Thursday assorted links
1. Those new service sector jobs.
2. Optimism is correlated with exceptional longevity.
3. “Kosmos has made 7 discoveries so far, which we are releasing today, in areas ranging from neuroscience to material science and clinical genetics, in collaboration with our academic beta testers. Three of these discoveries reproduced unpublished findings; four are net new, validated contributions to the scientific literature.” Link here.
4. The wisdom of Ross Douthat.
5. “Following large exogenous minimum wage increases, schedule unpredictability increases by 20% per week and schedules become even more responsive to weather shocks. This highlights how some of the welfare gains workers realize from a minimum wage may be offset by increased schedule unpredictability.” Link here to the paper.
6. Kevin Kelly’s essentials for independent travel in China.
7. New colossal statue for Alcatraz island? Recall Alex’s proposal.
8. Ross dialogue on feminization with Leah Libresco Sargent and Helen Andrews (NYT).
Creative Stagnation
Legislation requiring cars and trucks, including electric vehicles, to have AM radios easily cleared a House committee Wednesday, although it could run into opposition going forward.
H.R. 979, the “AM Radio for Every Vehicle Act,” would require the Department of Transportation to enforce the mandate through a rulemaking. It passed the Energy and Commerce Committee by a 50-1 vote. Rep. Jay Obernolte (R-Calif.) was the only “no.”
What’s next—mandating 8-track players in every car? Fax machines in every home? Floppy disks in every laptop? If Congress actually cared about emergency communication, it would strengthen cellular networks, not cling to obsolete technology. Congress is a den of old busybodies.
Hat tip: Nick Gillespie.
Addendum: If AM radio is so valuable for emergencies then the market will provide or you could, you know, put an AM radio in your glove box. No need for a mandate. We already have FM, broadcast TV, cable, satellite, cell, and Wireless Emergency Alerts; resilience can be met without specifying AM hardware.
Here Comes the Sun—If We Let It: Cutting Tariffs and Red Tape for Rooftop Solar
Australia has so much rooftop solar power that some states are offering free electricity during peak hours:
TechCrunch: For years, Australians have been been installing solar panels at a rapid clip. Now that investment is paying off.
The Australian government announced this week that electricity customers in three states will get free electricity for up to three hours per day starting in July 2026.
Solar power has boomed in Australia in recent years. Rooftop solar installations cost about $840 (U.S.) per kilowatt of capacity before rebates, about a third of what U.S. households pay. As a result, more than one in three Australian homes have solar panels on their roof.
Why is rooftop solar adoption in the U.S. lagging behind Australia, Europe, and much of Asia? Australia has roughly as many rooftop installations as the entire United States, despite having less than a tenth of its population.
First, tariffs. U.S. tariffs on imported solar panels mean American buyers pay double to triple the global market rate.
Second, permits. The U.S. permitting process is slow, fragmented, and expensive. In Australia, no permit is required for a standard installation—you simply notify the distributor and have an accredited installer certify safety. In Australia, a rooftop solar panel is treated like an appliance; in the U.S., it’s treated like a mini power plant. Germany takes a similar approach to Australia, with national standards and an “as-of-right” presumption for rooftop solar that removes red tape.
By contrast, the U.S. system involves multiple layers of approval—building and electrical permits, several inspections, and a Permission-to-Operate from the local utility, which may not be eager to speed things up just to lose your business. Moreover, each of thousands of jurisdictions has different requirements, creating long delays and high costs.
High costs suppress adoption, limiting economies of scale and forcing installers to spend more on sales than installation. Yet Australia and Germany are not so different from the United States—they simply made solar easy. If the U.S. eliminated tariffs, standardized/nationalized rules, and accelerated approvals, rooftop solar would take off, costs would fall, and innovation would follow.
The benefits extend beyond cheaper power. Distributed rooftop generation makes the grid more resilient. Streamlining solar policy would thus cut energy costs, strengthen protection against disasters and disruptions, and speed the transition to a future with more abundant and cleaner power.
