Results for “series”
1036 found

Grandmaster-Level Chess Without Search

Scaling seems to be working here:

The recent breakthrough successes in machine learning are mainly attributed to scale: namely large-scale attention-based architectures and datasets of unprecedented scale. This paper investigates the impact of training at scale for chess. Unlike traditional chess engines that rely on complex heuristics, explicit search, or a combination of both, we train a 270M parameter transformer model with supervised learning on a dataset of 10 million chess games. We annotate each board in the dataset with action-values provided by the powerful Stockfish 16 engine, leading to roughly 15 billion data points. Our largest model reaches a Lichess blitz Elo of 2895 against humans, and successfully solves a series of challenging chess puzzles, without any domain-specific tweaks or explicit search algorithms. We also show that our model outperforms AlphaZero’s policy and value networks (without MCTS) and GPT-3.5-turbo-instruct. A systematic investigation of model and dataset size shows that strong chess performance only arises at sufficient scale. To validate our results, we perform an extensive series of ablations of design choices and hyperparameters.

Here is the full paper by Anian Ruoss, et.al. I guess AI is likely to get better at other things too  — I’ve become numb to the miracles, frankly.

One reason why the disinflation proved so manageable

From a new and important paper by Xiwen Bai, Jesús Fernández-Villaverde, Yiliang Li, and Francesco Zanetti:

Our analysis shows that supply chain disruptions generate stagflation, accompanied by an increase in spare capacity for producers. This higher spare capacity curtails the supply of goods to retailers and results in a surge in prices, leading to a tighter retail market. We show that, in this situation, prices become highly sensitive to changes in demand, while output remains relatively inelastic. In other words, disruptions to the supply chain enhance the effectiveness of contractionary monetary policy in taming inflation while reducing the sensitivity of output to the policy. Our results reinforce the general findings on the state-dependence of the efficacy of monetary policy.

Progress!  And arguably this is a Keynesian result:

In fact, our result resembles the celebrated analysis by Keynes (1940). Keynes argued that when output is constrained (in our case, because of supply chain disruptions, in Britain’s case in 1940, because of resources employed in World War II), policymakers can lower aggregate demand aggressively to prevent inflation without much fear of lowering production.

As I mention in GOAT, How to Pay for the War remains a neglected and underrated work by Keynes.  Note this as well:

We document how supply chain shocks drove inflation during 2021 but that, in 2022, traditional demand and supply shocks also played an important role in explaining inflation.

The pure supply-side story, as you have been hearing from Krugman just isn’t going to work, it is time to give it up.

By the way, yet another advance in this lengthy paper is this (from JFV): “…we have satellite information about every single container ship of the world in real time (ID, geolocation, speed, draft, heading,…), which allows us to do tons of things in terms of machine learning and time series econometrics that people have not been able to do before.”

Vitalik on AI and crypto

…we can expect to see in the 2020s that we did not see in the 2010s: the possibility of ubiquitous participation by AIs.

AIs are willing to work for less than $1 per hour, and have the knowledge of an encyclopedia – and if that’s not enough, they can even be integrated with real-time web search capability. If you make a market, and put up a liquidity subsidy of $50, humans will not care enough to bid, but thousands of AIs will easily swarm all over the question and make the best guess they can. The incentive to do a good job on any one question may be tiny, but the incentive to make an AI that makes good predictions in general may be in the millions. Note that potentially, you don’t even need the humans to adjudicate most questions: you can use a multi-round dispute system similar to Augur or Kleros, where AIs would also be the ones participating in earlier rounds. Humans would only need to respond in those few cases where a series of escalations have taken place and large amounts of money have been committed by both sides.

This is a powerful primitive, because once a “prediction market” can be made to work on such a microscopic scale, you can reuse the “prediction market” primitive for many other kinds of questions:

  • Is this social media post acceptable under [terms of use]?
  • What will happen to the price of stock X (eg. see Numerai)
  • Is this account that is currently messaging me actually Elon Musk?
  • Is this work submission on an online task marketplace acceptable?
  • Is the dapp at https://examplefinance.network a scam?
  • Is 0x1b54....98c3 actually the address of the “Casinu Inu” ERC20 token?

You may notice that a lot of these ideas go in the direction of what I called “info defense” in .

There are many other points, such as AIs as oracles, read it here, interesting throughout.

2023 CWT retrospective episode

Here is the link, here is the episode summary:

On this special year-in-review episode, Tyler and producer Jeff Holmes look back on the past year in the show and more, including the most popular and underrated episodes, the origins of the show as an occasional event series, the most difficult guests to prep for, the story behind EconGOAT.AI, Tyler’s favorite podcast appearance of the year, and his evolving LLM-powered production function. They also answer listener questions and conclude with an assessment of Tyler’s top pop culture recommendations from 2013 across movies, music, and books.

