Learn to Ode 2026

Joke: https://x.com/TheLincoln/status/2027215235103207693

Writing about the Citrini Research report on February 28 feels like a being 6 years behind (it was only 6 days ago).

THE 2028 GLOBAL INTELLIGENCE CRISIS: A Thought Exercise in Financial History, from the Future”

Two things the white-collar chattering class fears is that their jobs will disappear or their stock portfolios will crash. The Citrini note put that feared scenario in a picture frame so we could stare at it, like Annie Jacobsen’s book on nuclear war. The post imagines a 2028 scenario: AI automates white-collar work, companies collapse, private credit blows up, mortgages default, unemployment hits 10%.

Brian Albrecht responded: “We don’t need to just make up fantasy stories: Using economics to read Citrini Research’s AI”

Tyler encouraged us to consider a response put out by Citadel “The 2026 Global Intelligence Crisis

Even cognitive automation faces coordination frictions, liability constraints, and trust barriers. It seems more likely that AI will be a complement rather than a substitute for labor is many areas.

One barrier to AI taking all the white-collar jobs as quickly as 2028 is just physical scaling constraints.

Having done research on “learn to code” (Buchanan 2022), I always watch new developments with interest. In 2023, I told an auditorium full of students in Indiana to learn to code if they don’t hate the work too much. At that time I had forecast that AI tools would make coding less miserable but not eliminate the need for technical human workers. Even if that was good advice at the time, is it still good advice today? I wish I had time to put up a blog on this topic every week.

Adjustments can happen along the margin of price as well as quantity. Wages to programmers can come down from their previously exalted heights, which could help the market absorb some of the young professionals who listened to “learn to code” in 2023.

So, now that the value of coding skills is in question, people are turning back to the value of the maligned English degree. It has been true for a long time that employers felt soft skills were more scarce than STEM degrees. I might add that an economics degree conveys a highly marketable blend of hard and soft skills.

Buchanan, Joy (2022). “Willingness to be paid: Who trains for tech jobs?” Labour Economics,
79, Article 102267.

Learning the Bitter Lesson at EconLog

I’m in EconLog with:

Learning the Bitter Lesson in 2026

At the link, I speculate on doom, hardware, human jobs, the jagged edge (via a Joshua Gans working paper), and the Manhattan Project. The fun thing about being 6 years late to a seminal paper is that you can consider how its predictions are doing.

Sutton draws from decades of AI history to argue that researchers have learned a “bitter” truth. Researchers repeatedly assume that computers will make the next advance in intelligence by relying on specialized human expertise. Recent history shows that methods that scale with computation outperform those reliant on human expertise. For example, in computer chess, brute-force search on specialized hardware triumphed over knowledge-based approaches. Sutton warns that researchers resist learning this lesson because building in knowledge feels satisfying, but true breakthroughs come from computation’s relentless scaling. 

The article has been up for a week and some intelligent comments have already come in. Folks are pointing out that I might be underrating the models’ ability to improve themselves going forward.

Second, with the frontier AI labs driving toward automating AI research the direct human involvement in developing such algorithms/architectures may be much less than it seems that you’re positing.

If that commenter is correct, there will be less need for humans than I said.

Also, Jim Caton over on LinkedIn (James, are we all there now?) pointed out that more efficient models might not need more hardware. If the AIs figure out ways to make themselves more efficient, then is “scaling” even going to be the right word anymore for improvement? The fun thing about writing about AI is that you will probably be wrong within weeks.

Between the time I proposed this to Econlog and publication, Ilya Sutskever suggested on Dwarkesh that “We’re moving from the age of scaling to the age of research“.

Telephone Classroom Game for Teaching Large Language Models

Use the above game to generate interaction in a class setting. Students collectively form an LLM and have fun seeing the final sentence that gets produced. I call this game “LLM Telephone” based on the classic game of telephone. I suggest downloading the file LLM_Telephone_Game_Sheet and handing out printed copies. However, this game could be adapted to a virtual setting.

