TimeGPT-1

In this paper, we introduce TimeGPT, the first foundation model for time series, capable of generating accurate predictions for diverse datasets not seen during training. We evaluate our pre-trained model against established statistical, machine learning, and deep learning methods, demonstrating that TimeGPT zero-shot inference excels in performance, efficiency, and simplicity. Our study provides compelling evidence that insights from other domains of artificial intelligence can be effectively applied to time series analysis. We conclude that large-scale time series models offer an exciting opportunity to democratize access to precise predictions and reduce uncertainty by leveraging the capabilities of contemporary advancements in deep learning.

That is from a new paper by Azul Garza and Max Mergenthaler-Canseco.  A few of you may be needing a new job soon!

Robert Whaples reviews *GOAT*

An excellent piece, here is one excerpt I enjoyed in particular:

Cowen reads the John Maynard Keynes of The General Theory “as writing about an economy where uncertainty was much higher than usual, investment was highly unstable, fiscal policy was unable to fill in the gap, there was a risk or even reality of a downward spiral of prices and wages, monetary and exchange rate policies were out of whack, multipliers operate, the quest for savings could lower incomes overall, and the influence of liquidity factors on money demand and interest rates was especially high. All at once” (p. 72, emphasis in the original). In other words, Cowen drives home the point that this “general theory” isn’t actually general, it’s about very special, very unusual circumstances.

He considers Lord Keynes the GOAT contender whom he would most “want to hang around with” (p. 54). I had exactly the opposite reaction. The Keynes he portrays is virtually an egotistical monster. One who, for example, “kept an extended spreadsheet of his lovers and sexual encounters … each one rated by number” (p. 58). Anyone who treats other human beings this way—let alone writing it down—isn’t the kind of person I want to hang around with.

Recommended.

Wednesday assorted links

1. Terminator metal.

2. An RCT on the economic benefits of vision correction.

3. “Claude 3 Opus is roughly as persuasive as humans.

4. Jon Haidt responds to the Nature review.  And Greg Lukianoff with his First Amendment concerns.

5. Interview with Nicholas Tabarrok, who works in the movies.

6. Samo Burja skeptical on nuclear.

7. Henry Oliver on smart phones.

8. Noah worries about WWIII.

The $20 bill gets picked up, body parts markets in everything

At the 1815 Battle of Waterloo, Napoleon Bonaparte’s final battle, more than 10,000 men and as many horses were killed in a single day. Yet today, archaeologists often struggle to find physical evidence of the dead from that bloody time period. Plowing and construction are usually the culprits behind missing historical remains, but they can’t explain the loss here. How did so many bones up and vanish?

In a new book, an international team of historians and archaeologists argues the bones were depleted by industrial-scale grave robbing. The introduction of phosphates for fertilizer and bone char as an ingredient in beet sugar processing at the beginning of the 19th century transformed bones into a hot commodity. Skyrocketing prices prompted raids on mass graves across Europe—and beyond.

Here is the full article, via William Meller.  And, as Alex has stressed in the past, never underestimate the elasticity of supply!

Why is there a movement to ban lab-grown beef?

That is the topic of my latest Bloomberg column, here is one bit:

…let me offer another theory: The anti-lab-grown-meat movement is about conservative cultural insecurity — the fear that, without the force of law, some conservative cultural norms will fade away…

Imagine that lab-grown meat proves feasible at a reasonable cost. It might end up as cheaper than beef from a cow, and it might also be better for the climate. In such a world, there might be growing pressures to abandon real meat for the lab-grown kind. There could even be a political movement to tax or ban real meat, similar to carbon taxes or plans to phase out fossil fuels.

Currently there is no momentum in that direction. For all the talk of vegetarianism and veganism, the percentage of Americans who practice those beliefs seems to be roughly flat. Many Americans like eating meat, for better or worse. But if real meat had a true substitute, perhaps the political calculus would differ.

This is the real fear — not of lab-grown meat itself, but of the changing culture its popularity would represent. Whether conservatives find the meat substitute to be adequate is beside the point. Society would have decided that some of their most cherished beliefs can be disposed of. Both humankind’s dominion over nature, which runs strong in the Christian strand of conservative thought, and the masculinized meat-eating culture — more specifically, the meat-grilling culture — would be under threat.

If artificial meat is banned, of course, none of that can happen.

In one sense, critics of conservatism should be heartened by the campaign against lab-grown meat. If I were a mainstream animal-rights advocate, I would revise upwards my estimate of my own power and influence.

I then consider how we might use science to arrive at a better resolution of these disputes.

Modi and investment

Yes, India is doing well but the picture could be much better:

During Modi’s tenure, GFCF [gross fixed capital formation] as a percentage of GDP declined and has remained low until the post-pandemic recovery. In fact the highest level of GFCF as a percentage of GDP during the first nine years of Modi’s leadership is lower than the lowest level in PM Singh’s tenure.

