Category: Uncategorized

Okie-dokie, Casey Handmer edition

I’ve been prototyping an automated peer review project. With current models you get human level peer review for about $0.10 of tokens. For $5 you can rederive every result, check every statistical inference, and audit every paper cited in both directions.

There are about 100m papers in the scientific corpus. Most are not important. Many are sloppy. For structural reasons it’s essentially impossible to change published work, even to correct a typo.

For $500m today you can thoroughly audit every paper ever written. Every p-hacked result, every plagiarist, every fabrication, every citation club, everything.

I don’t have $500m so instead I build a tree, focusing on the 1% of papers that carry 99% of the signal, doing a cheap first pass looking for obvious problems. I can find 90% of the nonsense for a few hundred dollars of tokens.

Any guesses where the biggest surprises will be found?

Here is the link.

The economics of agents

From Patrick Collison:

Assorted things I’ve been mulling over regarding agents in the economy:

• Companies are in some sense the original superintelligence. It’s often the case that they can use their amortized cognitive surplus to advance their interests against those of certain consumers: relative to the optimization capacity of companies, consumers can be inattentive, underinformed, irrational, etc. (Think, say, of some of the classic insurance upsells at the end of a booking process, which is typically priced substantially above what a regular policy would cost.) Herbert Simon introduced the idea of bounded rationality, and firms often end up benefiting from it. Thanks to personal agents, we will probably get somewhat more rational in how we engage with companies.

• How exactly will outcomes change? Well, certain kinds of price discrimination will work less well (how do coupons work if every agent hunts fastidiously for them?), and certain kinds of customer cross-subsidy will change (today, people who forget to cancel subscriptions subsidize those who don’t). Taken across the economy as a whole, I’m not sure what the incidence of all of this will be.

• In corporate strategy, people have long debated the relative importance of distribution and product quality. Perhaps you’ve made something better, but does anyone know about it? Won’t people just continue to buy the ACME Corp Mousetraps? Personal agents will, I think, act as a kind of structural subsidy for product quality. How would markets look different if individuals always spent at least 10 hours researching their purchases? On the margin, focusing on making a product better is, I think, going to become a more effective strategy for companies. This is good!

• If true, a corollary is that information about product quality will become more important and impactful. Today, most people have very limited ability to aggregate and integrate such data, but agents will have insatiable thirst for such signals. (It’s often said that data is becoming more valuable because it helps in the training of models. That’s obviously true, but I think there’s a secondary way in which it’s more valuable simply as a complement to cheaper intelligence.)

• A big question is whether agents will increase superstar effects (rational agents all settle on the same product) or increase dispersion (because everyone has slightly different tastes and preferences, which agents are good at eliciting, finally overcoming the artificial clustering we’ve been subjected to).

• Many internet companies have monetized demand routing, which is in turn often predicated on short attention spans. “I’ll click on the first thing I see.” This has led to some amount of inefficiency, as companies try to propitiate capricious ranking algorithms and incur various placement fees. It’s not all deadweight loss (willingness to pay is itself a kind of signal, which is valuable to buyers), but it’s not great. How will the nature and value of this routing change in a world of personal agents? I don’t know the answer, but I think that the effects might be pretty big.

• Much of what I’m writing hinges on the assumption that personal agents will be on the consumer’s side. I suspect that that will be the winning strategy, but whether it’s true, and what it means to be true, involves a lot of other questions. Taken as a whole, the market is a massively parallelized, decentralized, and highly imperfect value attribution machine, and, taking stock of all of the effects, it seems to me that the quality point may well be the biggest: that agents will change the nature of the backpropagation such that the rewards for better products increase. Maybe this is too optimistic, but I suspect (and hope) that companies that make things that they’re proud of will feel that the universe is a little more partisan in their favor.

In addition to those mechanisms, I might also expect some prices to be more volatile.  Suppliers will attempt to use agents to collude, and the law may not have a simple way of stopping this practice, due to illegibility.  The practice could be illegible and unintended by the merchants themselves.  But collusion of course tends not to be stable.  So (some) prices may swoon upwards for a while, and then take sudden dips as either the collusion breaks down or new entry occurs…

A Sentiment Analysis of Cowen, Hanson, Caplan, and Krugman

Supplied by Bryan Caplan, performed by ChatGPT, excerpt:

So Cowen isn’t well described as either “positive” or “negative.” A much better description is:

High appreciation + high concern + very low emotional agitation.

He seems to think there is an astonishing amount of wonderful stuff in the world and an astonishing number of things worth worrying about.

That combination is unusual.

Insightful throughout.  And have a nice day!

Predictions for economics, given AI

From Ingar Haaland:

With math essentially being delegated to OpenAI, here’s what I predict for economics and the social sciences more generally:

The top tier of research will just become better and it will be normal human-led research where AI is used for scale (e.g. conducting qualitative interviews with relevant populations, running behavioral interventions in the field, analyzing massive text data, etc.)

Field experiments will rise in value and generally making “connections” with firms and being able to run stuff (potentially testing AI pipelines) will be in high demand

Research with administrative data will become even more valuable, but the benefits might be concentrated among some prolific authors who are allowed by the government agencies to use local models to analyze the data at scale. PhD students can probably forget about it

Economic history will have a very exciting boom and there will be an enormous race and big rewards for digitizing archives as a source to both identify new research strategies and document new descriptive facts.

Review and verification systems will become massively improved with AI assistance and it will become standard practice to do a 360 review of the paper + code + data at the *submission* stage.

Research that in principle can already be almost completely outsourced to the AI (download and analyze public data, run simple survey experiments, write theory models) is in for a big shock. This would be very bad from the perspective of researchers who depend on this “bread and butter” research, but I think this type of work will just be outsourced to public agencies who can answer their own questions without the need for “peer-reviewed” research.

