OpenAI Operator research preview is now out
I am pleased to have been given an early look at this new project, I think in less than a year’s time many of us will be using an updated version for many ordinary tasks: “Operator is one of our first agents, which are AIs capable of doing work for you independently—you give it a task and it will execute it.” And:
Operator is powered by a new model called Computer-Using Agent (CUA). Combining GPT-4o’s vision capabilities with advanced reasoning through reinforcement learning, CUA is trained to interact with graphical user interfaces (GUIs)—the buttons, menus, and text fields people see on a screen.
Operator can “see” (through screenshots) and “interact” (using all the actions a mouse and keyboard allow) with a browser, enabling it to take action on the web without requiring custom API integrations.
Here is the associated OpenAI blog post. Exciting times.
Thursday assorted links
1. Do growth rates have to decline for aging societies?
2. Conversations with Tyler Boston meet-up.
3. ChinaTalk is expanding into Chinese AI issues.
4. Nate Silver on the zigzag of American politics.
5. Dwarkesh podcast with Adam Brown: “Bubble Universes, Black Holes, & AdS/CFT.”
6. UFO okie-dokie.
7. The politics of LLM content? Tweet storm. David Rozado’s Substack summary. Full report.
8. James Broughel on why people don’t like Dylan so much today. I agree with this descriptive point. But oddly my objection to a lot of music today is how bad it sounds (by my standards), not that the songs are not good.
9. Humanity’s last exam. Here are some of he questions, how well would you do? And NYT coverage. It seems o1 pro was not tested? (I tried the Kant question on o1 pro and it seemed to get it right.) As measured in the study, DeepSeek-R1 (!) does the best.
Milei Implements Peer Approval for Food
Reason: In a sweeping move to overhaul Argentina’s food trade policies, Javier Milei’s administration officially deregulated food imports and exports on Monday. The reform, outlined in Decree 35/2025, seeks to boost foreign trade, cut bureaucratic red tape, and lower consumer prices.
Federico Sturzenegger, head of the Ministry of Deregulation and State Transformation, explained in a post on X that the measure “seeks cheaper food for Argentines and more Argentine food for the world.”
Under the new policy, food products and packaging certified by countries with “high sanitary surveillance” can now enter Argentina without any additional registration or approval processes. These items will be automatically recognized under the Argentine Food Code, cutting down on administrative delays and costs for importers.
The legislation identifies countries such as Australia, New Zealand, Canada, the United States, Israel, Japan, Switzerland, and the United Kingdom, as well as the European Union, as having similar or higher sanitary standards than Argentina.
As Sturzenegger explains in his post, this measure “eliminates requirements to register and authorize: samples, products, establishments, warehouses, utensils, and containers (32 pages of paperwork).”
An excellent “peer approval” policy and one that I have long supported when it comes to the FDA and drug approvals. In fact, since 2010 the US FDA has begun to recognize other countries as having comparable food safety systems. To date, Canada, Australia and New Zealand have been recognized with a Systems Recognition partnership.
Systems Recognition (SR) is a partnership between the U.S. Food and Drug Administration (FDA) and a foreign regulatory counterpart, in which the agencies have concluded that they operate comparable regulatory programs that yield similar food safety outcomes.
Argentina’s policy is unilateral and assumes equivalence if a country uses recognized standards (e.g., Codex Alimentarius) or has high sanitary vigilance while the FDA’s SR policy is bilateral and involves more regulatory harmonization and investigation. I prefer the Argentinian approach. Nevertheless, both programs have the goals of simplifying trade, avoiding duplicate inspections, and helping to prioritize scarce inspection resources.
I encourage the FDA to build on SR for food and extend it to drugs. This could be done in a minor and major way, both of which would useful. The minor reform would be peer approval for already-approved US drugs. In this way, importation could ease drug shortages. The FDA has done this in the past on an ad-hoc basis but it should be made permanent. The second, more major reform, would to extend peer-approval to any drug or device approved by a stringent authority.