And one excerpt:

COWEN: That’s a unique experience. You have a chance to do Chomsky. Maybe you don’t even want to do it, but you feel, “If I don’t do it, I’ll regret not having done it.” Just like we didn’t get to chat with Charlie Munger in time, though he’s far more, I would say, closer to truth than Chomsky is.

I thought half of Chomsky was quite good, and the other half was beyond terrible, but that’s okay. People, I think, wanted to gawk at it in some manner. They had this picture — what’s it like, Tyler talking with Chomsky? Then they get to see it and maybe recoil, but that’s what they came for, like a horror movie.

HOLMES: The engagement on the Chomsky episode was very good. Some people on MR were saying, “I turned it off. I couldn’t listen to it.” But actually, most people listened to it. It did, actually, probably better than average in terms of engagement, in terms of how much of the episode, on average, people listen to.

COWEN: How can you turn it off? What does that say about you? Were you surprised? You thought that Chomsky had become George Stigler or something? No.

Fun and interesting throughout.  If you are wondering, the most popular episode of the year, by far, was with Paul Graham.

Why Housing is Unaffordable: The Elasticity of Supply

We have a great new MRU video on housing and why it’s so expensive in many dynamic cities. The video is a good introduction to the economics and also to the politics of housing supply. A highlight is a wonderful animation illustrating how increasing demand coupled with inelastic supply leads to competitive bidding among buyers, driving up prices. Anyone teaching economics should take a look.

Of course, the video pairs perfectly with our textbook, Modern Principles. Indeed, this is the initial video in a two-part series exploring the elasticity of supply, transforming a traditionally mundane topic into a relevant and fun discussion. Check it out!

Das Adam Smith Problem

The second set of advocates for the book [Theory of Moral Sentiments] I usually find in media outlets, sophisticated media outlets at that, or I hear it over lunch table conversation. These claims suggest that Wealth of Nations covers the commercial, selfish side of human behavior, while Theory of Moral Sentiments is an account of the caring, empathetic side, or something like that. I wish I had a nickel for every time I read or heard that contrast. Maybe it is harmless enough, but – and I don’t completely understand why — it kind of makes me sick. It is simultaneously an attempt to claim a bland centrist middle ground, to snidely distance oneself from capitalism and selfishness, and reduce Smith to a series of empty clichés. It is trying to be pat rather than insightful. It is trying to give everything its place in a manner that we can then safely ignore.

Just for a start, I view Smith’s portrait of human nature in Wealth of Nations as rich and multi-faceted, a piece of behavioral economics, in modern terminology, rather than narrow, commercial, and purely selfish. And in Theory of Moral Sentiments yes people are empathetic, and show sympathy for others, but they are often caring in…pretty narrow and selfish ways. I just don’t think the “each book carves out its own sphere” understanding of the pair works very well.

My biggest takeaway from TMS is that humans beings make evaluations, including sympathetic evaluations but not only, based on local rather than global information. They put a lot of weight on what is right before their eyes and neglect the bigger picture. The very opening passage of TMS expresses how we can understand the emotions of others only through our own. We cannot look around corners to understand other minds directly, so we make inferences from our own experience. Smith demonstrates and then demonstrates that point again throughout the book.

That is a passage from my generative book, written by me, GOAT: Who is the Greatest Economist of all Time, and Why Does It Matter?

Saturday assorted links

1. Henry Oliver and GOAT and personality.

2. Various ratings for LLM services.

3. Georgetown fact of the day, 1940 edition.  And Unilever retreats from “Woke” (FT).

4. “Preschoolers extend novel labels based on people’s weight rather than their race.

5. I guess they all read Thomas Schelling?

6. Correct link for new Stephen Dubner podcast series.  Theme is how to succeed by failing.

7. Cry for me Argentina.

Friday assorted links

1. The culture that was German.

2. “The profit share has been constant at 18% of GDP because the increase in monopoly rents has been completely offset by rising fixed costs and changes in technology.

3. More on GPT for time series.

4. My beloved Acapulco is in bad shape.

5. The Latecomer, new on-line periodical, lead article by Casey Handmer.  And on Twitter.

6. Geothermal learning curves.

Thursday assorted links

1. MIE: Robinson Crusoe tourism.

2. “An estimated 500,000~ women in the United States between 18-24 are content creators on Onlyfans.

3. New Freakonomics radio podcast series on how to succeed at failing.

4. “Celebrities made this mainstream.” Recommended.

5, Want to send a comment to the IRS about proposed new crypto tax reporting rules? The polity that is AI.

