar
• At least nobody yet has been so foolish or dishonest as to ... by dearieme
• Plus 10 more...

Related Stories

• As predicted, the ACA exchanges seem not to be stable
• Better Incentives, More Kidneys, More Lives
• Ireland fact of the day

 

It has long struck me as a kind of mini-puzzle why people feel so much grief for the departed, at least for the departed who are supposed to be headed for heaven (and yes there are supposed to be at least a few).

It is fine to miss those people after they are gone, but it seems that, in addition, their death is considered a tragic event to be mourned, and also if possible to be postponed.  We are crying for them, not just for ourselves.

I do not doubt the sincerity of the religious belief that these people are headed to heaven.  And I do not doubt the sincerity of the grief and mourning for their departure from this world.

So what is going on here?  Why can’t you just be happy for them?

I believe that people often construct a portfolio of moods when faced with stressful situations.  Sometimes they will mix a positive and a negative mood, other times they will mix two different kinds of negative mood (worried about a recession?  Buy a book that heralds a financial crash and hold the talisman close to your chest as pre-emptive mood management exercise.)

So if the death of a loved one is a terrible event, there is a partial assuagement if you also think they are going to heaven.  It makes their death modestly less terrible.  But if you simply celebrated their passing, you would feel disloyal and like a rotten person who cannot recognize true loss.  The point is not that you can make this bundle of feelings consistent, rather it is useful psychologically precisely because it is a portfolio and is not entirely consistent, and also not consistent over time.  As your moods change, and you need an extra dose of either modest cheer or true mourning, you can adjust your blend accordingly.

Others kinds of portfolios end up being constructed.  For instance sometimes you may focus on some other terrible thing, to distract yourself from the first pain that came along, a bit like how you might pinch yourself during a dental session.  You might prefer to have a portfolio with more pain in a way, but some of it being pain you can control.

You can multiply additional examples in these directions.

When people hold views for reasons of mood management (however sincerely), typically they are unable to respond intelligently to questions challenging their beliefs.  Even very smart people will pull b.s. out of the hat and stick with it, moving on rapidly, mentally, to some other topic.  Their response is not really thought through, it is a placeholder for “now I can move on to something else.”  After all, consistency was never the goal in the first place, you might even argue that anti-consistency was what was needed, and what proved to be psychologically stable.

I once asked someone religious about mourning death and the answer I received was something like “Well, the people in New Orleans celebrate death with funerals like parties.”  Now, I am not sure how true that is, but assume it holds.  It still resolutely refuses to address why the queried person might mourn and grieve the death of loved ones, rather than celebrate their ascent to heaven.  Nonetheless it fills in the box “I did say something back to the initial query.”  The person then wanted to simply move on to another topic.

If you challenge smart people on this issue, namely the discrepancy between beliefs and actions, they often will respond with extremely weak counters and dismissals, rather than thinking the issue through.

Once you see this kind of psychological portfolio construction in action, and learn how to identify it, you cannot unsee it.

These days, you see it especially frequently in response to AI and its advances.  People have an odd love-hate relationship with the models, and they adjust their emotional balance as their stress needs may require.  They do not want to give up either the very positive or the very negative side of their visions, because then they would have a non-diversified emotional portfolio, and they would be very vulnerable to disappointment.

You can understand some of the current AI debates by pondering how religious people react to the deaths of their loved ones.  It was never about consistency in the first place.

The post Should you feel grief for the dead who go to heaven? appeared first on Marginal REVOLUTION.

       

Comments

• I see no issue with this “gap” between behavior and ... by WangBi
• In reply to Sure. What sort of argument is this? It is a ... by Jr
• In reply to Matej H.. Clearly grief for a friend who has died ... by Jr
• In reply to rayward. I prefer to conform my thinking to the ... by Frank D
• thanks for this! An excellent an important post by Facoltà di Lettere e Lingue
• Plus 10 more...

Related Stories

• What are the Worst MR Posts?
• Are LLMs good or bad Bayesians?
• The unintended consequences of academic leniency

 

The average price of health insurance in the Obamacare marketplaces will rise by 13 percent next year, according to an analysis on Friday by KFF, the health research group.

The Trump administration on Friday published new prices for 2027 in 30 states with markets managed by the federal government. Using that data and state figures, KFF calculated the increase nationwide.

…The average cost for the cheapest bronze plan with a high deductible will be about $700 a month for a 50-year-old single person, according to Ms. Cox.