*Abundance*, by Ezra Klein and Derek Thompson
The NIH’s own research indicates that Pioneer Award recipients seem to produce influential, highly cited research. But despite efforts to help younger scientists, the share of basic NIH funding going to scientists under thirty-five continues to decline. In the 2004 fiscal year, the High-Risk, High-Reward Research program allocated about $200 million to scientists, a moderate decline since 2019. The amount was an almost negligible fraction — less than half of 1 percent — of the NIH’s annual budget for that year.
Self-recommending, you can pre-order here.
My Conversation with the excellent Joe Boyd
Here is the audio, video, and transcript. Here is the episode summary:
Joe Boyd was there when Dylan went electric, when Pink Floyd was born, and when Paul Simon brought Graceland to the world. But far from being just another music industry insider, Boyd has spent decades exploring how the world’s musical traditions connect and transform each other. His new book And the Roots of Rhythm Remain, is seventeen years in the making, and is in Tyler’s words “the most substantive, complete, thorough, and well-informed book on world music ever written.” From producing Albanian folk recordings to discovering the hidden links between Mississippi Delta blues and Indian classical music, Boyd’s journey reveals how musical innovation often emerges when traditions collide.
He joins Tyler to discuss why Zulu music became politically charged in South Africa, what makes Albanian choral music distinct from Bulgarian polyphony, what it was like producing Toots and the Maytals, his role in the famous “Dueling Banjos” scene in Deliverance, his work with Stanley Kubrick on A Clockwork Orange, his experiences with Syd Barrett and Pink Floyd, how he shaped R.E.M.’s sound on Fables of the Reconstruction, what really happened when Dylan went electric at Newport, how the Beatles integrated Indian music, what makes the Kinshasa guitar sound impossible to replicate, and how he maintains his collection of 6,000 vinyl LPs and 30,000 CDs, what he’ll do next, and more.
There are many, many segments of interest, here is the discussion of Dylan at Newport 1965:
COWEN: Now, as I’m sure you know, there’s a new Bob Dylan movie out called A Complete Unknown. The climactic scene in the movie is all about the Newport Folk Festival in 1965 where “Dylan goes electric.” You were the sound producer there, right?
BOYD: No, I was a production manager. There’s a character in the film who is credited with playing the part of Joe Boyd, the sound engineer. I think the actor who’s supposed to be playing me is at the sound controls. I haven’t seen the picture yet. But I was the production manager.
I was very concerned with the sound because I had been to the ’63 Newport Festival, and I thought it was a fantastic event. It was a never-to-be-forgotten, seeing Mississippi John Hurt and Doc Watson through the fog coming in off Narragansett Bay and Dylan linking arms with Joni and Pete and singing “We Shall Overcome.” But the sound was terrible. All through this festival of ’63, I felt the sound was really crap. You’d have a bluegrass band with a guy playing the fiddle, and you couldn’t hear the fiddle!
The first thing I did when I got behind my desk in June of ’65 in New York at George Wayne’s office was call up Paul Rothchild, the great producer, the guy who produced The Doors and Janis Joplin and so many things. I said, “Hey, Paul, why don’t you come up to Newport and mix the sound?” He said, “Okay, can I have three kin passes?” Meaning for his family: places to stay, passes to every event. I said, “Deal. You got it.”
So, Paul and I together sound checked everybody. Every single artist that appeared at Newport was sound checked in the morning by me and Paul except for Dylan, who we sound checked in the evening, six o’clock, between the afternoon show and the evening show, because Dylan wouldn’t get up in the morning to be sound checked. The guy on the board, the guy whose hands were on those mixers was Paul Rothchild, not me. I’ve never been a sound engineer. I don’t have any technical qualification to be a sound engineer. Neither did Paul for that matter, but he was better at it than I was.
COWEN: The controversy at the time — was it really about Dylan playing electric? Was it just about the poor quality of the sound? Was it about Pete Seeger being upset? What actually happened at that time?
BOYD: I think the controversy — you could see it coming for a month, if not more. To me, you can see it. Have you seen that film, The Other Side of the Mirror?