The geopolitical determinants of economic growth, 1960-2019
This paper establishes geopolitical relations as a first-order determinant of economic growth. We construct a novel event-based measure of bilateral geopolitical alignment by employing large language models with web search capabilities to analyze over 440,000 political events across 196 countries from 1960 to 2019. This comprehensive measure enables us to identify the precise timing and magnitude of geopolitical shifts. Exploiting within-country temporal variation, we find that a one-standard-deviation improvement in geopolitical relations increases GDP per capita by 10 percent over 15 years. These persistent effects operate through multiple reinforcing channels: enhanced political stability, increased investment, expanded trade, and productivity gains. Geopolitical factors account for GDP variations ranging from −35 to +30 percent across countries over our sample period, with developing nations exhibiting particularly severe penalties from international isolation.
That is from a recent paper by Tianyu Fan of Yale University, who is on the job market this year. Here is his job market paper on the job market incidence of AI, it actually has a new and significant idea. One of the most interesting profiles I have seen of any candidate this year, he also has a triply-authored piece in Econometrica 2023.
I worry about “affordability politics”
It focuses the listener’s attention on price rather than quantity. To many people it sounds better than “economic growth,” though often for the wrong reasons. I cover related points in my latest Free Press article:
Rather than suggesting beneficial economic reforms, the affordability mantra too often leads to “free lunch” thinking and political giveaways. It is a new form of economic populism—a more general rubric of which, after the volatility and disruptions of the Trump tariffs, I have had enough.
I am not opposed to “affordability” as an abstract concept. For instance, I find food prices shockingly high, even in so-so restaurants, and I wish the prices were lower. And if I were in charge of the economy, I would try to lower costs. But there are only so many ways to do that. One option would be to deregulate the energy sector, easing permitting for solar, wind, and nuclear power. Over a five- to 10-year time horizon, that would lead to cheaper energy and, indirectly, to modestly lower food prices. I would also repeal the Trump tariffs, which artificially inflate the cost of foreign goods. I might also refrain from minimum wage increases, which only cause the price of food to rise further.
But even in the best-case scenario, all of those actions would make food just a bit cheaper than it would be otherwise. I would hardly expect voters to hail my reign as a major triumph, or ask for more of the same. Instead, they might flock to the candidate who promises government-run grocery stores or price controls. Sound familiar?
I do not expect this problem to go away anytime soon, and now it is the Trump people too.
UATX Is Ending Tuition Forever
Thanks to a $100 million gift from Jeff Yass — the largest donation since UATX was founded in 2021 — we’re breaking the chains. His gift marks the launch of a $300 million campaign to build a university that sets students free.
Our bet: Create graduates so exceptional they’ll pay it forward when they succeed, financing the tuition of the next generation. When our students build important companies, defend our nation, advance scientific frontiers, build families, and create works that elicit awe, they’ll remember who made their excellence possible. And they’ll give back.
Here is the full announcement.
My excellent Conversation with Sam Altman
Recorded live in Berkeley, at the Roots of Progress conference (an amazing event), here is the material with transcript, here is the episode summary:
Sam Altman makes his second appearance on the show to discuss how he’s managing OpenAI’s explosive growth, what he’s learned about hiring hardware people, what makes roon special, how far they are from an AI-driven replacement to Slack, what GPT-6 might enable for scientific research, when we’ll see entire divisions of companies run mostly by AI, what he looks for in hires to gauge their AI-resistance, how OpenAI is thinking about commerce, whether GPT-6 will write great poetry, why energy is the binding constraint to chip-building and where it’ll come from, his updated plan for how he’d revitalize St. Louis, why he’s not worried about teaching normies to use AI, what will happen to the price of healthcare and hosing, his evolving views on freedom of expression, why accidental AI persuasion worries him more than intentional takeover, the question he posed to the Dalai Lama about superintelligence, and more.
Excerpt:
COWEN: What is it about GPT-6 that makes that special to you?