Here is more from Shruti.

A graph with a line and a red circle

Description automatically generated

Tuesday assorted links

1. Skepticism about the beauty premium?

2. “On a cellular level, younger generations seem to be aging faster than their forebears.” (speculative)

3. Interview with Merve Emre.

4. William Stanley Jevons and eclipses (NYT).  And an Amtrak train ride across the country is less carbon-efficient than flying (NYT).

5. The ascent of high school wrestlers.

6. A short video on how the Great Pyramids may have been built.

7. Was more spent on eclipse tourism than on the Taylor Swift tour?

What to Watch

3 Body Problem (Netflix): Great! A captivating mix of big ideas, a compelling mystery, and spectacular set-pieces like the Cultural Revolution, strange worlds, the ship cutting and more. Of course, there are some weaknesses. 3 Body Problem falters in its portrayal of genius, rendering the British scientists as too normal, overlooking the obsessiveness, ambition, and unconventionality often found in real-world geniuses. Ironically, in its effort to diversify gender and race, the series inadvertently narrows the spectrum of personality and neurodiversity. Only Ye Wenjie, traumatized by the cultural revolution, obsessed by physics and revenge, and with a messianic personality hits the right notes. Regardless, I am eager for Season 2.

Shogun (Hulu): Great meeting of cultures. Compelling plot, based on the excellent Clavell novel. I didn’t know that some of the warlords of the time (1600) had converted to Christianity. (Later banned and repressed as in Silence). Shogun avoids two traps, the Japanese have agency and so does the European. Much of it is in Japanese with subtitles.

Monsieur Spade: It starts with a great premise, twenty years after the events of “The Maltese Falcon,” Sam Spade has retired in a small town in southern France still riven by World War II and Algeria. Clive Owen is excellent as Spade and there are some good noir lines:

Henri Thibaut: You were in the army, Mr. Spade?

Sam Spade: No, I was a conscientious objector.

Henri Thibaut: You don’t believe in killing your fellow man?

Sam Spade: Oh, I think there’s plenty of men worth killing, as well as plenty of wars worth fighting, I’d just rather choose myself.

Yet for all the promise, I didn’t finish the series. In addition to being set in France, Monsieur Spade has a French cinema atmosphere, boring, long, vaguely pretentious. There is also a weird fascination with smoking, does it pay off with anything? I don’t know. Didn’t finish it.

What I’ve been reading

1. Roger Lewis, Erotic Vagrancy: Everything about Richard Burton and Elizabeth Taylor.  An amazing book, full of life and energy on every page, and yes there are 605 of them.  Imagine if Camille Paglia had stuck with it and produced case studies.  The main problem is simply that most people don’t know or care about Burton and Taylor any more?

2. David Caron, Michael Healy, 1873-1941, An Túr Gloine’s Stained Glass PioneerAn excellent book, can it be said that Michael Healy is Ireland’s fourth greatest stained glass artist?  Clarke, Geddes, and Hone would be the top three?  It is good to see him getting this attention, but what will happen when so many Irish churches are decommissioned or abandoned or simply never seen?  What does that equilibrium look like?  All the more reason to invest in this book.  What an underrated European tradition.

3. Paul Seabright, the subtitle says it all, The Divine Economy: How Religions Compete for Wealth, Power, and People.  I’ve just started to crack this one open, Paul’s books are always very smart.

4. Sahar Akhtar, Immigration & Discrimination: (un)welcoming others.  Can the idea of wrongful discrimination be applied to immigration decisions?  Maybe you believe this is a pure and simple matter of national autonomy, but what if the potential immigrants are from a former and wronged colony?  From an island nation perishing due to climate change?  Or they were previously pushed off territory that is now part of the host nation?  And yet open borders as an idea also does not work — how should one fit all these pieces together?

5. Austin Bush, The Food of Southern Thailand.  The best book I know of on southern Thailand flat out.  This one has recipes of course, but also photos, maps, anecdotes, and plenty of history.  The food is explained in conceptual terms.  Recommended, for all those with an interest.

6. Michael Cook, A History of the Muslim World: From its Origins to the Dawn of Modernity.  Mostly ends at 1800, this will become one of the standard, must-read histories of Islam and its multiple homes.  The section on India, which is what I have been reading, is strongly conceptual and novel compared to other survey books such as Hourani.  At the very least a good book, possibly a great book.

Monday assorted links

1. NYT symposium on smart phones and childhood.  And when do young female suicide rates start rising?  Tweet here.  And “Gen Alpha (my kid’s generation) has already optimized out of it and have figured out how to do the social play they need to in the new medium.

