Triumph of the Data Hoarders

James Bailey3 Comments

Several major datasets produced by the federal government went offline this week. Some, like the Behavioral Risk Factor Surveillance Survey and the American Community Survey, are now back online; probably most others will soon join them. But some datasets that the current administration considers too DEI-inflected could stay down indefinitely.

This serves as a reminder of the value of redundancy- keeping datasets on multiple sites as well as in local storage. Because you never really know when one site will go down- whether due to ideological changes, mistakes, natural disasters, or key personnel moving on.

External hard drives are an affordable option for anyone who wants to build up their own local data hoard going forward. The Open Science Foundation site allows you to upload datasets up to 50 GB to share publicly; that’s how I’ve been sharing cleaned-up versions of the BRFSS, state-levle NSDUH, National Health Expenditure Accounts, Statistics of US Business, and more. If you have a dataset that isn’t online anywhere, or one that you’ve cleaned or improved to the point it is better than the versions currently online, I encourage you to post it on OSF.

If you are currently looking for a federal dataset that got taken down, some good places to check are IPUMS, NBER, Archive.org, or my data page. PolicyMap has posted some of the federal datasets that seem particularly likely to stay down; if you know of other pages hosting federal datasets that have been taken down, please share them in the comments.

Was the US at Our Richest in the 1890s?

Jeremy Horpedahl2 Comments

Donald Trump has repeatedly said that the US was at our “richest” or “wealthiest” in the high-tariff period from 1870-1913, and sometimes he says more specifically in the 1890s. Is this true?

First, in terms of personal income or wealth, this is nowhere near true. I’ve looked at the purchasing power of wages in the 1890s in a prior post, and Ernie Tedeschi recently put together data on average wealth back to the 1880s. As you can probably guess, by these measures Trump is quite clearly wrong.

So what might he mean?

One possibility is tax revenue, since he often says this in the context of tariffs versus an income tax. Broadly this also can’t be true, as federal revenue was just about 3% of GDP in the 1890s, but is around 16% in recent years.

But perhaps it is true in a narrower sense, if we look at taxes collected relative to the country’s spending needs. Trump has referenced the “Great Tariff Debate of 1888” which he summarized as “the debate was: We didn’t know what to do with all of the money we were making. We were so rich.” Indeed, this characterization is not completely wrong. As economic historian and trade expert Doug Irwin has summarized the debate: “The two main political parties agreed that a significant reduction of the budget surplus was an urgent priority. The Republicans and the Democrats also agreed that a large expansion in government expenditures was undesirable.” The difference was just over how to reduce surpluses: do we lower or raise tariffs?

It does seem that in Trump’s mind being “rich” in this period was about budget surpluses. Let’s look at the data (I have truncated the y-axis so you can actually read it without the WW1 deficits distorting the picture, but they were huge: over 200% of revenues!):

It is certainly true that under parts of the high-tariff period, we did collect a lot of revenue from tariffs! In some years, federal surpluses were over 1% of GDP and 30% of revenues collected. But notice that this is not true during Trump’s favored decade, the 1890s. Following the McKinley Tariff of 1890, tariff revenue fell sharply (though probably not likely due to the tariff rates, but due to moving items like sugar to the duty-free list, as Irwin points out). The 1890s were not a decade of being “rich” with tariff revenue and surpluses.

Finally, also notice that during the 1920s the US once again had large budget surpluses. The income tax was still fairly new in the 1920s, but it raised around 40-50% of federal revenue during that decade. By the Trump standard, we (the US federal government) were once again “rich” in the 1920s — this is true even after the tax cuts of the 1920s, which eventually reduced the top rate to 25% from the high of 73% during WW1.

If we define a country as being “rich” when it runs large budget surpluses, the US was indeed rich by this standard in the 1870s and 1880s (though not the 1890s). But it was rich again by this standard in the 1920s. This is just a function of government revenue growing faster than government spending. And the growth of revenue during the 1870s and 1880s was largely driven by a rise in internal revenue — specifically, excise taxes on alcohol and tobacco (these taxes largely didn’t exist before the Civil War).