ALTMAN: If GPT-3 was the first moment where you saw a glimmer of something that felt like the spiritual Turing test getting passed, GPT-5 is the first moment where you see a glimmer of AI doing new science. It’s very tiny things, but here and there someone’s posting like, “Oh, it figured this thing out,” or “Oh, it came up with this new idea,” or “Oh, it was a useful collaborator on this paper.” There is a chance that GPT-6 will be a GPT-3 to 4-like leap that happened for Turing test-like stuff for science, where 5 has these tiny glimmers and 6 can really do it.
COWEN: Let’s say I run a science lab, and I know GPT-6 is coming. What should I be doing now to prepare for that?
ALTMAN: It’s always a very hard question. Even if you know this thing is coming, if you adapt your —
COWEN: Let’s say I even had it now, right? What exactly would I do the next morning?
ALTMAN: I guess the first thing you would do is just type in the current research questions you’re struggling with, and maybe it’ll say, “Here’s an idea,” or “Run this experiment,” or “Go do this other thing.”
COWEN: If I’m thinking about restructuring an entire organization to have GPT-6 or 7 or whatever at the center of it, what is it I should be doing organizationally, rather than just having all my top people use it as add-ons to their current stock of knowledge?
ALTMAN: I’ve thought about this more for the context of companies than scientists, just because I understand that better. I think it’s a very important question. Right now, I have met some orgs that are really saying, “Okay, we’re going to adopt AI and let AI do this.” I’m very interested in this, because shame on me if OpenAI is not the first big company run by an AI CEO, right?
COWEN: Just parts of it. Not the whole thing.
ALTMAN: No, the whole thing.
COWEN: That’s very ambitious. Just the finance department, whatever.
ALTMAN: Well, but eventually it should get to the whole thing, right? So we can use this and then try to work backwards from that. I find this a very interesting thought experiment of what would have to happen for an AI CEO to be able to do a much better job of running OpenAI than me, which clearly will happen someday. How can we accelerate that? What’s in the way of that? I have found that to be a super useful thought experiment for how we design our org over time and what the other pieces and roadblocks will be. I assume someone running a science lab should try to think the same way, and they’ll come to different conclusions.
COWEN: How far off do you think it is that just, say, one division of OpenAI is 85 percent run by AIs?
ALTMAN: Any single division?
COWEN: Not a tiny, insignificant division, mostly run by the AIs.
ALTMAN: Some small single-digit number of years, not very far. When do you think I can be like, “Okay, Mr. AI CEO, you take over”?
Of course we discuss roon as well, not to mention life on the moons of Saturn…
Wednesday assorted links
1. If Robin Hanson wrote a piece on charitable giving.
2. Are AI products boring? And when LLMs make all cover letters boring.
3. Citation gender gaps in top economics journals.
4. A guide to job market presentations in economics.
5. The AI “Minnesota goodbye.”
6. Big tech cashflow is still bigger than capex.
Incentives matter, and supply is elastic
Now we read that in a Harvard job market paper, by Olivia Zhao with co-author Edward Kong:
Pharmaceutical firms’ incentives to develop new drugs stem from expected profitability. We explore how market exclusivity, a policy that shapes these expectations, influences pharmaceutical innovation. First, we estimate the effects of extending market exclusivity for antibiotics, a drug class where private returns to development historically had not internalized the high social value of new innovation. Using a difference-in-differences approach, we find that a policy that approximately doubled the market exclusivity period for certain antibiotics increased innovative activity. Patent f ilings for antibiotics increased by 47%, and we find suggestive evidence that preclinical studies and phase 3 trial initiations also increased under the policy. Building on these empirical findings, we estimate a structural model of firms’ drug development decisions to predict how market exclusivity extensions of varying lengths would affect innovation in antibiotics and other therapeutic areas. Our reduced form and structural findings suggest that market exclusivity—especially if targeted to high-social-value, low-market-return areas—can be an effective tool for realigning incentives and stimulating innovation, but stress that baseline market size and interactions between market and patent exclusivities affect this policy lever’s impacts.