2. Prize money for prompts.

3. Bridgewater x Metaculus forecasting contest.

4. Louise Perry on Andrea Dworkin.

5. Chechenya bans all music deemed too fast or too slow.

6. Is Pakistan seeking normalization with India?

7. Eclipse songs.  And from India.

8. Levitt and Donohue defend their abortion-crime results.

Cultivating Minds: The Psychological Consequences of Rice versus Wheat Farming

It’s long been argued that the means of production influence social, cultural and psychological processes. Rice farming, for example, requires complex irrigation systems under communal management and intense, coordinated labor. Thus, it has been argued that successful rice farming communities tend to develop people with collectivist orientations, and cultural ways of thinking that emphasize group harmony and interdependence. In contrast, wheat farming, which requires less labor and coordination is associated with more individualistic cultures that value independence and personal autonomy. Implicit in Turner’s Frontier hypothesis, for example, is the idea that not only could a young man say ‘take this job and shove it’ and go west but once there they could establish a small, viable wheat farm (or other dry crop).

There is plenty of evidence for these theories. Rice cultures around the world do tend to exhibit similar cultural characteristics, including less focus on self, more relational or holistic thinking and greater in-group favoritism than wheat cultures. Similar differences exist between the rice and dry crop areas of China. The differences exist but is the explanation rice and wheat farming or are there are other genetic, historical or random factors at play?

A new paper by Talhelm and Dong in Nature Communications uses the craziness of China’s Cultural Revolution to provide causal evidence in favor of the rice and wheat farming theory of culture. After World War II ended, the communist government in China turned soldiers into farmers arbitrarily assigning them to newly created farms around the country–including two farms in Northern Ningxia province that were nearly identical in temperature, rainfall and acreage but one of the firms lay slightly above the river and one slightly below the river making the latter more suitable for rice farming and the former for wheat. During the Cultural Revolution, youth were shipped off to the farms “with very little preparation or forethought”. Thus, the two farms ended up in similar environments with similar people but different modes of production.

Talhelm and Dong measure thought style with a variety of simple experiments which have been shown in earlier work to be associated with collectivist and individualist thinking. When asked to draw circles representing themselves and friends or family, for example, people tend to self-inflate their own circle but they self-inflate more in individualist cultures.

The authors find that consistent with the differences across East and West and across rice and wheat areas in China, the people on the rice farm in Ningxia are more collectivistic in their thinking than the people on the wheat farm.

The differences are all in the same direction but somewhat moderated suggesting that the effects can be created quite quickly (a few generations) but become stronger the longer and more embedded they are in the wider culture.

I am reminded of an another great paper, this one by Leibbrandt, Gneezy, and List (LGL) that I wrote about in Learning to Compete and Cooperate. LGL look at two types of fishing villages in Brazil. The villages are close to one another but some of them are on the lake and some of them are on the sea coast. Lake fishing is individualistic but sea fishing requires a collective effort. LGL find that the lake fishermen are much more willing to engage in competition–perhaps having seen that individual effort pays off–than the sea fishermen for whom individual effort is much less efficacious. Unlike Talhelm and Dong, LGL don’t have random assignment, although I see no reason why the lake and sea fishermen should otherwise be different, but they do find that women, who neither lake nor sea fish, do not show the same differences. Thus, the differences seem to be tied quite closely to production learning rather than to broader culture.

How long does it take to imprint these styles of thinking? How long does it last? Is imprinting during child or young adulthood more effective than later imprinting? Can one find the same sorts of differences between athletes of different sports–e.g. rowing versus running? It’s telling, for example, that the only famous rowers I can think are the Winklevoss twins. Are attempts to inculcate these types of thinking successful on a more than surface level. I have difficulty believing that “you didn’t build that,” changes say relational versus holistic thinking but would styles of thinking change during a war?

350+ coauthors study reproducibility in economics

Jon Hartley is one I know, here is the abstract:

This study pushes our understanding of research reliability by reproducing and replicating claims from 110 papers in leading economic and political science journals. The analysis involves computational reproducibility checks and robustness assessments. It reveals several patterns. First, we uncover a high rate of fully computationally reproducible results (over 85%). Second, excluding minor issues like missing packages or broken pathways, we uncover coding errors for about 25% of studies, with some studies containing multiple errors. Third, we test the robustness of the results to 5,511 re-analyses. We find a robustness reproducibility of about 70%. Robustness reproducibility rates are relatively higher for re-analyses that introduce new data and lower for re-analyses that change the sample or the definition of the dependent variable. Fourth, 52% of re-analysis effect size estimates are smaller than the original published estimates and the average statistical significance of a re-analysis is 77% of the original. Lastly, we rely on six teams of researchers working independently to answer eight additional research questions on the determinants of robustness reproducibility. Most teams find a negative relationship between replicators’ experience and reproducibility, while finding no relationship between reproducibility and the provision of intermediate or even raw data combined with the necessary cleaning codes.

Here is the full paper, here are some Twitter images.  I have added the emphasis on the last sentence.