1890 was the last year of big surpluses in the nineteenth century, and in that year the federal government spent $318 million. Tariff revenue (customs) was just $230 million. There was only a surplus in that year because the federal government also collected $108 million of alcohol excise taxes and $34 million of tobacco excise taxes. In fact, throughout the period 1870-1899, tariff revenues are never enough to cover all of federal spending, though they do hit 80% in a few years (source: Historical Statistics of the US, Tables Ea584-587, Ea588-593, and Ea594-608):

One more thing: in some of these speeches, Trump blames the Great Depression on the switch from tariffs to income taxes. In addition to there really being no theory for why this would be the case, it just doesn’t line up with the facts. The 1890s were plagued by financial crises and recessions. The 1920s (the first decade of experience with the income tax) was a period of growth (a few short downturns) and as we saw above, large budget surpluses. The Great Depression had other causes.

After the Fall: What Next for Nvidia and AI, In the Light of DeepSeek

Scott BuchananLeave a comment

Anyone not living under a rock the last two weeks has heard of DeepSeek, the cheap Chinese knock-off of ChatGPT that was supposedly trained using much lower resources that most American Artificial Intelligence efforts have been using. The bearish narrative flowing from this is that AI users will be able to get along with far fewer of Nvidia’s expensive, powerful chips, and so Nvidia sales and profit margins will sag.

The stock market seems to be agreeing with this story. The Nvidia share price crashed with a mighty crash last Monday, and it has continued to trend downward since then, with plenty of zig-zags.

I am not an expert in this area, but have done a bit of reading. There seems to be an emerging consensus that DeepSeek got to where it got to largely by using what was already developed by ChatGPT and similar prior models. For this and other reasons, the claim for fantastic savings in model training has been largely discounted. DeepSeek did do a nice job making use of limited chip resources, but those advances will be incorporated into everyone else’s models now.

Concerns remain regarding built-in bias and censorship to support the Chinese communist government’s point of view, and regarding the safety of user data kept on servers in China. Even apart from nefarious purposes for collecting user data, ChatGPT has apparently been very sloppy in protecting user information:

Wiz Research has identified a publicly accessible ClickHouse database belonging to DeepSeek, which allows full control over database operations, including the ability to access internal data. The exposure includes over a million lines of log streams containing chat history, secret keys, backend details, and other highly sensitive information.

Shifting focus to Nvidia – – my take is that DeepSeek will have little impact on its sales. The bullish narrative is that the more efficient algos developed by DeepSeek will enable more players to enter the AI arena.

The big power users like Meta and Amazon and Google have moved beyond limited chatbots like ChatGPT or DeepSeek. They are aiming beyond “AI” to “AGI” (Artificial General Intelligence), that matches or surpasses human cognitive capabilities across a wide range of cognitive tasks. Zuck plans to replace mid-level software engineers at Meta with code-bots before the year is out.

For AGI they will still need gobs of high-end chips, and these companies show no signs of throttling back their efforts. Nvidia remains sold out through the end of 2025. I suspect that when the company reports earnings on Feb 26, it will continue to demonstrate high profits and project high earnings growth.

Its price to earnings is higher than its peers, but that appears to be justified by its earnings growth. For a growth stock, a key metric is price/earnings-growth (PEG), and by that standard, Nvidia looks downright cheap:

Source: Marc Gerstein on Seeking Alpha

How the fickle market will react to these realities, I have no idea.

The high volatility in the stock makes for high options premiums. I have been selling puts and covered calls to capture roughly 20% yields, at the expense of missing out on any rise in share price from here.

Disclaimer: Nothing here should be considered as advice to buy or sell any security.

