War by Other Means
In The Trial of The Chicago Seven, the Aaron Sorkin movie about the group of anti–Vietnam War protesters charged with inciting riots at the 1968 Democratic National Convention, the focus is on the antics of Abbie Hoffman (Sacha Baron Cohen) and Jerry Rubin (Jeremy Strong). It’s a good movie but their story is not the only story. Among the Chicago Seven was an older, quieter, more bemused David Dellinger (played in the movie by John Carroll Lynch). It was not Dellinger’s first trial. In 1940 Dellinger had refused to register for the then-new draft, the first peacetime draft in America’s history, and he had been imprisoned as a conscientious objector and pro-pacifism protester. Dellinger served a year in federal prison in Danbury, CT, and upon his release, he adamantly refused to register once more. He was imprisoned for an additional two years in the maximum-security facility at Lewisburg, Pennsylvania, where he engaged in hunger strikes and endured periods of solitary confinement. Dellinger was the real deal.
In War by Other Means, Daniel Akst recaptures an older generation of anti-war, pro-pacifism protesters; people like Dellinger, the radical Catholic Dorothy Day, Bayard Rustin, Dwight MacDonald and others. This earlier group grew out of the disillusionment that many Americans felt after World War One–they resolved to never again be entangled in European death and destruction.
During the interwar period, moreover, the United States had developed perhaps the largest and best-organized pacifist movement in the world. Pacifism was part of the curriculum at some schools and firmly on the agenda of the mainline Protestant denominations that were such important institutions in the life of this churchgoing nation at the time….Pacifism was well established on campuses thanks to a massive and diverse national student anti-war movement….During the thirties pacifism, probably surpassed even the Depression as the dominant social issue in American liberal Protestantism…
It wasn’t just liberal Protestants, Pacifism also drew on the isolationist tradition:
…isolationism simply wanted to keep America out of other peoples’ bloody conflicts; it advocated strength through preparedness and put faith in the vastness of the oceans to keep us safe. “Isolationism” has become a dirty word since its heyday in the thirties, when it came into common usage. But in fact it started life as a pejorative, one used by American expansionists in the late nineteenth century to tar the righteous killjoys who objected to burgeoning US imperialism…
Regrettably, the unique blend of “left-wing” pacifism and “right-wing” isolationism, once prevalent in America, has largely vanished. Mostly vanished also is–for want of better terms–the Christian and left-wing libertarianism of people like Dellinger, Day and MacDonald, who although being on the left and hardly pro-market had a deep appreciation for individualism and civil society and a fear of the homogenizing and brutalizing role of the state.
Daniel Akst’s War by Other Means is an important and engaging look at a cast of remarkable American characters and their unique blend of ideological pacifism.
Addendum: Nick Gillespie at Reason has a very good interview with Akst.
Social Media as a Bank Run Catalyst
Social media fueled a bank run on Silicon Valley Bank (SVB), and the effects were felt broadly in the U.S. banking industry. We employ comprehensive Twitter data to show that preexisting exposure to social media predicts bank stock market losses in the run period even after controlling for bank characteristics related to run risk (i.e., mark-to-market losses and uninsured deposits). Moreover, we show that social media amplifies these bank run risk factors. During the run period, we find the intensity of Twitter conversation about a bank predicts stock market losses at the hourly frequency. This effect is stronger for banks with bank run risk factors. At even higher frequency, tweets in the run period with negative sentiment translate into immediate stock market losses. These high frequency effects are stronger when tweets are authored by members of the Twitter startup community (who are likely depositors) and contain keywords related to contagion. These results are consistent with depositors using Twitter to communicate in real time during the bank run.
That is from a new paper by J. Anthony Cookson, et.al. Via the excellent Kevin Lewis.
Brian Slesinsky on AI taxes (from my email)
My preferred AI tax would be a small tax on language model API calls, somewhat like a Tobin tax on currency transactions. This would discourage running language models in a loop or allowing them to “think” while idle.
For now, we mostly use large language models under human supervision, such as with AI chat. This is relatively safe because the AI is frozen most of the time [1]. It means you get as much time as you like to think about your next move, and the AI doesn’t get the same advantage. If you don’t like what the AI is saying, you can simply close the chat and walk away.