COWEN: I don’t think so.
BOYD: It’s basically Murray Lerner who shot that film festival, which is about the Newport Festival, has all the footage from ’63, ’64, ’65, ’66. The Other Side of the Mirror is all the Dylan footage from ’63, ’64, and ’65, and it’s fascinating. In ’63, he’s the idealistic singing about a coal miner, and Pete, everybody looking at him like he’s Woody Guthrie.
Then in the ’64, he does a workshop, and Pete Seeger introduces him as the voice of a generation, and he gets up to the microphone, and he sings “Mr. Tambourine Man.” You look at Seeger, who looks puzzled, slightly shocked. What is this? This isn’t a protest song. This isn’t a song you could sing at the barricades. This isn’t a song that’s going to move the youth to revolution. What is this?
That is the beginning of what happened in ’65, is Dylan moving away in a different direction, and he’d already recorded half an album with an electric band in the studio. Just before, in the weeks leading up to the festival, we had The Byrds’ “Mr. Tambourine Man,” electric version, on the Top 40 radio. We had Dylan, “Like A Rolling Stone” with an electric band on the radio.
It was Top 40 big business, mainstream popular culture moving into this delicate little idealistic corner called the Newport Folk Festival, which was based on mostly all-acoustic music and very pure, traditional, or idealistic. Everybody — Pete Seeger and Theodore Bikel and Alan Lomax, and a lot of people in the audience — sensed that this was a bull in a china shop, that this was big-time something moving into this delicate little world.
I was totally on Dylan’s side. Paul Rothchild and I were like, “Yes.” But in retrospect, I see Pete Seeger’s point, absolutely. I would contest — of course, I would, wouldn’t I — contest that the sound was awful. It was just very loud. Nobody had ever heard sound that loud. I think Rothchild pushed up the faders, but it had to be because it was the first moment of rock.
Nobody ever used the word “rock” before 1965. There was rock and roll, there was pop, there was rhythm and blues, but there wasn’t rock. This was rock because you had a drummer, Sam Lay, who was hitting the drums very hard. Mike Bloomfield — this was his moment. He cranked up the level on his guitar. You didn’t have direct connections from amps to the PA system in those days. You just had the sound coming straight out of the amp. So, with the sound of the drums, the sound of the bass, the sound of Bloomfield’s guitar, you had to turn the vocal up so that it would be heard over the guitar.
That escalation of volume is what shaped or defined the future of rock. It became really loud music. That was the first time anybody heard it. It was really shocking. There was probably a little distortion because the speakers weren’t used to it, but it was the kind of sound that would be normal two years later. But that night it wasn’t, and I think Newport and folk music and jazz never really recovered. Every young person who used to become a folk or a jazz fan became a rock fan.
Joe has an encyclopedic knowledge of so many areas of music, and I was honored to do this episode with him. Interesting throughout. Again I will recommend Joe’s new and extraordinarily thorough book And the Roots of Rhythm Remain: A Journey Through Global Music.
*Superagency: What Could Possibly Go Right with Our AI Future*
By Reid Hoffman and Greg Beato. As you might expect, I am in synch with the basic message of this book.
I received a review copy which on the front says “Tyler Cowen edition.” There is a forward, made out to me personally and highly intelligent, relating the book to my own work. There is then a gallery of images of me, very well done by AI image generators.
This is all yet another way in which many books will change, I am all for the innovation.
Wednesday assorted links
1. Why is everyone suddenly reading Middlemarch?
2. Good thread on a great book: Revel’s Without Marx or Jesus.
3. Richard Hanania on the day one Executive Orders. And Doug Irwin on tariffs (FT).
4. Ashley Mehra on Executive AI orders and the defense production act.
5. How many exceptional people are there?
6. The culture that is Ireland: “Protestantism, particularly Presbyterianism, is associated with higher levels of human capital. This denominational effect is remarkably robust, even when accounting for various control variables and alternative modelling specifications.”
