Category: Uncategorized
The Tyranny of the Complainers
Some years ago, Dourado and Russell pointed out a stunning fact about airport noise complaints: A very large number come from a single individual or household.
In 2015, for example, 6,852 of the 8,760 complaints submitted to Ronald Reagan Washington National Airport originated from one residence in the affluent Foxhall neighborhood of northwest Washington, DC. The residents of that particular house called Reagan National to express irritation about aircraft noise an average of almost 19 times per day during 2015.
Since then, total complaint volumes have exploded—but they are still coming from a tiny number of now apparently more “productive” individuals. In 2024, for example, one individual alone submitted 20,089 complaints, accounting for 25% of all complaints! Indeed, the total number of complainants was only 188 but they complained 79,918 times (an average of 425 per individual or more than one per day.)
What I learned recently is that it’s not just airport noise complaints. We see the same pattern in data from the US Department of Education’s Office for Civil Rights which enforces federal civil rights laws related to education funding. In 2023, for example, 5059 sexual discrimination complaints came from a single individual–from a total of 8151 complaints. Thus, one individual accounted for 68.5% of all sexual discrimination complaints in that year.
In the annual reports for 2022-2024 the OCR identifies what type of complaint the single-individual with multiple complaints was making, a sex discrimination complaint, while in previous years they just give data on the number of complaints from single individuals compared to the total of all types of complaints. I’ve collated this data in this graph which presents totals compared to multiple complaints from a single individual without regard to the type of complaint. Do note, that there are also single individuals filing hundreds of other types of complaints such as age discrimination complaints so the data from more recent years may actually be an underestimate.
In any case, it’s clear that a single individual often accounts for 10-30% of all complaints! These complaints have to be investigated so this single individual may be costing taxpayers millions. It’s as if a single individual were pulling a fire alarm thousands of times a year, mobilizing emergency services on demand, and never facing repercussions.
Does this strategy work? Probably. When complaints are summarized for Congress or reported in the media, are totals presented as-is, or adjusted for spam?
Increasingly, public institutions seem to exist to manage the obsessions of a tiny number of neurotic—and possibly malicious—complainers.
My excellent Conversation with Brendan Foody
Here is the audio, video, and transcript. Here is the episode summary:
At 22, Brendan Foody is both the youngest Conversations with Tyler guest ever and the youngest unicorn founder on record. His company Mercor hires the experts who train frontier AI models—from poets grading verse to economists building evaluation frameworks—and has become one of the fastest-growing startups in history.
Tyler and Brendan discuss why Mercor pays poets $150 an hour, why AI labs need rubrics more than raw text, whether we should enshrine the aesthetic standards of past eras rather than current ones, how quickly models are improving at economically valuable tasks, how long until AI can stump Cass Sunstein, the coming shift toward knowledge workers building RL environments instead of doing repetitive analysis, how to interview without falling for vibes, why nepotism might make a comeback as AI optimizes everyone’s cover letters, scaling the Thiel Fellowship 100,000X, what his 8th-grade donut empire taught him about driving out competition, the link between dyslexia and entrepreneurship, dining out and dating in San Francisco, Mercor’s next steps, and more.
And an excerpt:
COWEN: Now, I saw an ad online not too long ago from Mercor, and it said $150 an hour for a poet. Why would you pay a poet $150 an hour?
FOODY: That’s a phenomenal place to start. For background on what the company does — we hire all of the experts that teach the leading AI models. When one of the AI labs wants to teach their models how to be better at poetry, we’ll find some of the best poets in the world that can help to measure success via creating evals and examples of how the model should behave.
One of the reasons that we’re able to pay so well to attract the best talent is that when we have these phenomenal poets that teach the models how to do things once, they’re then able to apply those skills and that knowledge across billions of users, hence allowing us to pay $150 an hour for some of the best poets in the world.
COWEN: The poets grade the poetry of the models or they grade the writing? What is it they’re grading?
FOODY: It could be some combination depending on the project. An example might be similar to how a professor in English class would create a rubric to grade an essay or a poem that they might have for the students. We could have a poet that creates a rubric to grade how well is the model creating whatever poetry you would like, and a response that would be desirable to a given user.
COWEN: How do you know when you have a good poet, or a great poet?
FOODY: That’s so much of the challenge of it, especially with these very subjective domains in the liberal arts. So much of it is this question of taste, where you want some degree of consensus of different exceptional people believing that they’re each doing a good job, but you probably don’t want too much consensus because you also want to get all of these edge case scenarios of what are the models doing that might deviate a little bit from what the norm is.
COWEN: So, you want your poet graders to disagree with each other some amount.