Under such conditions, a sorcerer’s apprentice shouldn’t be able to start anything they can’t stop. But many people are experimenting with running AI in fully automatic mode and that seems much more dangerous. It’s not yet as dangerous as experimenting with computer viruses, but that could change.
Such a tax doesn’t seem necessary today because the best language models are very expensive [2]. But making and implementing tax policy takes time, and we should be concerned about what happens when costs drop.
Another limit that would tend to discourage dangerous experiments would be a minimum reaction time. Today, language models are slow. It reminds me of using a dial-up modem in the old days. But we should be concerned about what happens when AI’s start reacting to events much quicker than people.
Different language models quickly reacting to each other in a marketplace or forum could cause cascading effects, similar to a “flash crash” in a financial market. On social networks, it’s already the case that volume is far higher than we can keep up with. But it could get worse when conversations between AI’s start running at superhuman speeds.
Financial markets don’t have limits on reaction time, but there are trading hours and circuit breakers that give investors time to think about what’s happening in unusual situations. Social networks sometimes have rate limits too, but limiting latency at the language model API seems more comprehensive.
Limits on transaction costs and latency won’t make AI safe, but they should reduce some risks better than attempting to keep AI’s from getting smarter. Machine intelligence isn’t defined well enough to regulate. There are many benchmarks and it seems unlikely that researchers will agree on a one-dimensional measurement, like IQ in humans.
[1] https://skybrian.substack.com/p/ai-chats-are-turn-based-games
[2] Each API call to GPT4 costs several cents, depending on how much input you give it.
Running a smaller language model on your own computer is cheaper, but they are lower quality, and it has opportunity costs since it keeps the computer busy.
Fertility and mimetic desire
There is a new and excellent NYT column by Peter Coy on this topic, here is one excerpt:
There is a town in western Japan named Nagi that’s famous for making babies. Its fertility rate in 2021 was 2.68 lifetime births per woman, compared with 1.3 for Japan as a whole, according to an article in The Wall Street Journal that my Opinion colleague Jessica Grose recently cited. Delegations from elsewhere in Japan and abroad have come to Nagi to learn its secret formula. Is it the free medical care for all children? The affordable child care? The cash gifts to new mothers?
I’ve been considering another theory. Maybe people in Nagi are having babies because other people in Nagi are having babies.
And:
I read several papers on peer effects on fertility with Angrist’s caveats in mind. One, by Jason Fletcher and Olga Yakusheva, looked at American teenagers and found that a 10 percentage point increase in pregnancies of classmates is associated with a 2 to 5 percentage point greater likelihood of a teenager herself becoming pregnant.
The complications behind such inferences are considered in detail.
Saturday assorted links
1. “Schumacher family planning legal action over AI ‘interview’ with F1 great.”
2. More Brian Potter on how solar power got cheap.
3. “In a study to be published this summer, they find that the median ai expert gave a 3.9% chance to an existential catastrophe (where fewer than 5,000 humans survive) owing to ai by 2100. The median superforecaster, by contrast, gave a chance of 0.38%.” From The Economist.
4. New charter city for Nigeria? (p.s. silly header on the article). And some CCI corrections to the piece.
The link between economic concentration and political power?
Our findings do not support the political antitrust movement’s central hypothesis that there is an association between economic concentration and the concentration of lobbying power. We do not find a strong relationship between economic concentration and the concentration of lobbying expenditure at the industry level. Nor do we find a significant difference between top firms’ and other firms’ allocation of additional revenues to lobbying. And we find no evidence that increasing economic concentration has appreciably restricted the ability of smaller players to seek political influence through lobbying. Ultimately, our findings show that the political antitrust movement’s claims do not rest on a solid empirical foundation in the lobbying context. Our findings do not allay all concerns about transformation of economic power into political power, but they show that such transformation is not straightforward, and they counsel caution about reshaping antitrust law in the name of protecting democracy.
Here is the recent paper by Sepehr Shahshahani and Nolan McCarthy. Via the excellent Kevin Lewis. And yes, yes I know there is much more here than just lobbying expenditures, but that it doesn’t show up in that area…isn’t supportive.