Shocked

mdmakowsky1 Comment

Two weeks in and it’s safe to say the United States federal government has been shocked out of it’s previous equilibrium (whether that shock is “exogenous” is honestly besides the point). Some thoughts, in no particular order

The federal talent drain is going to get even worse

At some point in the last 100 years the equilibrium strategy for the government has been to pay employees in the non-pecuniary benefits of a) job security, b) status, c) retirement d) pro-social civic pride, and e) still more job security. Almost none of that remains wholly intact. The previous bundle of non-pecuniaries resulted in a federal labor force where, glibly estimated, 20% of the employees did 80% of the work. The federal government functioned off the talent and committment of employees whose non-pecuniary preferences led them to forego considerable amounts of income in the private sector. Not sure who’s going to stick around or start a career in the federal government at this point, but I expect the selection effects to be sometimes darkly tragicomic, but mostly just tragic. People have already been hurt. More people will continue to be hurt.

The shrill cranks were right

It’s time for a lot of people to start publicly accepting the fact that the new administration is actually running an authoritarian playbook. Words like “fascism” are neither shrill nor overwrought. Is it unfortunate that people have being making accusations of fascist intent everyday for the last 30 years? Yes, but just because they were wrong then doesn’t mean it’s inappropriate now. The stopped clock is in fact right twice a day. Guess what time it is?

On raptors and resistance

If you’re looking for metaphor instead of adjectives, the new administration are raptors testing their cages for weakpoints, seeing what they can get away with. The bad news is that they are finding no shortage of potential weaknesses to advance their agenda. The good news is that finding and exploiting weaknesses takes time. If we are willing to accept that tariffs are going to impose a lot of price-related pain on consumers and, as the previous round of elections around the world has evidenced, voters do in fact punish incumbents for consumer pain, then the optimal strategy is to merely survive the next 206 weeks with as little damage as possible. So how do we do that?

Put simply, waste time. The entire opposition strategy should be to force the administration to use as much time as possible at every step. Procedural, judicial, and legislative moves are all available. Aspiring fascists they may be, but they are not particularly competent fascists. These people are not grinding out 16 hour work days to write air tight executive orders. They are not career bureacrats who know exactly what buttons to push. They are carnival barkers, reality tv producers, third-rate social media influencers, and niche celebrities. Every time they make a misstep, design something poorly, and have to rescind it 44 hours later? That’s a win. It’s wasted time on a ticking clock that they will never get back. It doesn’t feel like a win because it imposed a lot of pain on a lot of people, but that pain fell well short of the administration’s ambitions.

This works for the tariffs as well. This is not the 18th century where you would simply put someone with a coin purse on the docks to inspect and collect tariffs from every ship that came to port. Modern supply chains are outrageously complex. Collecting tariffs effectively requires institutional infrastructure closer to a VAT tax. Do you really they think these people are going to design it in a manner impervious to bureaucratic and market resistance on the first or second try? Resistance means tying things up in courts on one side while publicly broadcasting the loopholes for the marketplace on the other. Resistance means not just smiling when Canada designs retaliatory tariffs that target “red state” produced goods, but actively broadcasting and supporting that targeting (he wrote while living in a red state and knows he should probably stock up on maple syrup).

Complaining in Stereo

Incumbents lost around the world because nothing pierces rational voter ignorance quite like inflation. Unemployment is salient, but until you hit ~8% or more it might not be sufficiently pervasive to move enough votes. The converse is even more true – it’s almost impossible to get credit for high employment because all you really know is that you have a job which you would have had anyway because you are good and smart and deserve to have a job. Higher prices though, those are always someone else’s fault. The current adminstration blamed Democrats and foreigners. Now it’s the new opposition’s turn to blame Republicans and incompetent public figures in the bureacracy. When consumers take it on the chin, the opposition needs to amplify, amplify, amplify. If there is one thing that seems to be universally true in the modern social media age, it’s that few things are as welcomed by the audience as anxiety and anger. People love to complain. I see no reason not to feed that complaining.

Using Taylor Swift to teach about Adam Smith

Joy BuchananLeave a comment

It’s a niche thing, but Art Carden and I wrote a collection of Taylor Swift/Adam Smith essays. I’m going to use some for teaching this semester, so I wanted to post this in case it’s useful for other teachers.

