South Appalachia > North Appalachia
The ARC classifies 27.2 percent of North Appalachian counties as distressed but only 9.6 percent of South Appalachian counties that way. Over 70 percent of counties in South Appalachia have grown in population since the 2020 Census. North Appalachia lost 17,131 people in total, while South Appalachia gained 127,585. The difference in net in-migration is even more stark. While the North posted positive net domestic in-migration of 22,563, the South tallied almost 300,000—13 times as high. The story is similar for jobs, with the North losing 227,049 positions since the pre-pandemic year of 2019, while the South actually exceeded its pre-Covid levels by 66,377. In other words, much of South Appalachia is seeing a population inflow and is growing in both population and employment.
Here is much more from Aaron M. Renn, of interest and with good maps, and for the pointer I thank Terry O’Connor.
Friday assorted links
Driverless Cars May Already Be Safer Than Human Drivers
Tim Lee runs the numbers:
Waymo and Cruise have driven a combined total of 8 million driverless miles, including more than 4 million in San Francisco since the start of 2023.
And because California law requires self-driving companies to report every significant crash, we know a lot about how they’ve performed.
For this story, I read through every crash report Waymo and Cruise filed in California this year, as well as reports each company filed about the performance of their driverless vehicles (with no safety drivers) prior to 2023. In total, the two companies reported 102 crashes involving driverless vehicles. That may sound like a lot, but they happened over roughly 6 million miles of driving. That works out to one crash for every 60,000 miles, which is about five years of driving for a typical human motorist.
These were overwhelmingly low-speed collisions that did not pose a serious safety risk. A large majority appeared to be the fault of the other driver. This was particularly true for Waymo, whose biggest driving errors included side-swiping an abandoned shopping cart and clipping a parked car’s bumper while pulling over to the curb.
Cruise’s record is not impressive as Waymo’s, but there’s still reason to think its technology is on par with—and perhaps better than—a human driver.
Human beings drive close to 100 million miles between fatal crashes, so it’s going to take hundreds of millions of driverless miles for 100 percent certainty on this question. But the evidence for better-than-human performance is starting to pile up, especially for Waymo. And so it’s important for policymakers to allow this experiment to continue. Because at scale, safer-than-human driving technology would save a lot of lives.
Driverless cars never break the speed limit, the driver is never drunk, nor distracted by their cell phone or the fight they had with their spouse. Another advantage that people might not think of is that these cars are far better for cyclists as Parker Conrad notes:
It’s so, so obvious to anyone riding a bike in SF that autonomous vehicles are WAY safer for bicyclists than human drivers. They see me every time; human drivers constantly turn right into the bike lane without thinking.
Why? Because driverless cars literally have eyes in the back of their heads.
Driverless cars are in general less good at edge cases but the advantages add up.
I would qualify this only slightly by noting that some locations are more difficult than others and while San Francisco is quite difficult terrain, Phoenix, Arizona was chosen because of flat terrain and sunny weather. Still, the bottom line is absolutely correct. Driverless cars are safer and more capable than many people think and we should always measure their defects relative to realistic alternatives and not to some idealized notion of perfection.
Latin America fact of the day
The year 2022 was a vintage one for political turmoil in Latin America. Colombia elected a leftwing former guerrilla as president, Chile considered (and rejected) a radical new constitution, Peru’s elected president was impeached and imprisoned pending trial after after a failed attempt to seize extraordinary powers and Brazil’s far-right leader Jair Bolsonaro narrowly lost a bid for re-election.
It was also a record year for foreign direct investment. Investors committed $225bn to Latin America and the Caribbean in 2022, according to ECLAC, the UN’s economic agency for the region. That was 55 per cent more than the previous year and comfortably surpassed the previous peak a decade earlier. Part of the increase was a post-pandemic rebound but the number of future projects announced also rose, though more modestly.
And this:
Brazil was so popular with investors that it became the world’s number five destination for foreign investment last year, behind the US, China, Hong Kong and Singapore, according to Unctad. Brazil is also the top developing economy for international renewable energy investment, according to Unctad, with $115bn in projects.
Here is more from Michael Stott at the FT. Latin America is such a mess it is hard to be flat out optimistic. Yet I feel the highlights, such as southern Brazil and parts of Mexico, are going to surprise people on the upside. Mexico has survived terrible governance, and Colombia is likely to as well. Chile has turned back from the abyss. Argentina may be willing to try something different, even if it doesn’t seem on track to succeed. So these are heady days down south, noting that most of the major problems still have not gone away.