AI and economic liability
I’ve seen a number of calls lately to place significant liability on the major LLM models and their corporate owners, and so I cover that topic in my latest Bloomberg column. There are numerous complications, and I cover a mere smidgen of them, but still more analytics are needed here. Excerpt:
Imagine a bank robbery that is organized through emails and texts. Would the email providers or phone manufacturers be held responsible? Of course not. Any punishment or penalties would be meted out to the criminals…
In the case of the bank robbery, the providers of the communications medium or general-purpose technology (i.e., the email account or mobile device) are not the lowest-cost avoiders and have no control over the harm. And since general-purpose technologies — such as mobile devices or, more to the point, AI large language models — have so many practical uses, the law shouldn’t discourage their production with an additional liability burden.
Of course there are many more complications, and I am not saying zero corporate liability is always correct. But we do need to start with the analytics, and a simple fear of AI-related consequences does settle the matter. There is this:
On a more practical level, liability assignment to the AI service just isn’t going to work in a lot of areas. The US legal system, even when functioning well, is not always able to determine which information is sufficiently harmful. A lot of good and productive information — such as teaching people how to generate and manipulate energy — can also be used for bad purposes.
Placing full liability on AI providers for all their different kinds of output, and the consequences of those outputs, would probably bankrupt them. Current LLMs can produce a near-infinite variety of content across many languages, including coding and mathematics. If bankruptcy is indeed the goal, it would be better for proponents of greater liability to say so.
Here is a case where partial corporate liability may well make sense:
It could be that there is a simple fix to LLMs that will prevent them from generating some kinds of harmful information, in which case partial or joint liability might make sense to induce the additional safety. If we decide to go this route, we should adopt a much more positive attitude toward AI — the goal, and the language, should be more about supporting AI than regulating it or slowing it down. In this scenario, the companies might even voluntarily adopt the beneficial fixes to their output, to improve their market position and protect against further regulatory reprisals.
Again, not the final answers but I am imploring people to explore the real analytics on these questions.
Friday assorted links
1. New YouTube channel with interviews of very good (but typically neglected) economists, run by a high school student, Kennedy Owen.
3. People are more dishonest against big business than small business.
4. Anti-malaria vaccine now approved in both Nigeria and Ghana.
5. A Tulsa story.
6. AI proposals before Congress.
7. Devin Helton.
The pro-immigration argument that everyone hates
Fortunately people hate it because it is wrong, otherwise they would have to hate it for less intellectually honest reasons. The basic context of course is that native rates of fertility are in irreversible decline. Here goes:
Immigration is not going so well today in terms of assimilation. Yet in the future it will go worse yet, because the native-borns will be smaller in number and also older and less energetic. Nonetheless we need to take in a lot of immigrants today, as a kind of practice, so we can get used to the much greater number of immigrants we will need to take in a generation or two from now. It is better to be a crummy country than a country of 33,000 people. And so we must become crummier now, so that later on our rise in crumminess is modestly tempered, though it still will happen. Open the gates!
See?
Questions pondered by Gwern
Here is one:
Why does writing in the morning (anecdotally so far) seem to be so effective for writers, even ones who are not morning persons? While programmers, which seems like a similar occupation, are invariably owls?
Here is a much longer list.
Using AI in politics
Could AI be used to generate strategic advantage in politics and elections?
Without doubt. We used it to improve prediction of the true critical voters in 2016 (but not to improve the execution of digital marketing, per the Cadwalladr conspiracy) and the true critical voters and true marginal seats in 2019. Competent campaigns everywhere could already, pre-GPT, use AI tools to improve performance.
We did some simple experiments last year to see if you could run ‘synthetic’ focus groups and ‘synthetic’ polls inside a LLM. Yes you can. We interrogated synthetic swing voters and synthetic MAGA fans on, for example, Trump running again. Responses are indistinguishable from real people as you might expect. And polling experiments similarly produced results very close to actual polls. Some academic papers have been published showing similar ideas to what we experimented with. There is no doubt that a competent team could use these emerging tools to improve tools for politics and perform existing tasks faster and cheaper. And one can already see this starting (look at who David Shor is hiring).