In introductory economics courses, students often encounter Adam Smith as a one-dimensional figure – the patron saint of self-interest who wrote about the “invisible hand” of the market. But Smith was a far more nuanced thinker, and his insights about human nature remain relevant today. The challenge is making these complex ideas accessible to modern undergraduates.

That’s where this comes in as a teaching aid. Through three recent articles examining Swift’s very public decisions and artistic output, we can introduce students to Smith’s key ideas in a way that feels immediately relevant and engaging. From Swift’s struggles with public perception in “Anti-Hero” to the economic implications of her homemade cinnamon rolls, these pieces provide concrete, contemporary examples that illuminate Smith’s dual role as both moral philosopher and economic thinker. Many undergraduates are already familiar with Swift’s music and public persona, providing an accessible entry point to Smith’s more abstract concepts.

Here’s the recommended order to introduce our articles and a blurb on what you can learn (seriously).

“Anti-Hero as a Smithian Anthem” – This article introduces Smith’s foundational concept of the impartial spectator and his sophisticated view of human nature through a contemporary example. The article demonstrates that Smith wasn’t just an economist but a moral philosopher who understood how deeply humans care about others’ perceptions of them, showing students that economics isn’t just about money.

“Taylor Swift & The World’s Most Expensive Cinnamon Rolls” – This piece provides a bridge between Smith’s moral philosophy and his economic thinking, using opportunity cost analysis while simultaneously showing how rational economic actors might “inefficiently” spend time on activities that build social bonds. The article illustrates how Smith’s ideas about sympathy and social connection exist alongside, not in opposition to, his economic insights about specialization and efficiency.

“Would Adam Smith Tell Taylor Swift to Attend the Super Bowl?” – This article builds on the previous readings to explore the full complexity of Smith’s thought, showing how his ideas from both The Theory of Moral Sentiments and The Wealth of Nations can be applied to analyze real-world decisions.

How FRASER Enhances Economic Research and Analysis

Zachary BartschLeave a comment

Most of us know about FRED, the Federal Reserve Economic Data hosted by the Federal Reserve of St. Louis. It provides data and graphs at your fingertips. You can quickly grab a graph for a report or for a online argument. Of course, you can learn from it too. I’ve talked in the past about the Excel and Stata plugins.

But you may not know about the FRED FRASER. From their about page, “FRASER is a digital library of U.S. economic, financial, and banking history—particularly the history of the Federal Reserve System”. It’s a treasure trove of documents. Just as with any library, you’re not meant to read it all. But you can read some of it.

I can’t tell you how many times I’ve read a news story and lamented the lack of citations –  linked or unlinked.  Some journalists seem to do a google search or reddit dive and then summarize their journey. That’s sometimes helpful, but it often provides only surface level content and includes errors – much like AI. The better journalists at least talk to an expert. That is better, but authorities often repeat 2nd hand false claims too. Or, because no one has read the source material, they couch their language in unfalsifiable imprecision that merely implies a false claim.

A topical example would be the oft repeated blanket Trump-tariffs. That part is not up for dispute. Trump has been very clear about his desire for more and broader tariffs. Rather, economic news often refers back to the Smoot-Hawley tariffs of 1930 as an example of tariffs running amuck. While it is true that the 1930 tariffs applied to many items, they weren’t exactly a historical version of what Trump is currently proposing (though those details tend to change).

How do I know? Well, I looked. If you visit FRASER and search for “Smoot-Hawley”, then the tariff of 1930 is the first search result. It’s a congressional document, so it’s not an exciting read. But, you can see with your own eyes the diversity of duties that were placed on various imported goods. Since we often use the example of imported steel and since the foreign acquisition of US Steel was denied, let’s look at metals on page 20 of the 1930 act. But before we do, notice that we can link to particular pages of legislation and reports – nice! Reading the Smoot-Hawley Tariff Act’s original language, we can see the diverse duties on various metals. Here are a few:

Continue reading →

The Big Ideas

James Bailey2 Comments

Do I really think that the things I write about here and in my papers are the most important things in the world? No. Like most academics, I tend to emphasize the issues where I think I bring a unique perspective, rather than most important issues. But if you don’t realize this, you might get the impression that I think the things I normally talk about are the most important, rather than simply the most neglected and tractable / publishable. I don’t work on the most important issues because I see no good way for me to attack them- but if you do see a way, that is where you should focus. So what are the big issues of the 2020’s?