An aggregate Bayesian approach to more (artificial) intelligence?
It is not disputed that current AI is bringing more intelligence into the world, with more to follow yet. Of course not everyone believes that augmentation is a good thing, or will be a good thing if we remain on our current path.
To continue in aggregative terms, if you think “more intelligence” will be bad for humanity, which of the following views might you also hold?:
1. More stupidity will be good for humanity.
2. More cheap energy will be bad for humanity.
3. More land will be bad for humanity.
4. More people (“N”) will be bad for humanity.
5. More capital (“K”) will be bad for humanity.
6. More innovation (the Solow residual, the non-AI part) will be bad for humanity.
Interestingly, while there are many critics of generative AI, few defend the apparent converse about more stupidity, namely #1, that we should prefer it.
I am much more worried about #2 — more cheap energy — than I am about more generative AI.
I don’t know anyone worried about “too much land.” Maybe the dolphins?
There have been many people in the past worried about #4 — too many people — but world population will be shrinking soon enough, so that is a moot point.
I do not hear that “more capital” will be bad for humanity. As for innovation, the biggest innovation worriers seem to be the AI worriers, which brings us back to the original topic.
My general view is that if you are worried that more intelligence in the world will bring terrible outcomes, you should be at least as worried about too much cheap energy. What exactly then is it you should want more of?
More land? Maybe we should pave over more ocean, as the Netherlands has done, but check AI and cheap energy, which in turn ends up meaning limiting most subsequent innovation, doesn’t it?
If I don’t worry more about that scenario, it is only because I think it isn’t very likely.
If you worry about bringing too much intelligence into the world, I think you have to be a pretty big pessimist no matter what happens with AI. How many other feasible augmentations can have positive social marginal products if intelligence does not?
Addendum: I have taken an aggregative approach. You might think we need “more intelligence” and also “more AI,” but perhaps in different hands or at different times. In contrast, I think we are remarkably fortunate to be facing the particular combination of parties and opportunities that stand before us today.
Political Sentiment and Innovation
Are these the “animal spirits” they like to talk about? And is this also evidence of the growing politicization of the U.S. economy?:
We document political sentiment effects on US inventors. Democratic inventors are more likely to patent (relative to Republicans) after the 2008 election of Obama but less likely after the 2016 election of Trump. These effects are 2-3 times as strong among politically active partisans and are present even within firms over time. Patenting by immigrant inventors (relative to non-immigrants) also falls following Trump’s election. Finally, we show partisan concentration by technology class and firm. This concentration aggregates up to more patenting in Democrat-dominated technologies (e.g., Biotechnology) compared to Republican-dominated technologies (e.g., Weapons) following the 2008 election of Obama.
That is from a new NBER working paper by Joseph Engleberg, Runjing Lu, William Mullins, and Richard R. Townsend.
Thursday assorted links
1. Scott Alexander defends the notion of automaticity.
2. Well, the Rolling Stones needed a bass player and a drummer… They have to cut “I Wanna’ Be Your Man,” right?
3. “Specifically, we show that men who post often on social media are seen as feminine, a phenomenon we refer to as the “frequent-posting femininity stereotype.”” Link here.
4. Caltech allows new, non-high school class admissions requirements.
5. Zvi on Gemini, and many other AI matters.
New Emergent Ventures Vertical Supports Talent Identification Projects
Here is the press release and more detail, I thank Schmidt Futures and Eddie Mandhry for their support, and there is a parallel initiative led by David Deming and Heidi Williams, focusing on the research side of talent identification (for that do not apply to Mercatus/EV). Applications on the practitioner side for finding and developing talent are welcome at the usual Emergent Ventures site.
Italy fact of the day
Before becoming Italian prime minister, Giorgia Meloni was one of the most strident voices on migration in the European Union. As an opposition politician, she warned darkly of efforts to substitute native Italians with ethnic minorities and promised to put in place a naval blockade to stop migrants crossing the Mediterranean.
During her time in office, she has taken a markedly different tack — presiding over a sharp spike in irregular arrivals and introducing legislation that could see as many as 1.5 million new migrants arrive through legal channels.
Do note this:
Meloni is presiding over a country that is economically stagnant and in demographic decline. Over the last decade, Italy has shrunk by some 1.5 million people (more than the population of Milan). In 39 of its 107 provinces, there are more retirees than workers. ..