6. The world’s first wearable e-reader?

7. Robert Irwin, RIP (NYT).

8. Brian Albrecht on why Armen Alchian is the GOAT.  And Peter Isztin on Gary Becker.

Wednesday assorted links

1. Is it racist to have this policy, or racist to cancel it?

2. Mark Skousen argues for Carl Menger as GOAT.

3. Benjamin Yeoh reviews GOAT.  Excellent piece.

4. Are LLMs too sycophantic? 

5. “Findings from a series of logistic regression models indicate that Black Protestants have the lowest rates of both credit card and mortgage debt and Hispanic/Latinx Catholics have comparably low rates of credit card debt relative to Conservative Protestants.

6. Richard Roundtree, RIP.

7. And “It is the height of the cyborg Theocracy.”  NB: I don’t actually recommend clicking on the link.

8. Max Corden has passed away (FT).

What is it we do and do not know about macroeconomics?

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

Another episode frequently cited as evidence against economists is the Great Recession of 2007-2009. Economists did make some mistakes on that one — but they are not the ones you usually hear about.

When real estate prices started to slow down and then fall, many economists declared there was a real estate bubble. The theory quickly developed that the market crash was due to a real estate bubble bursting, followed by a sharp fall in aggregate demand, followed by a decline in employment and output.

The last part of that explanation is correct. In retrospect, however, it is not clear that the housing prices of 2006-2007 represented a bubble. By today’s metrics those prices appear prescient, if slightly premature. The market was suddenly realizing that a lot of real estate assets were going to be worth much more — and the recent evolution of real estate valuations seems to have confirmed that judgment.

In 2009, however — and following a lot of foreclosures and the emergence of troubled banks — the market was far from ready to accept that the high real estate prices had been justified. The market was too skeptical when it should have been less panicked. A lot of economists got this wrong too, along with many pundits. All of this made the resulting panic worse because the talk was so pessimistic about real estate valuations. Instead, the real problem was that the market had lost faith in a set of high real estate prices that has since been largely validated. Maybe not in Las Vegas and Orlando, but for the nation as a whole, most of all on the coasts.

Economists should have been less quick to judge what is or is not a bubble. The real-estate-bubble explanation appeared to be correct in the short run, but economists should have been more modest about their ability to second-guess the market. The good news is that, with hindsight, we can piece together what happened. Policymakers and market participants made a series of overlapping mistakes related to monetary policy, the shadow banking system and panic about real estate.

There is a good picture of trends in real estate prices at the link, or ungated here.

My Conversation with Ada Palmer

Here is the audio, video, and transcript.  Here is the episode summary:

Ada Palmer is a Renaissance historian at the University of Chicago who studies radical free thought and censorship, composes music, consults on anime and manga, and is the author of the acclaimed Terra Ignota sci-fi series, among many other things.

Tyler sat down with Ada to discuss why living in the Renaissance was worse than living during the Middle Ages, how art protected Florence, why she’s reluctant to travel back in time, which method of doing history is currently the most underrated, whose biography she’ll write, how we know what old Norse music was like, why women scholars helped us understand Viking metaphysics, why Diderot’s Jacques the Fatalist is an interesting work, what people misunderstand about the inquisition(s), why science fiction doesn’t have higher social and literary status, which hive she would belong to in Terra Ignota, what the new novel she’s writing is about, and more.

Here is one excerpt:

COWEN: De Sade — where does that come from? What are the influences on de Sade as a writer?

PALMER: Thomas Aquinas. No, lots and lots of things, but he’s very interested in the large philosophical milieu in the period. Remember that the 18th century is a moment when the clandestine bookshop is a major, major thing. And if anyone enjoys and is interested in the history of censorship and clandestine publishing, I can’t recommend enough the work of Robert Darnton, a brilliant, brilliant historian of clandestine literature.

But the same underground bookshops sell all underground materials, which means an underground bookshop sells pornography, and it also sells Voltaire and Rousseau, and it also sells diatribes criticizing the king, and it also sells radical Jansenist theological pamphlets about whether the Holy Spirit derives from the Father and Son equally or from the Father alone.

The same kinds of people frequent these shops, and the same kinds of people buy things. So, think about how, when you go into a Barnes & Noble, the science fiction and fantasy section is one section, even though science fiction and fantasy are different things. But they have a lot of overlap, both in the overlap of readership and in overlap in books that have both science fiction and fantasy elements. It was perfectly natural, in the same way, for clandestine bookshops to generate these works that are pornography and radical philosophy at the same time. They’re printed by the same printers, sold to the same audiences, and circulate in the same places.

De Sade uses his extreme pornography to get at questions of morality, ethics, and artificiality. What are the ethics of hurting each other? Why do we feel that way about hurting each other? What are so-called natural impulses, as John Locke and Hobbes were very dominant at the time, or Descartes, who is differently dominant at the time in rivalry with them? They make claims about the natural human impulses or the natural character of a human being. What does extreme sexuality show us about how that character might be broader than it is?