Here is the paper, I guess for such work, intended for such an audience, one does not refer to history’s greatest villain? And here is their paper on market incentives and antibiotics. Via Nicholas Decker.
Track the 2025-2026 Economics Job Market
“Econ Now aggregates economics PhD job market papers, job postings for economists, and economics conference information all in one place.”
Link here, I have yet to try it.
Tuesday assorted links
The MR Podcast: Our Favorite Models, Session 3: Compensating Differentials and Selective Incentives
On The Marginal Revolution Podcast this week, Tyler and I discuss compensating differentials and Olsonian selective incentives. Here’s one bit:
If you think about the gender wage gap, it’s sometimes said that women earn—it varies—80 cents for every dollar that a man earns. That doesn’t control for anything. Once you control for education and skill and so forth, this gets smaller. Then you also have to control for these quite difficult, elusive sometimes, job amenities. Claudia Goldin, for example, has pointed out that men are much more willing to take jobs requiring inflexible hours.
COWEN: And longer hours, too.
TABARROK: Longer hours and inflexible hours, where your hours are less under your control. That’s what I mean by inflexible. For example, in one study of train and bus drivers, the train and bus drivers are paid equally by gender. There’s no differences whatsoever in what they’re paid on an hourly basis. It turns out that the male drivers, their wages, their returns are much higher because they take a lot more overtime. They take 83% more overtime than their female colleagues. They’re much more likely to accept an overtime shift, which pays time and a half. The male workers also take fewer unpaid hours off. The male salaries on a yearly basis end up being higher, even though males and females are paid equally.
Now, you can roll this back and say that’s because of the unfair demands on women of childcare or something like that, but it’s not a market discrimination. It’s not market discrimination. It’s a compensating differential. Males earn more because they’re more willing to take the inflexible overtime hours and so forth.
One of the most interesting ones is that Uber drivers, male drivers earn a little bit more. Now, obviously, there’s no gender difference whatsoever in how the drivers are paid. It just turns out that male drivers just drive a little bit faster.
COWEN: I’ve noticed this, by the way, when I take Ubers.
TABARROK: On an annual basis, they make about 7% more. Now, again, it’s not entirely obvious that this is even better for the male drivers. Maybe they’re taking a little bit more risk. Maybe they’re a little bit more likely to get into an accident as well.
….COWEN: …Someone gets the short end of the stick. Not only women, but maybe women on average would be more likely to suffer.
TABARROK: I’m not sure it’s the short end of the stick, though I agree with increasing returns, that the people who work longer hours will also earn higher salaries and maybe have plush offices and so forth. Let me put it this way. One of the things which I think the feminism story sometimes gets a little bit wrong is to actually underestimate the value that women get, and that men can get as well, of childcare, of looking after kids, of spending more time at home, or spending more time doing childcare. That can be extremely valuable. At the end of life, who writes on their tombstone, “I wish I could have worked more”?
COWEN: You’re looking at one.
TABARROK: Present company excepted.
Here’s the episode. Subscribe now to take a small step toward a much better world: Apple Podcasts | Spotify | YouTube.
Andrej and Dwarkesh as philosophy
If you follow AI at all, you probably do not need another recommendation of the Andrej Karpathy and Dwarkesh Patel podcast, linked to here:
I hardly ever listen to podcasts, but at almost two and a half hours I found this one worthwhile and that was at 1x (I don’t listen to podcasts at higher speed, not wanting to disrupt the drama of the personalities). What struck me is how philosophical so many aspects of the discussion were. Will this end up being the best “piece of philosophy” done this year? Probably. Neither participant of course is a trained philosopher, but neither were Plato or Kierkagaard. They are both very focused on real issues however, and new issues at that. And dialogue is hardly a disqualifying medium when it comes to philosphy.