FOODY: Some amount, exactly, but still a response that is conducive with what most users would want to see in their model responses.
COWEN: Are you ever tempted to ask the AI models, “How good are the poet graders?”
[laughter]
FOODY: We often are. We do a lot of this. It’s where we’ll have the humans create a rubric or some eval to measure success, and then have the models say their perspective. You actually can get a little bit of signal from that, especially if you have an expert — we have tens of thousands of people that are working on our platform at any given time. Oftentimes, there’ll be someone that is tired or not putting a lot of effort into their work, and the models are able to help us with catching that.
And:
COWEN: Let’s say it’s poetry. Let’s say you can get it for free, grab what you want from the known universe. What’s the data that’s going to make the models, working through your company, better at poetry?
FOODY: I think that it’s people that have phenomenal taste of what would users of the end products, users of these frontier models want to see. Someone that understands that when a prompt is given to the model, what is the type of response that people are going to be amazed with? How we define the characteristics of those responses is imperative.
Probably more than just poets that have spent a lot of time in school, we would want people that know how to write work that gets a lot of traction from readers, that gains broad popularity and interest, drives the impact, so to speak, in whatever dimension that we define it within poetry.
COWEN: But what’s the data you want concretely? Is it a tape of them sitting around a table, students come, bring their poems, the person says, “I like this one, here’s why, here’s why not.” Is it that tape or is it written reports? What’s the thing that would come in the mail when you get your wish?
FOODY: The best analog is a rubric. If you have some —
COWEN: A rubric for how to grade?
FOODY: A rubric for how to grade. If the poem evokes this idea that is inevitably going to come up in this prompt or is a characteristic of a really good response, we’ll reward the model a certain amount. If it says this thing, we’ll penalize the model. If it styles the response in this way, we’ll reward it. Those are the types of things, in many ways, very similar to the way that a professor might create a rubric to grade an essay or a poem.
Poetry is definitely a more difficult one because I feel like it’s very unbounded. With a lot of essays that you might grade from your students, it’s a relatively well-scoped prompt where you can probably create a rubric that’s easy to apply to all of them, versus I can only imagine in poetry classes how difficult it is to both create an accurate rubric as well as apply it. The people that are able to do that the best are certainly extremely valuable and exciting.
COWEN: To get all nerdy here, Immanuel Kant in his third critique, Critique of Judgment, said, in essence, taste is that which cannot be captured in a rubric. If the data you want is a rubric and taste is really important, maybe Kant was wrong, but how do I square that whole picture? Is it, by invoking taste, you’re being circular and wishing for a free lunch that comes from outside the model, in a sense?
FOODY: There are other kinds of data they could do if it can’t be captured in a rubric. Another kind is RLHF, where you could have the model generate two responses similar to what you might see in ChatGPT, and then have these people with a lot of taste choose which response they prefer, and do that many times until the model is able to understand their preferences. That could be one way of going about it as well.
Interesting throughout, and definitely recommended. Note the conversation was recorded in October (we have had a long queue), so a few parts of it sound slightly out of date. And here is Hollis Robbins on LLMs and poetry.
Wednesday assorted links
1. Under Australia’s social media ban, it will be harder for museums to reach young audiences.
2. The influence of Claremont Institute fellows.
4. The philosophy of Solvej Balle.
5. “Persuasion of humans in the bottleneck.”
6. Scott Sumner movie reviews. And more.
7. Bela Tarr, RIP (NYT).
8. David Zipper says “Europe Doesn’t Need Driverless Cars.” (FT)
A final remark on AGI and taxation
I’ve noted repeatedly in the past that the notion of AGI, as it is batted around these days, is not so well-defined. But that said, just imagine that any meaningful version of AGI is going to contain the concept “a lot more stuff gets produced.”
So say AGI comes along, what does that mean for taxation? There have been all these recent debates, some of them surveyed here, on labor, capital, perfect substitutability, and so on. But surely the most important first order answer is: “With AGI, we don’t need to raise taxes!”
Because otherwise we do need to raise taxes, given the state of American indebtedness, even with significant cuts to the trajectory of spending.
So the AGI types should in fact be going further and calling for tax cuts. Even if you think AGI is going to do us all in someday — all the more reason to have more consumption now. Of course that will include tax cuts for the rich, since they pay such a large share of America’s tax burden. (Effective Altruists, are you listening?”)
The rest of us can be more circumspect, and say “let’s wait and see.”
Yes, Western Europe will survive recent waves of migration
Over 1.2mn people came to the EU seeking protection in 2015, many displaced by worsening conflict in Syria. There were bitter political feuds in Brussels over asylum, border and relocation policies. January 2016 set a grim record for the number of migrants dying while attempting to cross the Mediterranean.