Democracy, Capitalism and Monarchy (Yarvin)
The Yarvin interview in the NYTimes magazine illustrates the change in vibes, but frankly, I was bored. It’s amusing when Yarvin tweaks liberals by pointing out that FDR was an authoritarian, but Liberal Fascism did it better.
More generally, much of Yarvin’s thinking is superficial. He thinks, for example, that capitalism works because firms are monarchies.
Yes. I think that having an effective government and an efficient government is better for people’s lives. When I ask people to answer that question, I ask them to look around the room and point out everything in the room that was made by a monarchy, because these things that we call companies are actually little monarchies. You’re looking around, and you see, for example, a laptop, and that laptop was made by Apple, which is a monarchy.
There are many errors here. First, Apple is one firm among countless others most of which do not produce hugely successful products. The big question is not how Apple produces but how Apple is produced. Firms operate as planned entities but they are embedded in and constrained by a broader sea of market competition. It’s the competitive environment that drives innovation, efficiency, and consumer satisfaction.
Second, Mises was closer to the truth when he wrote in Planned Chaos that it’s the consumers not the producers who are monarchs:
In the market economy the consumers are supreme. Their buying and their abstention from buying ultimately determine what the entrepreneurs produce and in what quantity and quality. It determines directly the prices of consumer goods and indirectly the prices of all producer goods, viz., labor and material factors of production. It determines the emergence of profits and losses and the formation of the rate of interest. It determines every individual’s income…The market adjusts the efforts of all those engaged in supplying the needs of the consumers to the wishes of those for whom they produce, the consumers. It subjects production to consumption.
Capitalist firms are disciplined by the necessity of persuading consumers to purchase their products and by competition. Successful firms must continuously meet our desires and needs to survive. When Apple fails to do so, it will face the same fate as countless firms before it—obsolescence and failure.
Markets do hold lessons about governance, but Yarvin draws the wrong conclusions. Democracy, not monarchy, is the political system most analogous to capitalism. As Mises observed, “The market is a democracy in which every penny gives a right to vote.” The analogy works both ways: voting in a democracy mirrors spending in a market. Both systems empower individuals—consumers or voters—to shape outcomes, whether by determining market success or selecting leaders.
Democracy and capitalism are both examples of open-access orders, systems characterized by dispersed power, low barriers to entry, and transparent, universally applicable rules. Such features foster adaptability, accountability, and broad participation—qualities essential to both economic and political success.
The West does face a modest “crisis” of democracy, but the root of this crisis lies in expecting democracy to do too much. We have collectivized decisions which are best left in the hands of individuals and markets but democracy is not a good way of making collective decisions.
Democracy is best understood as a constraint on government power, akin to a Bill of Rights, federalism, and the separation of powers. Democracy’s virtue is in providing a mechanism to remove bad rulers without resorting to bloodshed and its primary value lies in preventing catastrophic outcomes like mass famines and democide—a significant and undeniable merit. Autocracies and monarchies perform much less well on the big issues and, contrary to what many people think, autocracies do not grow faster, win more wars, or perform better on any meaningful comparison that has been investigated.
It is also essential to recognize that “democracy” encompasses a wide range of structures—parliamentary, presidential, constitutional, and more—and there is plenty of room for improved choice within the broader category. We can improve our democracy.
The real lesson from markets is not to create monarchs but to design systems that create choice and competition and allow citizens to remove leaders when they fail.
Hat tip for discussion: Connor.
The 1920s immigration restrictions
The 1920s immigration restrictions in the US did not affect manufacturing wages.
The US immigration restrictions of the 1920s lowered the occupational standings of whites and incumbent immigrants.
US counties with more immigrants excluded by the quotas of the 1920s saw increased in-migration.
During the Great Black Migration of the US, black southerners moved to northern counties, filling roles left by excluded immigrants.
During the Great Black Migration, blacks who migrated to counties with more excluded immigrants experienced greater economic gains.
That is from a new piece by Bin Xie in the Journal of Comparative Economics. Via the excellent Kevin Lewis.