Some guy on Twitter felt I was slighting this book in my tweet on the matter. I’ll let history judge this one, as we’ll see which issues people are still talking about fifty years from now (note I said nothing against that book in my tweet, nor against contemporary philosophy, I just said this podcast was philosophical and very good). I’ve made the point before (pre-LLM) that current academic philosophers are losing rather dramatically in the fight for intellectual influence, and perhaps more of a serious engagement with these issues would help. I’ve seen plenty of philosophical work on AI, but none of it yet seems to be interesting. For that you have to go to the practitioners and the Bay Area obsessives.
Notes on Maputo (Mozambique)
An excellent piece by Joseph Levine, do read the whole thing, here is one excerpt:
Yango is the local ride-hailing app and makes life really easy. It’s a Dubai-based company, split off from Russian Yandex in 2024. Today it’s available as a ride-hailing app in 30+ African countries. Yango rides are really cheap, cheaper than taxis in Freetown or Kanifing, and quality was much higher. A 15 minute ride from my house to my office cost $0.23 on a motorbike, $0.58 in a tuk-tuk, $1.05 in a car, and $1.20 in the equivalent of an Uber Black car. I usually ended up taking the cheaper car option, because you can’t easily sit next to or talk to the driver of a tuk-tuk.
Yango eliminates the ability of taxi drivers to price discriminate. When I walked out of the airport, the taxi drivers wanted ~$20 for the drive to my flat, and I talked them down to $10. On Yango, the drive is $3. My general model of digital marketplaces is that they enable price discrimination (e.g., Uber surge pricing), but in Mozambique it’s the opposite.
Drivers are often professionals working a full-time job, or graduate students. I met a dentist, an engineer working at the new Total plant, several accountants. They also made up the majority of the Muslim Mozambicans whom I met. Most Muslims in Maputo are first- or second-generation immigrants from the northern provinces. One driver, Ismael, grew up on Mozambique Island, about 1,000km north of Maputo; it has a population of 10,000, but most of the young people move to Maputo. He came down to study engineering and now works at a terminal exporting natural gas.
Because I wasn’t on the local mobile money system, I paid for each Yango ride in cash. The car drivers rarely had change (tuk-tuk drivers always did). Alongside rating the driver in the Yango app out of 5 stars, you have to reply yes/no to the question “Did the driver ask for extra cash?” This led to the drivers being very reluctant whenever I told them to keep the change, even when I showed them I had already selected “no”. So we often went around to fruit sellers asking if they would break a twenty.
And:
Mozambique shares a timezone with all the other Southern African countries. As the easternmost country in the timezone, this feels really weird! When I arrived, sunrise was at 5:20am, and sunset about 12 hours later. The early sunsets move business and social events earlier throughout the day. I’m a morning person, but it would usually be considered rude to suggest a 7am in-person meeting. Not here!
The Kalashnikov is a prominent national symbol in Mozambique; it’s on the flag. In particularly touristy areas, hawkers would try to sell me flags, jerseys, banners, scarves with ever-larger AK-47s on them.
Recommended, there should be more travel writing like this.
The American economy is showing its flexibility
That is the theme of my recent Free Press column on AI. Excerpt:
The more quickly the world changes, the more the quality of our capital markets matters. And the world is indeed changing quickly: AI will soon be present in virtually every job. Many of us already use it informally for legal and medical advice, research, and even companionship.
And to close:
From the point of view of an investor, it matters whether or not we’re in an AI bubble. But if you are seeking to understand long-term social and economic trends, the bubble question is primarily a matter of short-term interest and timing. It will not decide where the economy is headed long-term.
Instead, what we are seeing is that America, at the drop of a hat, can turn on a dime and reallocate capital on an unparalleled scale, to our great and enduring benefit. Unless you were around to witness World War II, none of us have seen anything like this before. Do not expect the ride to be smooth or predictable, but feel free to sit back and enjoy: This is history in the making.
By some estimates, at least three-quarters of the world’s compute is in the United States.