Now things have changed, as European Commission president Ursula von der Leyen made clear in December when she took the stage at a conference on migrant smuggling. After a major policy overhaul over the past two years, “Europe is managing migration responsibly,” she said. “The figures speak for themselves.”
Irregular arrivals of migrants to the EU recorded by its border agency Frontex dropped by 25 per cent in the 11 months to November 2024, and have been continuously declining since a recent peak of 380,000 arrivals registered in 2023.
New asylum applications have also decreased by around 26 per cent in the first nine months of last year, according to Eurostat data, as fewer Syrians are applying for protection since the fall of the authoritarian regime of Bashar al-Assad in late 2024.
The Venezuelan stock market
Venezuela’s stock market is now up +73% since President Maduro was captured. Since December 23rd, as President Trump ramped up pressure on Maduro’s government, Venezuela’s stock market is up +148%.
Here is the link and chart. And up seventeen percent in the last day, and now some more on top of that. Note the bolivar is down only a small amount since December 23.
I see the reality as such:
a) Immoral actions were taken, leading up to the removal of Maduro, and immoral measures are likely to continue, both from the United States and from various Venezuelan replacement governments.
b) Trump’s actions have been some mix of unlawful and unconstitutional, to what degree you can debate.
c) In expected value terms, the people of Venezuela are now much better off.
It can and should be debated how much a) and b) should be weighted against c). But to deny c), or even to fail to mention it, is, I think, quite delusional.
Effective Altruists, are you paying attention?
Tuesday assorted links
1. 54 observations about Mexico.
2. What do economists mean when they say identification?
3. “Is the European Union now the world’s most underrated libertarian project?”
4. LLAMAS LLAMAS LLAMAS (music from Mexico).
5. Christina Cacioppo reading from 2026. And here is Klara Feenstra.
6. Scott Sumner on neoconservatism.
7. Long-term Treasuries have been losing their safe haven status.
The US Leads the World in Robots (Once You Count Correctly)
If you search for data on robots you will quickly find data from the International Federation of Robotics which places South Korea in the lead with ~818 robots per 10,000 manufacturing workers, followed by China, Japan, Germany and finally at 10th place the US at ~304 robots per 10,000. The IFR, however, misses the most sophisticated, impressive and versatile robots, namely Teslas with FSD capability. Teslas see the world, navigate complex environments, move tons of metal at high speeds and must perform at very high levels of tolerance and safety. If you included Teslas as robots, as you should, the US leaps to the top.
Moreover, once you understand Teslas as robots, Optimus, Tesla’s humanoid robot division, stops being a quixotic Elon side-project and becomes the obvious continuation of Tesla’s core work.
The puzzle of Pakistan’s poverty?
Until 2009, India was poorer than Pakistan on a per capita basis. India truly became richer than Pakistan after 2009 and since then it hasn’t looked back. If trends continue for a decade, India will be more than twice as rich as Pakistan soon…
So why has India pulled ahead in GDP per capita? The reason is simple. Pakistan’s high fertility has driven population growth faster than India’s. In 1952 Pakistan had about one-tenth of India’s population; by 2025 it had grown to nearly one-seventh.
In other words, many of the added Pakistanis have not started working yet, but they are on the books to lower the per capita esstimate. There is much in this Rohit Shinde essay I disagree with, but it is a useful corrective to those who simply wish to sing “policy, policy, policy.” Putting aside its per capita lag, Pakistan has done a better job keeping up with India than you might think at first.
In any case, I am not predicting that trend will continue in the future, I do not think so. So someday this essay might look especially “off,” nonetheless it is worth a moment of ponder.
O-Ring Automation
We study automation when tasks are quality complements rather than separable. Production requires numerous tasks whose qualities multiply as in an O-ring technology. A worker allocates a fixed endowment of time across the tasks performed; machines can replace tasks with given quality, and time is allocated across the remaining manual tasks. This “focus” mechanism generates three results. First, task-by-task substitution logic is incomplete because automating one task changes the return to automating others. Second, automation decisions are discrete and can require bundled adoption even when automation quality improves smoothly. Third, labour income can rise under partial automation because automation scales the value of remaining bottleneck tasks. These results imply that widely-used exposure indices, which aggregate task-level automation risk using linear formulas, will overstate displacement when tasks are complements. The relevant object is not average task exposure but the structure of bottlenecks and how automation reshapes worker time around them.