The nice thing about passing papers in the classroom is that you can have several sheets circulating in a quite room, so when the final sentence is read allowed it comes as a surprise to most people.

If you’d like to have a handout to follow the game with a more technical explanation, you can use this two-page PDF:

The game relies on a player presenting two tokens of which the next player can select their favorite. Participants should be bound by the rules of grammar and logic when making their selection and presenting two tokens to the next player.

This game works as a fun ice breaker for any type of class that touches on the topic of artificial intelligence. It is suitable for many ages and academic disciplines.

IP Paper on Econlog

My research on intellectual property is featured at

Everyone Take Copies (Econlog)

The title of this post, “everyone take copies,” comes from a conversation between the human subjects in an experiment in our lab, on which the paper is based. The experiment was studying how and when people take resources from one another.

Here’s a tip that doesn’t require any piracy. For those of you who are tired of the subscription economy fees, I think it’s safe to say in 2026 that anyone in the United States can find a local thrift store or annual rummage sale with oodles of nearly-free media. DVDs for a dollar. Used books for a dollar. Basically you are paying the transaction costs – the media itself is free. (I typed that dash myself, not AI!)

“Buying” a movie to stream on Amazon Prime can run over $20. Buying a used DVD is usually less than $10.

Something like the above observation probably lead to this parody news headline Awesome New Streaming Service Records Movie Streams Onto Cool Shiny Discs And You Can Buy Them And Own Them Forever

Here’s a response from the prompt “Make a picture of my office with AOL CD-ROMs decorating the wall.”

Update on School Valentines

I have principles. One of them is that school Valentines are indulgent and bad for the environment.

I have written Markets in Everything for school Valentine’s

And here are some quotes from Do Less for Preschool

Just fail, people. Don’t even put “crazy sock day” on your work Outlook calendar…

Oh no. If AI lifts the constraint of time, then what we are going to get is more crazy sock days. To stay ahead in the status competition, families will have to do Bluey-Crazy Sock Day every week. The ocean will become a thick soup of polyester Bluey-crazy-socks, size 3T, worn only once.

On principle, I did low-effort Valentine’s last year. I spent as little of my own time and money as I could. My kids wrote their friends’ names on the paper things.  Smugly, I imagined that I’d saved the dolphins in the tuna nets and helped some other mom feel like she was doing okay.

Who do you imagine is upset? Not the other moms. My kids. The older one especially feels in his body that failing this test in February of 2026 will result in expulsion from the tribe and death in the outer darkness beyond the reach of the campfire.

Guess what I care about more than dolphins? We will do more this year.

Preschool kids do not care and should not be asked to care. I still believe that parents and daycare directors should do less for preschool. Truly, I see no reason, at all, for a preschool dress up day or Valentine’s Day party. Can someone think of the dolphins before it’s too late and the kid grows up and starts caring about the status wars?

Casual Sourdough Baking

I have sourdough starter in my fridge and have tried making two rounds of bread. I write this as a civilian, for others who will never elevate baking above a hobby.

I will begin with a picture of my first bread:

That’s not very fluffy but not terrible. It was real bread to go with butter or soup. My neighbor’s starter was in great condition when she gave it to me, and I followed the standard recipe. I use 80% bread flour and 20% whole wheat flour to make it healthier.

Below I will speculate on whether sourdough projects are a good use of time or not. Call it home economics.

Continue reading

Thousands killed in Iran Protests

How Many People Were Killed in Iran’s Crackdown? (WSJ)

Efforts to establish the death toll in the Iranian protests are confounded by the regime’s internet blockade, but even the most conservative estimates take the tally way beyond some of the most brutal political crackdowns in modern history.

Even the lowest estimates—between 2,000 and 3,000—have surged past the death tolls in unrest during protests in 2019 and 2022.

I have sadly seen much higher estimates circulating which may be confirmed eventually.

