The Hidden Risks of AI: How Linguistic Diversity Can Make or Break Collective Intelligence

Posted on by
Reply

Diversity is a key ingredient in the recipe for collective intelligence because it brings together a range of perspectives, tools, and abilities; allowing for a more comprehensive approach to problem-solving and decision-making. Gender diversity on corporate boards improves firms’ performance, ethnic diversity produces more impactful scientific research, diverse groups are better at solving crimes, popular juries are less biased than professional judges, and politically diverse editorial teams produce higher-quality Wikipedia articles.

Large language models, like those powering AI systems, rely heavily on datasets or corpora, with a significant part of it based on English content. This dominance is consequential. Just as diverse groups of people yield richer outcomes, an AI trained on diverse linguistic data offers a broader perspective. Each language encapsulates unique thoughts, metaphors, and wisdom. Without diverse linguistic representation, we risk fostering AI systems with limited collective intelligence. The quality, diversity, and quantity of the data they are trained on directly influence their epistemic outputs. Unsurprisingly, large language models struggle to capture long-tail knowledge.

This comes with two major — at least hypothetically — risks: 1) systems that do not fully leverage the knowledge dispersed in the population, 2) the benefits of AI may be more accessible to some groups over others; for instance, speakers of less-dominant languages might not equally benefit from AI’s advancements. It’s not merely about translation; it’s the nuances and knowledge embedded in languages that might be overlooked.

There are also two additional dimensions that could reinforce biases in AI systems: 1) as future models are trained on content that might have been generated by AI, there may be a reinforcing effect where biases present in the initial training data are amplified over time; and 2) techniques such as guided transfer learning may also increase biases if the source model used in transfer learning is trained on biased data.

This introduces a nuanced dimension to the digital divide. Historically, the digital divide was characterized by access to technology, internet connectivity, digital skills, and the socio-economic variables shaping these factors. Yet, with AI, our understanding of what constitutes digital divide should expand. It’s a subtler yet crucial divide that policymakers and development practitioners might not yet fully recognize.

Generous Listening Symposium

Posted on by
Reply

This week, I am helping to lead a research symposium on “Generous Listening in Organizations” in partnership with the Vuslat Foundation. The main topic is listening to colleagues, including supervisors and employees. This practice is important for society and it matters to me, as a long-time middle-manager in a large organization (a university).

I also come into this conversation as someone who studies politics, broadly defined, and who believes that a better understanding of listening may create fruitful openings for strengthening democracy.

There is a vast literature on political communication, speech and rhetoric, deliberation, debate, and the public sphere–defined as the institutions in which people express their views. This topic is important because we do not automatically hold opinions, even about our own circumstances. We usually obtain our views through communication, which can go either well or badly. We can be persuaded to sacrifice for the common good or to become murderous racists. We can be persuaded that we are citizens of the world or members of narrow communities. Political action then follows communication.

Listening is one side of this exchange. It is certainly not absent in the literature but is less often discussed than the other side: speaking.

The research on political communication is diverse and nuanced, but quite a lot of it is critical. Survey-based research in the United States–and in many other countries–often finds that people are ill-informed, biased, and incoherent. I semi-facetiously summarize the overall message of current political science as: “People are stupid and they hate each other.”

The specific findings are often valid and worth consideration, but the overall message may hamper efforts to improve civic life. The message is most discouraging when the specific findings are linked to general claims about human beings, e.g., that we are naturally self-interested or that we evolved to use heuristics helpful in small groups of hunter-gatherers that fail to equip us for responsible self-government at a mass scale. If human beings have deep psychological limitations for participating in democracy, then perhaps we had better shore up our most basic safeguards (especially the peaceful transition of power after each election), and not be distracted by more ambitious democratic ideals.

I start with a different assumption. I presume that we exchange ideas in artificially designed settings that can help us to be wise or foolish: assembly halls, churches, newspapers, classrooms, laboratories, online networks, and many more. Designing and expanding good settings requires a degree of optimism about human potential. Therefore, research that implies we are hard-wired to be foolish can discourage people from working to build better institutions. And when our institutions are weak, we tend to think and behave in troubling ways that research then reveals, thereby reinforcing the researchers’ skepticism–a classic vicious cycle.

A focus on listening might help break the cycle. The evidence is pretty strong that when we form and state opinions, we are not as wise as we believe. We offer reasons for what we think and value that sound good to us–they sound like explanations of our views. But often, we have formed our views intuitively and then merely rationalized them in speech. Furthermore, our intuitions are unreliable, because they often reflect cognitive biases, selfishness, and limited empathy.

However, there is also evidence that we can be pretty good at listening. We can assess the reliability and competence of speakers and the cogency of their claims. In turn, our assessments of others’ statements can shift our intuitions. Indeed, Mercier & Sperber (2017) argue that we evolved to do this–to scan our human environment for people whose views are worthy of trust.

But then the questions include: How can we listen well? And how can we design institutions to enable and reward good listening?

Reference: Mercier, Hugo and Dan Sperber, The Enigma of Reason, Cambridge, MA: Harvard University Press 2017. See also: how intuitions relate to reasons: a social approach; how the structure of ideas affects a conversation; An agenda for R&D for democracy etc.