That is from a new paper by Joshua S. Gans and Avi Goldfarb. Once again people, the share of labor is unlikely to collapse…
Monday assorted links
1. Séb Krier.
2. Institute for Progress class in the economics of innovation.
3. It’s over, people. Alternately, here is Joshua Gans using AI for research.
6. The year in Neanderthals (NYT).
The wisdom of Garett Jones
Two cases for capital share going to zero in a strong AGI world: 1. Capital and labor are more like perfect complements than perfect substitutes, always will be as long as the economy is for humans, and so astronomical increases in capital shrink the capital share to zero.
Why capital share goes to zero in a strong AGI world: 2. Capital & labor are more like perfect substitutes than complements because AGI de facto replicates free humans. Astronomical increases in capital make capital so abundant it’s unpriced like air, so capital share is zero.
The link has a bit more. Of course this is a thought experiment and a reductio, not a prediction. (It seems people in the rationalist community systematically misunderstand how economists communicate? Maybe that is partly the fault of the economists, but they should not so dogmatically believe that the economists are wrong.) And here are very good comments from Basil Halperin.
The bottom line is that it is premature, to say the least, to expect that the share of labor falls to zero or near-zero.
Africa possibility of the day
Call it a statistical quirk if you must. But this year, with a bit of luck, Africa will grow faster than Asia. If the 54 African economies manage to outpace their Asian counterparts, it would be the first time in modern history that this has happened.
To achieve it, African economies will need to grow marginally faster on average than they did last year. In 2025, despite war in Sudan, insurgency in the Sahel and coups in Madagascar and Guinea Bissau, sub-Saharan Africa is expected to have mustered growth of about 4.1 per cent. The IMF expects this to notch up to 4.4 per cent as economies continue to reap the benefits of a weak dollar — good for cutting debt-service payments and easing inflationary pressure — and of high commodity prices, including for gold and copper.
At the same time, the IMF is predicting that, as the Chinese motor whirrs more slowly, the combined economies of Asia will slow in 2026 to around 4.1 per cent.
Here is more from David Pilling at the FT.
U.S. interventions in the New World, with leader removal
I can think of a few. I am not thinking of ongoing struggles, such as the funding of opposition to the Sandinistas, rather I wish to focus on cases where the key leaders actually were removed. After all, we know that is the case in Venezuela today. Maybe these efforts were rights violations, or unconstitutional, and yes that matters. But how did they fare in utilitarian terms?
Puerto Rico: 1898, a big success.
Mexican-American War: Removed Mexican leaders from what today is the American Southwest. Big utilitarian success, including for the many Mexicans who live there now.
Chile, and the coup against Allende: A utilitarian success, Chile is one of the wealthiest places in Latin America and a stable democracy today.
Grenada: Under Reagan, better than Marxism, not a huge success, but certainly an improvement.
Panama, under the first Bush, or for that matter much earlier to get the Canal built: Both times a big success.
Haiti, under Clinton, and also 1915-1934: Unclear what the counterfactuals should be, still this case has to be considered a terrible failure.
Cuba, 1906-1909: Unclear? Nor do I know enough to assess the counterfactual.
Dominican Republic, 1961-1954, starting with Trujillo. A success, as today the DR is one of the wealthiest countries in Latin America. But the positive developments took a long time.
I do not know enough about the U.S. occupation of the DR 1916-1924 to judge that instance. But not an obvious success?
Can we count the American Revolution itself? The Civil War? Both I would say were successes.
We played partial but perhaps non-decisive roles in regime changes in Ecuador 1963 and Brazil 1964, in any case I consider those results to be unclear. Maybe Nicaragua 1909-1933 counts here as well.
So the utilitarian in you, at least, should be happy about Venezuela, whether or not you should be happy on net.
You should note two things. First, the Latin interventions on the whole have gone much better than the Middle East interventions. Perhaps that is because the region has stronger ties to democracy, and also is closer to the United States, both geographically and culturally. Second, looking only at the successes, often they took a long time and/or were not exactly the exact kinds of successes the intervenors may have sought.
Absher, Grier, and Grier consider CIA activism in Latin America and find poor results. I think much of that is springing from cases where we failed to remove the actual leaders, such as Nicaragua and Cuba. Simply funding a conflict does seem to yield poor returns.
Adam Smith markets in everything
Gustavo Dudamel — the Oscar L. Tang & H.M. Agnes Hsu-Tang Music & Artistic Director Designate — conducts the World Premiere of the wealth of nations, a highly anticipated commission from the Pulitzer Prize–winning composer David Lang. Inspired by economist Adam Smith’s 1776 magnum opus, Lang dramatizes this foundational work about economics as inspired by Handel’s treatment of Biblical texts in Messiah. “I want this work to be enjoyable and thought-provoking,” says Lang, “encouraging audiences to consider what we truly value.”
Here is the link.