It’s a sign of how fast AI is moving that this idea was new last summer (I first heard it discussed among top people roughly July), we and others tested it, and focus has moved to new ideas without ~100% of those in mainstream politics today having any idea these possibilities exist.
“Almost space” markets in everything
The space race just got a new entrant. France’s Zephalto is offering passengers the chance to travel to the stratosphere in a balloon, starting at €120,000 ($132,000) per person in 2025.
“I partnered with the French space agency, and we worked on the concept of the balloon together,” says Zephalto founder and aerospace engineer Vincent Farret d’Astiès.
He tells Bloomberg that he’s planning on 60 flights a year, with just six passengers on board each flight. The company aims to provide an experience that brings the best bits of French hospitality—fine food, wine and design—to the edges of space for those who can afford the six-figure ticket.
Balloons filled with helium or hydrogen will depart from France with two pilots on board and rise 25 kilometers (15.5 miles) into the stratosphere for 1 1/2 hours. Once at peak altitude, which is about three times higher than for a commercial airliner, the balloon will stay for three hours, giving guests a chance to take in views previously seen only by astronauts. The descent will take a further hour and a half, for a six-hour round trip.
Here is more from Sarah Rappaport at Bloomberg. Via Daniel Lippman.
Thursday assorted links
1. How much are people spending on dates.
2. Why not buy an abandoned Japanese house? The price is right (NYT). Soon they may be cheaper than repeated dating.
3. At the local level, employment concentration is falling.
4. “Who is crazier? Me or them?” (Ukraine issues)
5. “We’re Not Going to Die,” Robin Hanson video. And ask the experts: good common sense from Tom Tugendhat, UK security minister, on AI safety. “China, with its vast datasets and fierce determination, is a strong rival.” Keep in mind we need to stay ahead of them for a long while, not just a few years. The fact that the Chinese might heavily regulate their private sector AI tells you nothing about what their government will do, or if anything it tells you they will emphasize developments in the military direction.
A Mosquito Factory?!
A “mosquito factory” might sound like the last thing you’d ever want, but Brazil is constructing a facility capable of producing five billion mosquitoes annually. The twist? The factory will breed mosquitoes carrying a special bacteria that significantly reduces their ability to transmit viruses. As far as I can tell, however, the new mosquitoes still suck your blood.
Nature: The bacterium Wolbachia pipientis naturally infects about half of all insect species. Aedes aegypti mosquitoes, which transmit dengue, Zika, chikungunya and other viruses, don’t normally carry the bacterium, however. O’Neill and his colleagues developed the WMP mosquitoes after discovering that A. aegypti infected with Wolbachia are much less likely to spread disease. The bacterium outcompetes the viruses that the insect is carrying.
When the modified mosquitoes are released into areas infested with wild A. aegypti, they slowly spread the bacteria to the wild mosquito population.
Several studies have demonstrated the insects’ success. The most comprehensive one, a randomized, controlled trial in Yogyakarta, Indonesia, showed that the technology could reduce the incidence of dengue by 77%1, and was met with enthusiasm by epidemiologists.
In Brazil, where the modified mosquitoes have so far been tested in five cities, results have been more modest. In Niterói, the intervention was associated with a 69% decrease of dengue cases2. In Rio de Janeiro, the reduction was 38%3.
Wolbachia-infected mosquitoes have already been approved by Brazilian regulatory agencies. But the technology has not yet been officially endorsed by the World Health Organization (WHO), which could be an obstacle to its use in other countries. The WHO’s Vector Control Advisory Group has been evaluating the modified mosquitoes, and a discussion about the technology is on the agenda for the group’s next meeting later this month.
Do older economists write differently?
The scholarly impact of academic research matters for academic promotions, influence, relevance to public policy, and others. Focusing on writing style in top-level professional journals, we examine how it changes with age, and how stylistic differences and age affect impact. As top-level scholars age, their writing style increasingly differs from others’. The impact (measured by citations) of each contribution decreases, due to the direct effect of age and the much smaller indirect effects through style. Non-native English-speakers write in different styles from others, in ways that reduce the impact of their research. Nobel laureates’ scholarly writing evinces less certainty about the conclusions of their research than that of other highly productive scholars.
Here is the full NBER paper by Lea-Rachel and Daniel S. Hamermesh.