Further complicated by the fact that Iran authorities demanding large sums for return of protesters’ bodies, BBC told

The value of reading great literature

I am pleased to have guest posted for Henry Oliver:

An economist asks: What is the value of reading great literature like Eliot and Tolstoy?

In 2025, after mostly feeling too busy for great literature for a few years, I picked up two books that come highly recommended by people with good taste: Middlemarch and Anna Karenina. They are excellent, and they are long. I propose two reasons why they need to be long.

My second reason for long novels:

The second powerful thing about a long novel, if they are written by geniuses like Eliot or Tolstoy, is that you have enough time to see how choices play out over years. You have space to even see the consequences of the consequences. You will experience moral formation from these novels in a way that you just cannot from a 2-hour movie or social media post. 

Quick bio of Henry: Henry Oliver writes the popular literary Substack The Common Reader, which has been quoted in the Atlantic and elsewhere. His book Second Act, a study of late-blooming talent, was released in 2024.

Lastly, I must thank my sister for engaging me in literature discussion over the Christmas break. She is reading Proust.

Summary of You Wouldn’t Steal a Car

I have a new working paper with Bart Wilson titled: “You Wouldn’t Steal a Car: Moral Intuition for Intellectual Property” 

This quote from the introduction explains the title:

… in the early 2000s… the Motion Picture Association of America (MPAA) released an anti-piracy trailer shown before films that argued: (1) “You wouldn’t steal a car,” (2) pirating movies constitutes “stealing,” and (3) piracy is a crime. The very need for this campaign, and the ridicule it attracted, signals persistent disagreement over whether digital copying constitutes a moral violation.

The main idea:

In contemporary economies, “idea goods” comprise a substantial share of value. Our paper examines how norms evolve when individuals evaluate harm after the taking of nonrivalous resources such as digital files.

We report experimental evidence on moral evaluations of unauthorized appropriation, contingent on whether the good is rivalrous or nonrivalrous. In a virtual environment, participants produce and exchange two types of resources: nonrivalrous “discs,” replicable at zero marginal cost, and rivalrous “seeds,” which entail positive marginal cost and cannot be simultaneously possessed or consumed by multiple individuals. Certain treatments allow unauthorized taking, which permits observation of whether participants classify such actions as moral transgressions.

Participants consistently label the taking of rivalrous goods as “stealing,” whereas they do not apply the same term to the taking of nonrivalrous goods.

To test the moral intuition for taking ideas, we create an environment where people can take from each other and we study their freeform chat. The people in the game each control a round avatar in a virtual environment, as you can see in this screenshot below.

In the experiment, “seeds” represent a rivalrous resource, meaning they operate like most physical goods. If the playerin the picture (Almond) takes a seed from the Blue player, then Blue will be deprived of the seed, functionally the equivalent of one’s car being stolen.

Thus, it is natural for the players to call the taking of seeds “stealing,” Our research question is whether similar claims will emerge after the taking of non-rivalrous goods that we call “discs.”

The following quote from our paper indicates that the subjects do not label or conceptualize the taking of digital goods (discs) as “stealing.”

Participants discuss discs often enough to reveal how they conceptualize the resource. In many instances, they articulate the positive-sum logic of zero-marginal-cost copying. For example, … farmer Almond reasons, “ok so disks cant be stolen so everyone take copies,” explicitly rejecting the application of “stolen” to discs.

Participants never instruct one another to stop taking disc copies, yet they frequently urge others to stop taking seeds. The objection targets the taking away of rivalrous goods, not the act of copying per se. As farmer Almond explains in noSeedPR2, “cuz if u give a disc u still keep it,” emphasizing that artists can replicate discs at zero marginal cost.

We encourage you to read the manuscript if you are interested in the details of how we set up the environment. The conclusion is that it is not intuitive for people to view piracy as a crime.