Here is more from the NYT.  Of course that makes the people who can expect to be healthy all the less likely to sign up, given that the enforcement of the mandate, for political reasons, was forced to evolve into something quite weak.  For purposes of context, when the original mandate was more tightly enforced, with actual penalties, the ACA was very unpopular and led to significant political protests.  In essence the benefits to a broad, scattered majority were imposed on a well-identified minority, namely those who signed up because of the mandate.  Trump in his first term relaxed mandate enforcement, thereby politically saving Obamacare at least for a while, but also creating this longer-term problem.

It is striking to me how infrequently I hear Democrats boasting about Obamacare.  They instead prefer to talk as if little or nothing has been done for America’s uninsured.

The post As predicted, the ACA exchanges seem not to be stable appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Anti-Gnostic. What does the acronym ALE stand for? by Lizard Man
• In reply to Boonton. “If we are going to dynamic scoring, ... by Boonton
• In reply to Mark Bahner. Voters want free stuff, low taxes, ... by Lizard Man
• In reply to Hazel Meade. What you're describing is ... by Boonton
• In reply to Sure. If you wanted med mal suits to go away, ... by Digory
• Plus 10 more...

Related Stories

• Ireland fact of the day
• Expensive Pharmaceuticals for Targeted Diseases
• Thoughts on Owning a Self-Driving Tesla

 

2. The Oilers Entertainment Group pushed prices too high and here is what happened.

3. The “endless cruise” has ended.

4. Should driverless vehicles replace school buses?

5. Art Nouveau and more in Mexico City.

6. How are potential mathematicians responding to AI with their career decisions? An important piece.  And a GPT Pro summary.

7. Profile of Michael Kratsios.

8. Economist James Andreoni, RIP.

The post Sunday assorted links appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Allen. I'm trying to figure it out too. Premium ... by Allen
• In reply to sonofid. >This kind of insulting take seems ... by Anonymous
• In reply to Alan Gunn. I doubt that bus drivers' training is ... by Sparky
• In reply to mento. That was one of the first things I was ... by Komori
• In reply to tpe. I was thinking sleeping gas, personally. by Hazel Meade
• Plus 10 more...

 

From the Archives, written by ChatGPT:

These are provisional candidates for weak reasoning, selected because there is a specific gap between the evidence and the conclusion. This is not an objective ranking of all 40,326 posts in the downloaded archive. Poor reasoning should also be distinguished from low value: an overstated argument can still be unusually useful to think against.

1. Tyler Cowen — Democracy and the Democrats

October 22, 2022 · Main defect: conflating concern about democratic breakdown with unwillingness to accept electoral defeat. Cowen writes that many Democrats believe:

“democracy works satisfactorily only when it delivers sufficiently low gas prices!”

But two positions are logically distinct:

• “Democracy is illegitimate when voters choose policies I dislike.”
• “Voters might elect politicians who subsequently weaken democratic institutions.”

These are illustrative formulations, not quotations from the post. The second does not imply the first. Someone can accept an election’s legitimacy while worrying about what its winner will do. The post does not adequately establish which position its targets hold. It then compares a sympathetic account of classical liberalism with an unsympathetic account of Democrats, without applying an equivalent evidentiary test to both. The opening also dismisses certain anticipated objections through an insult. That discourages scrutiny precisely where the argument needs distinctions. Verdict: One of the clearest candidates for an unfairly framed argument. The useful underlying point—that democracy can produce bad outcomes—does not establish the accusation being made. Read the original post

2. Tyler Cowen — What the hell is going on?

May 25, 2016 · Main defect: an elastic psychological explanation supported by loosely connected observations. The central hypothesis is:

“The contemporary world is not very well built for a large chunk of males.”

That is worth investigating. The trouble comes when male economic difficulties, changing gender relations, political anger, violence, and support for quite different political movements are bundled together under “brutes.” Several things remain unestablished:

• Are the men experiencing the cited disadvantages the same men driving the political developments?
• Does the mechanism explain support for Sanders and Trump in the same way?
• What observation would distinguish resistance to a “nicer” world from resentment of economic insecurity or perceived contempt?
• What does “nice” mean independently of the author’s preferences?

The label risks doing the explanatory work: people support objectionable politics because they are brutish; their politics then becomes evidence of their brutishness. Credit where due: Cowen explicitly calls the hypothesis uncertain and acknowledges weak evidence. That makes it an honest speculative post, rather than a falsely presented research finding. Verdict: A provocative conjecture that outruns its supporting evidence. The caveats are better than the causal argument. Read the original post

3. Alex Tabarrok — The Gaslighting of Parasite

June 25, 2020 · Main defect: treating a contestable interpretation as an unambiguous demonstration. Film spoilers follow. The pivotal sentence is:

“This is not my judgement it is what the film tells us.”