I see two issues that stand out above the many other important events of the day:

  • Artificial Intelligence: At minimum, the most important new technology in a generation; has the potential to bring about either utopia or dystopia. Do you have ideas for how to nudge it one way or another?
  • Rise of China: From extreme poverty to the world’s manufacturing powerhouse in two generations. What lessons should other countries learn from this for their own economic policy? How can we head off a world war and/or Chinese hegemony?

Focusing a bit more on economics, I see two perennial issues where there could be new opportunities to solve vital old questions:

  • Economic Development: We still don’t have a definitive answer to Adam Smith’s founding question of economics- why are some countries rich while other countries are poor, and how can the poor countries become rich? I think economic freedom is still an underrated answer, but even if you agree, the question remains of how to advance freedom in the face of entrenched interests who benefit from the status quo.
  • Robust Prediction: How can we make economics into something resembling a real science, one where predictions that include decimal places don’t deserve to be laughed at? Can you find a way to determine how much external validity an experiment has? Or how to use machine learning to get at causality? Or at least push existing empirical research to be more replicable?

I’ve added these points to my ideas page, since all this was inspired by me talking through the ideas on the page with my students and realizing how small and narrow they all seemed. Yes, small and narrow ideas are currently easier to publish in economics, but there is more to research and life than easy publications.

Reblog: One acceptable truth or a million fantasies

mdmakowskyLeave a comment

I’m in Houston to give a talk on “Ability to Pay” reforms for how fines and fees are assigned in the criminal justice system, so I’m taking the opportunity to economize on my scarce time i.e. be lazy.

This post received renewed interest in the last week thanks to a vastly superior stating of the hypothesis by Zach Weinersmith. I think it holds up pretty well, title aside, whose connection to the actual material is, at best, unnecessarily oblique and high-handed.

One acceptable truth or a million fantasies (12/28/20)

Humans are soft, slow, and (to the best of my knowledge) make for fairly nutritious meals. Brains for tool-making, and the opposable thumbs for using them, are significant evolutionary adaptations, but it is our capacity to act collectively that placed us at the top of the food chain.

By the end of a standard undergraduate economics curriculum, one couldn’t be blamed for coming to the conclusion that the failures of collective action are the greatest obstacle to mankind – Oh what we could have accomplished if only we had ever found a way to just cooperate. Alas, all those externalities, Prisoners’ Dilemmas, free riders, easy riders, market failures, government failures, they just stopped us at every turn…

I’m not doubting the pedagogical value of teaching any of these obstacles, I teach them myself, but I believe we spend insufficient time reminding students that humans have been solving collective action problems with great success for thousands of years. Every national government, book club, homeowners association, and sorority has managed to produce public goods. So has every military coup and angry mob (if only sometimes for fleeting moments), but collective action is collective action, regardless of how we may feel about the outcome.

More often than not the most interesting question to me isn’t can a collective action problem be solved, but rather i) how has it already been solved and ii) how is that solution going to be threatened or hijacked? When I look to the current political landscape and the only mildly-exaggerated state of political and social polarization, I see not just rival ideologies, but alternative strategies for engendering and ensuring cooperation. On the left, I observe greater recent emphasis on purity – there is a narrow band of acceptable truth and any deviation from that, be it however accidental or benign in intent, can lead to significant punishments, including purges colloquially referred to as cancellations. On the right, I see required public professing of incorrect, often seemingly absurd, beliefs. I might talk about purity tests and purges on the left another times. What I’m interested in at the moment are the public untruths of current right wing identities (broadly conceived) and how they fit into the sacrifice and stigma theory, or club theory, of religion.**

I’ve written a lot about sacrifice and stigma theory. It has become the hammer than has left me forever searching for nails. Originally put forth by Laurence Iannaccone in 1992, it is nothing short of brilliant to my mind. A tool for solving collective problems so profound that when it shows up we barely notice it, and where it shows up tends to be the most powerful clubs shaping our societies: the religious, martial, and extremist political groups that bend the arc of history.