The Stargate Project
The Stargate Project is a new company which intends to invest $500 billion over the next four years building new AI infrastructure for OpenAI in the United States. We will begin deploying $100 billion immediately. This infrastructure will secure American leadership in AI, create hundreds of thousands of American jobs, and generate massive economic benefit for the entire world. This project will not only support the re-industrialization of the United States but also provide a strategic capability to protect the national security of America and its allies.
The initial equity funders in Stargate are SoftBank, OpenAI, Oracle, and MGX. SoftBank and OpenAI are the lead partners for Stargate, with SoftBank having financial responsibility and OpenAI having operational responsibility. Masayoshi Son will be the chairman.
Arm, Microsoft, NVIDIA, Oracle, and OpenAI are the key initial technology partners. The buildout is currently underway, starting in Texas, and we are evaluating potential sites across the country for more campuses as we finalize definitive agreements.
As part of Stargate, Oracle, NVIDIA, and OpenAI will closely collaborate to build and operate this computing system. This builds on a deep collaboration between OpenAI and NVIDIA going back to 2016 and a newer partnership between OpenAI and Oracle.
This also builds on the existing OpenAI partnership with Microsoft. OpenAI will continue to increase its consumption of Azure as OpenAI continues its work with Microsoft with this additional compute to train leading models and deliver great products and services.
All of us look forward to continuing to build and develop AI—and in particular AGI—for the benefit of all of humanity. We believe that this new step is critical on the path, and will enable creative people to figure out how to use AI to elevate humanity.
Here is the full OpenAI tweet, at the very least these are interesting times to be alive. Here are some comments from Jeff Stein.
Lower Artificial Intelligence Literacy Predicts Greater AI Receptivity
As artificial intelligence (AI) transforms society, understanding factors that influence AI receptivity is increasingly important. The current research investigates which types of consumers have greater AI receptivity. Contrary to expectations revealed in four surveys, cross country data and six additional studies find that people with lower AI literacy are typically more receptive to AI. This lower literacy-greater receptivity link is not explained by differences in perceptions of AI’s capability, ethicality, or feared impact on humanity. Instead, this link occurs because people with lower AI literacy are more likely to perceive AI as magical and experience feelings of awe in the face of AI’s execution of tasks that seem to require uniquely human attributes. In line with this theorizing, the lower literacy-higher receptivity link is mediated by perceptions of AI as magical and is moderated among tasks not assumed to require distinctly human attributes. These findings suggest that companies may benefit from shifting their marketing efforts and product development towards consumers with lower AI literacy. Additionally, efforts to demystify AI may inadvertently reduce its appeal, indicating that maintaining an aura of magic around AI could be beneficial for adoption.
That is from a new paper by Tully, Longoni, and Appel, via the magical Kevin Lewis.
Tuesday assorted links
1. Trump repeals the Biden Executive Order on AI.
2. Artificial Intelligence Asset Pricing Models.
3. No concrete news on tariffs is good news.
4. Less post-training on DeepSeek? And other claims about DeepSeek. And yet further claims — ask o1 pro to explain it to you!
5. Are there national security risks to a Tysons casino? If so, what does this say about our other systems?
7. Yes, there is such a thing as a double-landlocked country. Uzbekistan and Lichtenstein.
Keep an Eye on Crypto Regulation
Crypto regulation is likely to change very rapidly. I expect that SAB 121 will be overturned, perhaps even today. Overturning SAB 121 wouldn’t even be controversial because, as I wrote earlier, Democrats and Republicans in the House and Senate both voted to overturn SAB 121 which was saved only by Biden’s veto.
Essentially, SAB 121 made it prohibitive for banks to offer custody services for crypto because that service would then impact all kinds of risk and asset regulations on the bank. Aside from singling out crypto, the SEC is not a regulator of banks so this seemed like a regulatory overreach.
I also hope that the tax rules on staking are simplified. Staking rewards paid in tokens should not be taxed until sold. Just as apples aren’t taxed when they grow on the tree but only when sold.