Meloni’s legal migration decree estimates Italy needs 833,000 new migrants over the next three years to fill in the gap in its labor force. It opens the door to 452,000 workers over the same period to fill seasonal jobs in sectors like agriculture and tourism as well as long-term positions like plumbers, electricians, care workers and mechanics…
Given Italy’s rules on family reunification, which allow residents to bring in relatives, “it’s easy to predict that over something like 10 years, these figures will triple,” bringing in about 1.5 million migrants, said Maurizio Ambrosini, a professor of sociology and an expert on migration at Milan’s university.
The median voter surfaces yet again? Here is the full account, via Andrew McLoughlin.
Wednesday assorted links
My Conversation with Vishy Anand
In Chennai I recorded with chess great Vishy Anand, here is the transcript, audio, and video, note the chess analysis works best on YouTube, for those of you who follow such things (you don’t have to for most of the dialogue). Here is the episode summary:
Tyler and Vishy sat down in Chennai to discuss his breakthrough 1991 tournament win in Reggio Emilia, his technique for defeating Kasparov in rapid play, how he approached playing the volatile but brilliant Vassily Ivanchuk at his peak, a detailed breakdown of his brilliant 2013 game against Levon Aronian, dealing with distraction during a match, how he got out of a multi-year slump, Monty Python vs. Fawlty Towers, the most underrated Queen song, how far to take chess opening preparation, which style of chess will dominate in the next ten years, how AlphaZero changes what we know about the game, the key to staying a top ten player at age 53, why he thinks he’s a worse loser than Kasparov, qualities he looks for in talented young Indian chess players, picks for the best places to eat in Chennai, and more.
Here is one excerpt:
COWEN: Do you hate losing as much as Kasparov does?
ANAND: To me, it seems he isn’t even close to me, but I admit I can’t see him from the inside, and he probably can’t see me from the inside. When I lose, I can’t imagine anyone in the world who loses as badly as I do inside.
COWEN: You think you’re the worst at losing?
ANAND: At least that I know of. A couple of years ago, whenever people would say, “But how are you such a good loser?” I’d say, “I’m not a good loser. I’m a good actor.” I know how to stay composed in public. I can even pretend for five minutes, but I can only do it for five minutes because I know that once the press conference is over, once I can finish talking to you, I can go back to my room and hit my head against the wall because that’s what I’m longing to do now.
In fact, it’s gotten even worse because as you get on, you think, “I should have known that. I should have known that. I should have known not to do that. What is the point of doing this a thousand times and not learning anything?” You get angry with yourself much more. I hate losing much more, even than before.
COWEN: There’s an interview with Magnus on YouTube, and they ask him to rate your sanity on a scale of 1 to 10 — I don’t know if you’ve seen this — and he gives you a 10. Is he wrong?
ANAND: No, he’s completely right. He’s completely right. Sanity is being able to show the world that you are sane even when you’re insane. Therefore I’m 11.
COWEN: [laughs] Overall, how happy a lot do you think top chess players are? Say, top 20 players?
ANAND: I think they’re very happy.
Most of all, I was struck by how good a psychologist Vishy is. Highly recommended, and you also can see whether or not I can keep up with Vishy in his chess analysis. Note I picked a game of his from ten years ago (against Aronian), and didn’t tell him in advance which game it would be.
The Relentless Rise of Stablecoins
1. In 2022, stablecoins settled over $11tn onchain, dwarfing the volumes processed by PayPal ($1.4tn), almost surpassing the payment volume of Visa ($11.6tn), and reaching 14% of the volume settled by ACH and over 1% the volume settled by Fedwire. It is remarkable that in just a few years, a new global money movement rail can be compared with some of the world’s largest and most important payment systems.
2. Over 25mm blockchain addresses hold over $1 in stablecoins. Of these, ~80%, or close to 20mm addresses, hold between $1 and $100. For a sense of scale, a US bank with 25mm accounts would rank as the 5th largest bank in the US by number of accounts. The massive number of small-dollar stablecoin holdings indicates the potential for stablecoins to provide global financial services to customers underserved by traditional financial institutions.
3. Approximately 5mm blockchain addresses send stablecoins each week. This number provides a very rough proxy for global users regularly interacting with stablecoins. These ~5mm weekly active addresses send ~38mm stablecoin transactions each week, representing an average of over 7 weekly transactions per active address.