All of those make sense to me.

Thursday assorted links

1. War in space?

2. How the math breakthroughs might matter.

3. Short proof of quasi-Riemann.

4. App for finding art exhibitions.

5. Podcast on African economic growth.

6. New and very good book: The Madrid Model: How Freedom and Openness Created an Economic Powerhouse, by Diego Sánchez de la Cruz.

7. This is only the beginning.

8. Democracy thwarted, pufffin to be featured on banknotes.

Why most stereotypes are negative

Stereotypes are a foundational construct in psychological science, often defined as beliefs concerning characteristic group attributes. We present a cognitive-ecological theory of social perception that predicts and explains why such characteristic attributes are likely negative, that is, why most stereotypes are negative. The theory assumes that, cognitively, people characterize groups by attributes that distinguish them from other groups. Ecologically, positive social information is more frequent (i.e., positivity  prevalence), and negative social information is more diverse (i.e., negativity diversity. If negative attributes are less frequent and more diverse relative to positive attributes, then they must be more likely to distinguish one group from another. Consequently, the attributes people see as characteristic of a group are likely to be negative. We illustrate the predicted stereotype negativity across two studies and two existing data sets. We then formalize the theory and illustrate it with simulations showing that rare attributes can be highly diagnostic of group membership while applying to only a minority of group members. Thus, the theory explains why most stereotype content is likely negative, despite the well-documented positivity prevalence, and why stereotypes may be diagnostic and comparatively accurate, despite being descriptively inaccurate. The theory explains stereotype negativity without requiring motivational derogation, essentialist group differences, or a general negativity bias, and it clarifies when positive stereotypes should occur. We also discuss the theory’s relation to dimensional models of stereotype content and consequences for interventions aimed at reducing stereotype negativity.

That is from a recent paper by Christian Unkelbach, Anne Irena Weitzel, and Hans Alves, via the excellent Kevin Lewis.

Brazil election notes (from my email)

From Diego Costa:

“Hi Tyler,

If you’re still interested in the fallout from Brazil’s elections, here are some observations that add texture to the usual narratives:

  • Nine of the 10 candidates who received the most votes for the Lower Chamber are under 40. The exception is 41. Their average age is 31.6. They’re all very online and cultural-war centered. Four are pro-woke and six are anti-woke. The biggest vote-winner, anti-woke Nikolas Ferreira, will still be too young to run for president in 2030.

  • There is a divergence between evangelical influencers and the big evangelical churches. While evangelical politicians with strong personal followings did well, several major big church machines lost seats, particularly in Rio.

  • Brazil is expected to have its most female Senate ever, with 18 of 81 senators, driven by women on the right. All three senators representing the Federal District will be right-wing, including Jair Bolsonaro’s wife, Michelle Bolsonaro.

  • This might be the end of Lulismo. Lula or his chosen candidate has run in every presidential election since 1989, but he has no obvious successor. If he loses the runoff, his personal presidential record will stand at three wins and four losses. Several historical PT historical figures also lost their bids. And Lula, who began his career with votes from the educated middle class in Brazil’s industrial Southeast, is ending it by leading only in the poorest income bracket and the poorer Northeast region.

  • Flávio Bolsonaro’s Liberal Party won 121 seats in the Chamber and should hold 28 in the Senate, a record for any party since the 1988 Constitution. It will nevertheless remain short of a majority in both houses. The Chamber will feature 20 parties, down from the 30 elected in 2018.

  • The Bolsonaro family elected two senators (Jair’s wife Michelle and son Carlos) and two federal deputies (his son Jair Renan and brother Renato). His son Flávio is leading going into the presidential runoff, while his other son Eduardo might become minister. That is possibly the highest concentration of national political power in one immediate family since the Brazilian Republic was founded in 1889.

  • Lava Jato’s leading figures made a strong electoral comeback. Sergio Moro, the former judge who convicted Lula, won the governorship of Paraná in the first round. Deltan Dallagnol, who led the prosecution, won enough votes for a Senate seat, although his eligibility may be contested in the superior courts.

  • Anger over corruption and power grabs by Supreme Court justices was a defining factor in the first-round result. I expect a confrontation between the new Senate and the court over limits on its powers to become a major story in 2027. Brazil’s Supreme Court issued over 100,000 decisions in 2025 alone.”

Tuesday assorted links

1. What is the real rate of Chinese economic growth?

2. An Abundance caucus rolls out a bipartisan agenda.

3. Canada fell to 18th from 9th in global ranking of economic freedom.

4. Will there ever be a Latin Bomb?

5. Why didn’t you use an LLM?

6. “Not only does it now cost France more to borrow than it does Italy and Greece; a rising number of big French companies enjoy lower market interest rates than does the French state.” (FT)

7. Seb Krier!

8. Good review of the new Kevin Roose book (NYT).

9. Research taste in foundation models is rising rapidly.

10. “Shares of South Korean cybersecurity companies surged by as much as 30% Tuesday.” (WSJ)

China fact of the day

With surrogacy illegal in China, an industry of agencies, consultants and fertility clinics has emerged to connect clients with women overseas willing to carry their children. While there is no data on the number of children born to Chinese parents via surrogacy, a recent study showed nearly a third of intended parents for surrogate babies in the US were international and some 40 per cent of those were from China.

Early demand largely came from couples struggling with fertility. But there was growing interest from younger women physically capable of becoming pregnant but reluctant to accept its impact on their bodies and careers, said consultants, clients and doctors interviewed by the FT…

Other destinations include Georgia and Kyrgyzstan, where packages can cost as little as $63,000.

Here is more from Eleanor Olcott at the FT.