This has implications for how the modern information economy will be structured. Consider “the subscription economy.” Increasingly, consumers pay recurring fees for ongoing access to products/services (like Netflix, Adobe software) instead of one-time purchases. Gen Z has been complaining on TikTok that they feel trapped with so many recurring payments and lack a sense of ownership.

In a recent interview on a talk show called The Stream, I speculated that part of the reason companies are moving to the subscription model is that they do not trust consumers with “ownership” of digital goods. People will share copies of songs and software, if given the opportunity, to the point where creators cannot monetize their work by selling the full rights to digital goods anymore.

A feature of our experimental design is that creators of discs get credit as the author of their creation even when it is being passed around without their explicit permission. Future work could explore what would happen if that were altered.

Related Reading

An Experiment on Protecting Intellectual Property” (2014) with Bart Wilson. Experimental Economics, 17:4, 691-716.

You Wouldn’t Steal a Car: Moral Intuition for Intellectual Property,” with Bart Wilson (SSRN)

Joy on The Subscription Economy (EWED)

The Anthropic Settlement: A $1.5 Billion Precedent for AI and Copyright (EconLog)

Understanding Vulnerability: What Anna Karenina Can Teach Us About Grooming and Loneliness

Ever since becoming aware of the terrible news about “grooming gangs” in the UK, I’ve been wanting to write something about why men succeed in manipulating women. Having a moment to read fiction during my break from classes, I have picked up Tolstoy’s Anna Karenina. Now that I’m past the turning point in the book where Anna and her lover Vronsky have to think about a new life outside the care of Anna’s husband, it’s clear that Anna fell for a person who will struggle to take care of her and any children. How does it work?   

Most women are unprepared concerning how desperate they are for what I’ll call love. Many women receive very little attention. More than 40% of adult women in the US are single. We have statistics on marital status, but profound loneliness can also occur within an official relationship.

Articles about the UK grooming gangs often emphasize the disadvantaged economic backgrounds of the victims. That does matter, and it did make them more vulnerable to manipulation. Vulnerability within most people everywhere is underexplored.

Anna Karenina is married, privileged, admired in society, yet she feels lonely. When Vronsky shows her focused attention she falls hard, even though she knows there could be consequences. Tolstoy shows how powerful validation can be to almost any woman, not just those who might seem the most vulnerable.

Vronsky is charming and attractive on the surface, but ultimately self-centered. He ruins Anna’s life and deprives her children of a mother. Why did he succeed in the first place?

If Anna is beautiful, some would assume that she would not be lonely. There are theories going around about the advantages of being beautiful (lookism). Even Jennifer Garner and Jennifer Aniston get cheated on. Most women are not experiencing something that feels like love to them.

In the case of the grooming gangs, folks with an understanding of emotional deprivation hacked the system for evil. None of this diminishes the responsibility of perpetrators or the reality of coercion. Because the initial phase feels so validating, grooming victims often blame themselves later. We might do better by bringing the system out into the open.

Systems can be used for good by those who understands them. Especially young people with a better understanding of the system can have a better chance of making it work in their favor.

It’s a weird conversation to have with youths (easier to assign Anna Karenina in high schools, but kids are losing the ability to read a novel). Could we educate them with something like: “You have a desire to be loved that may never get fulfilled. That does not make you special. It’s the most unoriginal thing about you. Try to make the system work for you and not get tricked.”

The red flag for Vronsky should have been his lack of family and lack of care for his community (he does not pay his tailor). The classic advice for young women to observe how a prospective boyfriend treats his mother is still very good.

Tolstoy does not give specific advice about what people should do. A superficial reading of the story would be that Anna Karenina is about duty and sin and the wages of sin. But throughout, it is a meditation on happiness. The second word of the novel is “happy,” as in “All happy families…”

In the first line of the novel, Tolstoy situates happiness within a family, not as something experienced by individuals. The characters, Levin and Kitty, who seem happiest at the end, find each other and work toward something greater than themselves.

Continue reading