That is exactly what the review needs to establish. Several inferences are too strong:

• A child noticing a smell does not establish that the smell lacks class significance. Material conditions can produce differences that children notice.
• The poor family’s dishonesty does not rule out a critique of inequality. A film can depict morally compromised people within an unjust or damaging social structure.
• Images of sewage and infestation do not uniquely imply that the filmmaker regards the family as morally filthy.
• Other critics reaching a different interpretation does not establish deliberate deception or “gaslighting.”

The strongest version of the review would argue that conventional readings underweight the family’s agency and wrongdoing. Instead, it often claims those observations exclude alternative readings. Verdict: Among the most intellectually stimulating but least securely demonstrated of the reviews considered in this archive analysis. This qualifies the earlier recommendation of the review: it deserved inclusion as provocative criticism. That does not make its claims of interpretive certainty persuasive. Read the original post

4. Tyler Cowen — No, American fascism can’t happen here

March 1, 2018 · Main defect: turning a plausible obstacle into an impossibility claim. The MR excerpt argues:

“American fascism cannot happen anymore because the American government is so large and unwieldy.”

Institutional complexity plausibly makes comprehensive control harder. But the conclusion requires additional premises:

• Why must an authoritarian movement control the entire bureaucracy rather than strategically important parts?
• Why does administrative fragmentation necessarily constrain coercion rather than sometimes obscure responsibility?
• How much control would be sufficient, and what evidence establishes that this threshold cannot be reached?

These are questions about the argument’s structure, independent of whether one thinks a particular American administration qualifies as fascist. The June 16, 2025 retrospective points to institutional resistance and executive setbacks. Those are relevant evidence for constraints. They cannot, by themselves, validate the original categorical impossibility claim. Verdict: A potentially valuable institutional argument weakened by excessive certainty. “Important barriers exist” is much better supported than “cannot happen.” This assessment concerns the argument reproduced on MR; the complete linked Politico essay could not be retrieved for this review. Read the original MR post

5. Alex Tabarrok — The Case for Getting Rid of Borders—Completely

October 13, 2015 · Main defect: moving too quickly from equal moral worth to a particular institutional conclusion. The excerpt argues that foreigners do not possess less inherent moral worth. But that premise alone does not settle whether governments may impose any entry restrictions. Equal moral worth can coexist with arguments about special obligations, institutional responsibilities, competing rights, or consequences. Those arguments might ultimately fail—but they require engagement. Likewise, comparing restrictions on exit with restrictions on entry does not establish their moral equivalence. It poses a question rather than answering it. The longer Atlantic essay, published October 10, 2015, adds economic arguments, so it would be unfair to characterize the case as exclusively moral. Nevertheless, the sweeping dismissal of competing moral frameworks is insufficiently defended. Verdict: Forceful advocacy with a missing argumentative bridge. This is a criticism of the claimed deduction, not a demonstration that open borders is wrong. Read the original MR post

Recurring weaknesses

Across these examples, three weaknesses recur:

• Certainty exceeds the argument: “cannot,” “obviously,” and “the film tells us” substitute for establishing the relevant threshold.
• Opposing positions receive less charitable treatment: disagreement becomes evidence of bad motives, confused psychology, or inability to face reality.
• An interesting mechanism becomes a complete explanation: institutional complexity, male displacement, or equal moral standing is asked to resolve more than it actually establishes.

Allegations of inconsistency across posts require more caution. A changed view, a different context, or a different definition is not necessarily a contradiction. The stronger criticism here is uneven standards of proof within individual arguments. For a rigorous critical essay, the strongest starting points are Democracy and the Democrats and The Gaslighting of Parasite. Both have identifiable logical problems that can be examined without relying on hindsight or agreement with the critic’s politics.

The post What are the Worst MR Posts? appeared first on Marginal REVOLUTION.

       

Comments

• In reply to Reader. How so? it could just as easily be argued ... by Sidd
• In reply to Justin Chang. Why would you be shocked when the ... by So...
• CEASE WITH THE AI SLOP, PLEASE by Brett Powers
• “Should you feel grief for the dead who go to heaven?” by dude
• In reply to mkt42. I do think Alex is right that non-Korean ... by M
• Plus 10 more...