Groups produce what we call “club goods” i.e. public goods only accessible to members of the group. What Iannaccone demonstrated was that a group could actually increase their production of club goods by burdening its members with completely unproductive costs. Why do religious groups require clothing, behavior, or language that could stigmatize their members in broader society? Why are members required to sacrifice their resources at the literal or figurative altar of the group? Because if you impair members’ private productivity, or if the fruits of that private production are skimmed away, they will invest more of their resources into the group. If all group members face these same altered incentives, guess what, you’ve solved the collective action problem!

When I see educated women and men declaring the earth is 5,000 years old, that evolution isn’t real, that climate change is a hoax, or that Donald Trump is a brilliant human being, what I see is public profession of beliefs that might limit social or even occupational opportunities and, in turn, further commit them to a specific subset of affiliations. In the constellation of beliefs that might end up as political shibboleths, of course, there stand to be some more costly than others. In fact, there might even be beliefs that impose negative externalities on others, such antipathy towards vaccines or mask-wearing during a pandemic. Excessive burden might hurt the group, of course – remember, club membership must to be a net gain to persist. In a polarized society, however, vitriol created in rival factions by the externality-generating belief could actually intensify the commitment of group members. The liberals hate real-Americans like me so much now, they’d never accept me as anything but a dumb redneck, so the rational thing to do is double down on my commitment to the only group that will have me. Beliefs that reduce private productivity, increase group productivity, and create long-run antipathy in rival groups can serve to create something incredibly valuable to the group: a captured membership. If there is one thing that is evolutionarily hard-wired into human beings it is the knowledge that isolation is death. A member so stigmatized by past public behavior that rival groups would never accept them stands to be very committed to the group going forward.

The vulnerability of sacrifice and stigma born of public adherence to false beliefs, however, is the capacity of leaders to incept preferred false beliefs into the dogma. This is one way that minority groups can become scapegoated, the carbon costs of fossil fuels denied, quack remedies pedaled, or the reliability of electoral institutions undermined. Religious texts exist (mostly) unedited for long periods of time for a very important reason: core rules of behavior, methods of tithing, and sets of beliefs must be inoculated against opportunistic actors who would hijack the club goods they produce.

Sacrifice and stigma through club-specific false beliefs is a dangerous strategy for political parties for the simple reason that without the constraints of fact or scripture, leaders will feel the pull of their own preferences. Far more dangerous however, is the megalomaniacal conman that any political party institutionally designed to demand cognitive dissonance of its members will eventually attract. Political parties need to solve collective action problems, yes, but they also need immune systems. One might point to social norms, both within and outside the group, as key means of protection. Recent years, however, would seem to suggest that norms are not sufficiently robust in the long run. The US court system has held up well, and has in many ways served as the nations constitutional immune system. Perhaps the major political parties should consider updating and reinforcing their own constitutions, and put in place mechanisms to protect themselves from the next inevitable invasion.

American political parties need to update and upgrade their immune systems.

Inspiring research:

Iannaccone, Laurence R. “Sacrifice and stigma: Reducing free-riding in cults, communes, and other collectives.” Journal of political economy 100.2 (1992): 271-291.

Aimone, Jason A., Laurence R. Iannaccone, Michael D. Makowsky, and Jared Rubin. “Endogenous group formation via unproductive costs.” Review of Economic Studies 80, no. 4 (2013): 1215-1236.

**Note: this is not to suggest that left-wing identity affiliations don’t utilize sacrifice and stigma mechanisms. There is no shortage of what I suspect are completely ineffective, but highly visible, ostensibly pro-environment behaviors that are demanded. But the “headline” mechanisms of herding left-of-center identities under the progressive banner look more like threats of exile than sacrifice and stigma.

DeepSeek vs. ChatGPT: Has China Suddenly Caught or Surpassed the U.S. in AI?