There are also a number of interesting cases working the way through the courts. Lewellen v. Garland seeks to clarify that crypto projects that don’t custody funds are not money transmitters (they can’t be since they never control funds and have no way of knowing the customer information that money transmitters must provide to the government). The case is particularly interesting to me because Lewellen, the plaintiff, is suing to set up a crypto based assurance contract based, in part, on my work (see also here with Cason and Zubrickas):
Pharos fills an important gap in the existing cryptocurrency financial system. Lewellen has seen that there are “public goods” that many people would be happy to contribute to financially, but only if supporters can be assured that the full amount to fund the public good will be raised. In other words, they will contribute if they can be assured that the public good will be deployed. Partial fundraising for these projects would not be acceptable. Examples include building infrastructure such as a bridge or hospital, building a war monument, funding an event like a festival or conference, funding a medical trial or scientific study, filing an advocacy lawsuit, or funding a movie production or other cultural good. Nobody wants to pay for these endeavours without knowing that others will pay enough to complete the project.
To address this dilemma, Pharos would deploy the concept of “assurance contracts.” An assurance contract is a system in which contributors commit money that is released to the planned recipient only if the fundraising goal is met by a certain date. Otherwise the money is returned to the would-be contributors. By promising a refund if the required amount is not raised, assurance contracts encourage more public goods to be funded through voluntary contributions. See Tabarrok, The Private Provision of Public Goods via Dominant Assurance Contracts, 96 Pub. Choice 345, 345-48 (1998).
What should AI policy learn from DeepSeek?
That is a Bloomberg column of mine from about two weeks ago. I thought it would make more sense to people if I did not blog it right away. Here is one bit:
Now the world knows that a very high-quality AI system can be trained for a relatively small sum of money. That could bring comparable AI systems into realistic purview for nations such as Russia, Iran, Pakistan and others. It is possible to imagine a foreign billionaire initiating a similar program, although personnel would be a constraint. Whatever the dangers of the Chinese system and its potential uses, DeepSeek-inspired offshoots in other nations could be more worrying yet.
Finding cheaper ways to build AI systems was almost certainly going to happen anyway. But consider the tradeoff here: US policy succeeded in hampering China’s ability to deploy high-quality chips in AI systems, with the accompanying national-security benefits, but it also accelerated the development of effective AI systems that do not rely on the highest-quality chips.
It remains to be seen whether that tradeoff will prove to be a favorable one. Not just in the narrow sense — although there are many questions about DeepSeek’s motives, pricing strategy, plans for the future and its relation to the Chinese government that remain unanswered or unanswerable. The tradeoff is uncertain in a larger sense, too.
To paraphrase the Austrian economist Ludwig Mises: Government interventions have important unintended secondary consequences. To see if a policy will work, it is necessary to consider not only its immediate impact but also its second- and third-order effects.
Travis Fisher on electricity privatization (from my email)
I’m a long-time reader and first-time emailer. I just read your blog post from earlier this month about privatizing public services like water and electric utilities.
My colleague Glen Lyons and I are developing a way to introduce more competition into the electricity sector, which some believe to be hopelessly uncompetitive. The idea is to allow new, large electricity customers to form new electricity networks. The change to state statute would officially introduce contestability into many markets, and we think actual rival networks would be built to satisfy new load. They would probably have to be large, electrically, meaning they would likely need to serve multiple large customers (today you can go off-grid, but only to supply yourself).
We aren’t necessarily trying to revolutionize the existing grid or change the way a typical residential customer receives electric service, although there may be beneficial spillover effects for all customers. And the idea is not brand new (I find myself agreeing with many of Wayne Crews’ views from the late 1990s), but the concept’s technological feasibility is at an all-time high, and the flood of new demand from data centers and new manufacturers is creating the right political environment to enact new policies.
Here is my description of the policy: https://www.cato.org/blog/what-would-consumer-regulated-electricity-look
Plus an interview we did recently: https://secondpower.substack.com/p/wacc
Here is the Cato bio of Travis Fisher.