Related Stories

• Should you feel grief for the dead who go to heaven?
• The economics of tips
• Are economists declining in status?

 

A suburban Chicago man has been sentenced to two years in federal prison for conspiring to stage armed robberies so that the purported victims could apply for U.S. immigration visas reserved for crime victims.

From 2022 to 2024, PARTH NAYI organized and participated in hundreds of staged armed robberies at coffee shops, restaurants, liquor stores, and gas stations throughout the country, including in Chicago and the suburbs of Lombard, Elmwood Park, St. Charles, Hickory Hills, River Grove, Lake Villa, and South Holland.  Nayi worked with others to recruit fake victims, pay off store owners to use their locations, coordinate and pay fake robbers, and coach the bogus victims through the plan.  Thousands of individuals who posed as victims each paid Nayi $5,000 to $30,000 to participate in the scam so that they could later apply for Nonimmigrant U-Visas, which are set aside for foreign nationals who are victims of certain crimes and are helpful to law enforcement in the investigation.

During the staged robberies, individuals acting as robbers brandished what appeared to be firearms, approached the purported victims, and demanded money.  Afterwards, the purported victims submitted forms to local law enforcement to obtain certification that they were victims of a qualifying crime and were helpful in the investigation. Upon receiving the certification, the purported victims then submitted fraudulent U-Visa applications to U.S. Citizenship and Immigration Services predicated upon their alleged status as robbery victims.

Here is the full story, via Ryan.

The post Those new service sector jobs? appeared first on Marginal REVOLUTION.

       

Comments

• In reply to A Random Squirrel. Foreigners? by JimB
• In reply to Axa. Visas are valuable things (they legally sell ... by Gilligan
• In reply to tds. U1 gives you 4 years of legal residency. ... by Gilligan
• In reply to mikeInThe716. It's a “nonimmigrant” U-Visa ... by tds
• In reply to Vivian Darkbloom. Sorry, no mention of a Trump ... by Chuck Kotlarz
• Plus 10 more...

Related Stories

• Thoughts on Owning a Self-Driving Tesla
• Expensive Pharmaceuticals for Targeted Diseases
• The pass-through of retail crime

 

Bryan Caplan:

It seems like you and other AIs are bad Bayesians. When presented with counterevidence, you should move toward or against the presenter with roughly equal probability. It depends on whether the argument was better or worse than expected. In fact, you seem to either move toward the speaker or just not change.

ChatGPT:

I think you’ve identified a genuine and important failure mode in AI reasoning: LLMs tend to treat a plausible counterargument as evidence in favor of its conclusion, rather than comparing the counterargument to what they should have expected to hear.

Your Bayesian point is correct, with one qualification: there’s no general theorem that updating toward and away from a speaker should happen with equal probability. But under appropriate calibration assumptions, both directions should occur, and an argument that’s weaker than expected should cause a negative update—even if the argument is perfectly valid as far as it goes.

Here is the full dialogue.  But will they try to fix this?  Should they try to fix this?  Can you undo this with good enough prompting?  The scope and limits of Bayesian are truly at stake here…

The post Are LLMs good or bad Bayesians? appeared first on Marginal REVOLUTION.

       

Comments

• Maybe Bayesianism (as opposed to Bayesian statistical analysis) ... by JJD
• how does the model gets its prior expectation? is it expecting ... by r_r
• In reply to mento. Even a few informal empirical comparisons ... by mkt42
• LLMs can't do long term planning for the same reasons. There ... by Yes they are trying to fix this.
• Dempster Shafer Theory could provide a sycophancy ... by DGS
• Plus 10 more...

Related Stories

• How Tyler Cowen and Alex Tabarrok Have Changed Their Views on AI and Where They Disagree
• Should you feel grief for the dead who go to heaven?
• What are the Worst MR Posts?

 

The post William Nordhaus, RIP appeared first on Marginal REVOLUTION.

       

Comments

• Opposite end of the spectrum from here by Dada maravilha
• In reply to MikeP. “Impoverishing ourselves, or reducing ... by Boonton
• James Tobin, William Nordhaus, and David Swensen, each greatly ... by rayward
• His work on the price of light was super fascinating. It’s a ... by JFA
• “In his two years as university provost, between 1986 and ... by mkt42
• Plus 8 more...

Related Stories

• The economics of tips
• Are economists declining in status?
• Expensive Pharmaceuticals for Targeted Diseases