4. Stablecoin usage has decoupled from crypto exchange volumes, indicating that significant stablecoin transaction volumes may be driven by non-trading/speculative activity. Since December 2021, centralized exchange volumes are down 64%, and decentralized exchange volumes are down 60%. During this period, stablecoin volumes are down only 11%, and weekly active stablecoin addresses and weekly stablecoin transactions are up over 25%.
5. Of the ~5mm weekly active stablecoin addresses, ~75% transact less than $1k per week, indicating that small/retail users likely represent the majority of stablecoin users.
6. The outstanding supply of stablecoins has grown from less than $3bn five years ago to over $125bn today (after peaking at over $160bn) and has shown resilience to the market downturn with the market cap of stablecoins currently down ~24% from its peak, compared with a ~57% decline for the overall crypto market cap.
7. Less than 1/3rd of stablecoins are held on exchanges. Most are held in externally owned accounts (not exchanges or smart contracts).
8. The majority of stablecoin activity uses Tether (USDT). Tether represents 69% of stablecoin supply, and YTD has accounted for 80% of weekly active addresses, 75% of transactions, and 55% of volumes.
9. Most stablecoin activity occurs on the Tron and BSC blockchains. Year-to-date, the Tron and BSC blockchains collectively account for 77% of weekly active addresses, 75% of transactions, and 41% of volumes.
10. The Ethereum blockchain is used for higher value transactions (on average). Despite accounting for just 6% of active wallets and 3% of transactions, the Ethereum blockchain is home to 55% of stablecoin supply and settles close to 50% of weekly stablecoin $ volume.
These are all from a Bevan Howard report, The Relentless Rise of Stablecoins (requires email).
*The Conservative Futurist*
The author is James Pethokoukis, and the subtitle is How to Create the Sci-Fi World We Were Promised.
All excellent stuff…due out October 3rd…
What I’ve been reading
R.C. Zaehner, Concordant Discord: The Interdependence of Faiths Being the Gifford Lectures on Natural Religion Delivered at St. Andrews in 1967-1969. Half of this volume is amazing, the other half meandering. The best parts are on Hinduism and Buddhism, and how they can best be understood in relation to Western religions. Zaehner has an amazing Wikipeda page, and I have ordered other books by him, the ultimate act of literary flattery.
James Stafford, The Case of Ireland: Commerce, Empire and the European Order, 1750-1848. An excellent and well-researched books, most interesting on the Irish Union of 1800-1801 and how and why so many classical liberals favored it. What did they get wrong? Or did they? Consistently instructive on earlier Irish thought on trade as well.
Victoria Houseman, American Classicist: The Life and Loves of Edith Hamilton. A good and fun book. I hadn’t known that she was very likely bisexual, or that she was good friends with Felix Morley and Robert Taft. Interesting throughout, and drives home the point about just how early Hamilton did her most important work on mythology. It remains widely read today.
Harvey Sachs, Schoenberg: Why He Matters. A very good introduction to a composer who truly matters. Also a good (short) portrait of Vienna at that time. Maybe it won’t “sell you” on Schoenberg, but it will make his advocates (I am one of them) seem far less crazy. It also admits that a lot of his work wasn’t that good, and helps you separate the better from the worse.
Henry Farrell and Abraham Newman, Underground Empire: How America Weaponized the World Economy. This book in preprint form was well ahead of its time, and now it is coming out in a super-timely fashion.
Lorraine Byrne Bodley, Schubert: A Musical Wayfarer. Super-thorough, everything about Schubert and most of all his music.
I have not read the new novel by Bradley Tusk, namely Obvious in Hindsight, about the attempted introduction of flying cars and the regulatory obstacles that arose (among other dramatic events).
Lunana, A Yak in the Classroom is the only and also the best Bhutanese movie I have seen, ever. Recommended, gives you a real look at the country, both rural and urban [sic].
AI update
We’re in that “a lot of people have stopped caring” phase of the transition, don’t let the status quo and recency biases fool you:
Google’s new flagship AI model, “Gemini,” is set to be a direct competitor to GPT-4 and boasts computing power 5 times that of GPT-4. Trained on Google’s TPUv5 chips, it’s capable of simultaneous operations with a massive 16,384 chips. The dataset used for training this model is around 65 trillion tokens, and it’s multi-modal, accepting text, video, audio, and pictures. Moreover, it can produce both text and images. The training also included content from YouTube and used advanced training techniques similar to “AlphaGo-type” methods. Google plans to release the Gemini model to the public in December 2023.
And that may or may not prove the most important event of that half year…