Scott Buchanan1 Comment

The biggest single-day decline in stock market history occurred yesterday, as Nvidia plunged 17% to shave $589 billion off the AI chipmaker’s market cap. The cause of the panic was the surprisingly good performance of DeepSeek, a new Chinese AI application similar to ChatGPT.

Those who have tested DeepSeek find it to perform about as well as the best American AI models, with lower consumption of computer resources. It is also available much cheaper. What really stunned the tech world is that the developers claimed to have trained the model for only about six million dollars, which is way, way less than the billions that a large U.S. firm like OpenAI, Google, or Meta would spend on a leading AI model. All this despite the attempts by the U.S. to deny China the most advanced Nvidia chips. The developers of DeepSeek claim they worked with a modest number of chips, models with deliberately curtailed capacities which met U.S. export allowances.

One conclusion, drawn by the Nvidia bears, is that this shows you *don’t* need ever more of the most powerful and expensive chips to get good development done. The U.S. AI development model has been to build more, huge, power-hungry data centers and fill them up with the latest Nvidia chips. That has allowed Nvidia to charge huge profit premiums, as Google and other big tech companies slurp up all the chips that Nvidia can produce. If that supply/demand paradigm breaks, Nvidia’s profits could easily drop in half, e.g., from 60+% gross margins to a more normal (but still great) 30% margin.

The Nvidia bulls, on the other hand, claim that more efficient models will lead to even more usage of AI, and thus increase the demand for computing hardware – – a cyber instance of Jevons’ Paradox (where the increase in the efficiency of steam engines in burning coal led to more, not less, coal consumption, because it made steam engines more ubiquitous).

I read a bunch of articles to try to sort out hype from fact here. Folks who have tested DeepSeek find it to be as good as ChatGPT, and occasionally better. It can explain its reasoning explicitly, which can be helpful. It is open source, which I think means the code or at least the “weights” have been published. It does seem to be unusually efficient. Westerners have downloaded it onto (powerful) PCs and have run it there successfully, if a bit slowly. This means you can embed it in your own specialized code, or do your AI apart from the prying eyes of ChatGPT or other U.S. AI providers. In contrast, ChatGPT I think can only be run on a powerful remote server.

Unsurprisingly, in the past two weeks DeepSeek has been the most-uploaded free app, surpassing ChatGPT.

It turns out that being starved of computing power led the Chinese team to think their way to several important innovations that make much better use of computing. See here and here for gentle technical discussions of how they did that. Some of it involved hardware-ish things like improved memory management. Another key factor is they figured out a way to only do training on data which is relevant to the training query, instead of training each time on the entire universe of text.

A number of experts scoff at the claimed six million dollar figure for training, noting that if you include all the costs that were surely involved in the development cycle, it can’t be less than hundreds of millions of dollars. That said, it was still appreciably cheaper than the usual American way. Furthermore, it seems quite likely that making use of answers generated by ChatGPT helped DeepSeek to rapidly emulate ChatGPT’s performance. It is one thing to catch up to ChatGPT; it may be tougher to surpass it. Also, presumably the compute-efficient tricks devised by the DeepSeek team will now be applied in the West, as well. And there is speculation that DeepSeek actually has use of thousands of the advanced Nvidia chips, but they hide that fact since it involved end-running U.S. export restrictions. If so, then their accomplishment would be less amazing.

What happens now? I wish I knew. (I sold some Nvidia stock today, only to buy it back when it started to recover in after-hours trading). DeepSeek has Chinese censorship built into it. If you use DeepSeek, your information gets stored on servers in China, the better to serve the purposes of the government there.

Ironically, before this DeepSeek story broke, I was planning to write a post here this week pondering the business case for AI. For all the breathless hype about how AI will transform everything, it seems little money has been made except for Nvidia. Nvidia has been selling picks and shovels to the gold miners, but the gold miners themselves seem to have little to show for the billions and billions of dollars they are pouring into AI. A problem may be that there is not much of a moat here – – if lots of different tech groups can readily cobble together decent AI models, who will pay money to use them? Already, it is being given away for free in many cases. We shall see…