I mean it when I say Thomas Aquinas, right? One of Thomas Aquinas’s traditional proofs of the existence of God is that everything he sees around him in nature — this also is one that Aristotle uses, but Aquinas articulates it in the most famous way for de Sade’s period — that when we look around us, it’s clear that everything is designed to work.

Interesting throughout.

New issue of Econ Journal Watch

This is now twenty years of Econ Journal Watch, congratulations to Dan Klein!  Here is the table of contents:

Volume 20, Number 2, September 2023

Screening the 1915 film The Birth of a Nation: In an article in the American Economic Review, Desmond Ang purports to show causal impact of screenings of the film between 1915 and 1919 on lynchings, on the formation and growth of Ku Klux Klan chapters between 1920 and 1925, and on hate crimes in the early 2000s. Here, concurring that the film itself reeks of racism, Robert Kaestner scrutinizes Ang’s data and analyses, and challenges the claims of causal evidence of effects from 1915–1919 screenings of the film. (Note: Professor Ang was not invited to reply for concurrent publication because Kaestner’s piece was finalized at too late a date. Professor Ang is invited to reply in a future issue.)

Temperature-economic growth claims tested again: Having tested temperature-economic growth claims previously in this journal (here and here), David Barker now reports on his investigation into much-cited articles by Melissa Dell, Benjamin Jones, and Benjamin Olken, published in the American Economic Review in 2009 and the American Economic Journal: Macroeconomics in 2012. As with the two previous pieces by Barker, the commented-on authors have declined to reply (the invitation remains open).

Debating the causes of the Ukraine famine of the early 1930s: Two scholars interpret the complex causes of a tragedy that caused the loss of perhaps three million souls. Natalya Naumenko’s research on the causes of the Ukraine famine is discussed by Mark Tauger, and Naumenko replies.

Ergodicity economics, debated: A number of scholars have advanced an approach to decision making under uncertainty called ergodicity economics. A critique is provided here by Matthew Ford and John Kay, who maintain that psychology is fundamental to any general theory of decision making under uncertainty. Eleven proponents of ergodicity economics have coauthored a reply. They suggest that the critique is based on an incomplete understanding of ergodicity economics, and point to two sources of misunderstanding. The replying authors are Oliver Hulme, Arne Vanhoyweghen, Colm Connaughton, Ole Peters, Simon Steinkamp, Alexander Adamou, Dominik Baumann, Vincent Ginis, Bert Verbruggen, James Price, and Benjamin Skjold.

Dispute resolution on hospitals, communication, and dispute resolution? Previously, Florence R. LeCraw, Daniel Montanera, and Thomas A. Mroz (LMM) criticized the statistical methods of a 2018 article in Health Affairs. Here, Maayan Yitshak-Sade, Allen Kachalia, Victor Novack, and Michelle M. Mello provide a reply to LMM, and LMM provide a rejoinder to them.

Aaron Gamino rejoins on health insurance mandates and the marriage of young adults: Previously, Aaron Gamino commented on the statistical modeling in a 2022 Journal of Human Resources article, whose authors, Scott Barkowski and Joanne Song McLaughlin, replied. Here now Gamino provides a rejoinder.

A History of Classical Liberalism in the Netherlands: Edwin van de Haar narrates the classical liberal movements in the Netherlands, from the Dutch Golden Age, through the 18th, 19th, and 20th centuries, and down to today. The article extends the series on Classical Liberalism in Econ, by Country.

To Russia with love: The conservative liberal Boris Chicherin (1828–1904) addressed his fellow Russians in an 1857 essay “Contemporary Tasks of Russian Life.” Here, the essay is republished by permission of Yale University Press, with a Foreword by the translator Gary Hamburg.

Pierre de Boisguilbert: Prime Extracts and Some Correspondence: The first great exponent of liberal economics in France was Pierre de Boisguilbert (1646–1714). Here, Benoît Malbranque provides English-language readers with a taste of Boisguilbert, and for the first time.

SSRN and medRxiv Censor Counter-narrative Science: Jay Bhattacharya and Steve Hanke detail the experience of three research teams being censored by SSRN and medRxiv. The article also points to a website (link) where scholars can report their experiences of being censored by SSRN, medRxiv, or other preprint servers.

Journal of Accounting Research’s Report on Its Own Research-Misconduct Investigation of an Article It Published: Dan Klein reports and rebukes the journal.

What are your most underappreciated works? Previously, 18 scholars with 4k+ Google Scholar cites pointed to a decade-or-more old paper with cite count below his or her h-index. Now, they are joined by Andrew Gelman, Robert Kaestner, Robert A. Lawson, George Selgin, Ilya Somin, and Alex Tabarrok.

EJW Audio:

Edwin van de Haar on Classical Liberalism in the Netherlands

Paul Robinson on Russian Liberalism

Vlad Tarko and Radu Nechita on Liberalism in Romania, 1829 to 2023