Mani Rao’s innovative Sanskrit translations

Posted on by
Reply

How can you translate writing that’s densely allusive if you expect few of your readers to recognize the allusions? How can you translate poetry–or any other formally complex writing–into a totally different language while conveying some of the experience of the original form? And how can you translate passages from a language that has one kind of grammatical structure into an entirely different one?

One option is to translate loosely into a poem that works in your own language, sacrificing accuracy for experience. An alternative is to translate into literal prose and provide many footnotes or other explanations. That gives your readers some of the information they need, but it offers a very different experience from reading the original poem.

For instance, consider these two translations of the first stanza of a Sanskrit classic, The Meghaduta or Cloud Messenger by Kalidasa. First, H.H. Wilson (1786-1860) offers a loose translation into Victorian rhymed couplets. 

Spoiled from his glories, severed from his wife,
A banished Yaksha passed his lonely life:
Doomed, by his lord's stern sentence to sustain,
Twelve tedious months of solitude and pain.
To these drear hills, through circling days confined,
In dull, unvaried grief the god repined...

In contrast, E.H. Rick Jarrow (2021) translates “scrupulously and thoroughly,” avoiding constructions that would sound “choppy” in English but striving to convey each name and idea from the original text

A Yaksha, banished in grievous exile from his beloved for a year,
his power eclipsed by the curse of his Lord for having swerved from his duty,
made his dwelling among the hermitages of Ramagiri
whose waters were hallowed by the ablutions of Janaka's daughter
and whose trees were rich with shade.

“Drear hills” may convey some of the mood of the poem but gives us no sense of the importance of the particular hills where the Yaksha has to live. Jarrow’s “The hermitages of Ramagiri whose waters were hallowed by the ablutions of Janaka’s daughter” incorporates more information but sends us to notes if we want to make sense of the references. Both translations are poems, but Wilson’s is rhymed and metered in a way that is typical of Victorian verse and perhaps honors the formal regularity of the Sanskrit.

Both translations have merit, but I am excited by the innovative approach of Mani Rao (2014). Here is how she renders the same first stanza:

Some yaksha who made a mistake was cursed by his master: 
Suffer! 
One entire year 

   An ordinary yaksha 
   Not a hero 

   When even a season’s separation’s unbearable 
   Imagine six 

   What mistake 
   Kalidasa does not specify 
   Some lapse of duty 
   Same word for ‘duty’ and ‘right’ 

   Has the hero lost the reader’s heart
    In the very first line? 

Heavy the pangs of separation from his beloved 
His prowess gone like a sun that’s set 
   Year-long night 

He lived in hermitages on a mountain 
named after Rama 

Groves cool, waters pure 
Sita once bathed here 

    Remember Rama remembered Sita 
    Remember messenger Hanuman 
    Flying like a cloud 
    Why hermitages, in the plural? 
    More than exiled. Unsettled.

The plain text is her translation of the original poem into free English verse. The italicized text is her own commentary, also in verse. It’s as if we were listening to someone–or perhaps a chorus–recite the poem while another person interjected comments and questions from the side.

Rao’s additions are not exhaustively explanatory. For instance, she does not retell the story of Rama remembering the absent Sita, which is an allusion at the beginning of the Meghaduta. I do not know that story–but I can look it up once someone mentions the names. The explanation, being a poem itself, is marked by explicit emotion, irony, misdirection, and other literary features.

Rao’s approach may have precedents. (I would not necessarily know.) I have appreciated how Anne Carson supplies prefaces to her Greek translations that are themselves poems. For instance:

                      i wish i were two dogs then 
                             i could play with me 
         (translator’s note on euripides’ bakkhai) 

Dionysos is god 
of the beginning 
before the beginning.

What makes 
beginnings special? 
Think of 

your first sip of wine
from a really good bottle.
[...]

I see great potential in this general approach of using verse to convey context for verse.

Sources: The Megha d?ta or Cloud Messenger: A Poem in the Sanskrit Language
by K?lid?sa, translated by Horace Hayman Wilson, revised edition (London: R. Watts, 1843). E. H. Rick Jarow, The Cloud of Longing: A New Translation and Eco-Aesthetic Study of Kalidasa’s Meghaduta (Oxford University Press 2021); Mani Rao, Kalidasa for the 21st Century (New Delhi: Aleph Book Company, 2014); Anne Carson, Euripides: Bakkhai (New Directions, 2017). See also: The Kural; translations from Kuruntokai; there are tears of things.

NEW Florida Citizen Resources for 2023 Civics and Government Benchmarks

Posted on by
Reply

Good afternoon, friends! It has been awhile, for sure. Today’s post is to update you on resources for Florida’s new civics and government benchmarks. Good news! We have started the process of uploading all the 7th grade middle school civics lessons to Florida Citizen and expect that most if not all will be up by mid-week next week. As a reminder, you do need to register for Florida Citizen to access the lesson plans! Please email me if you have issues with registration or logging in.

Once you log in, click on ‘Resources’, then select ‘School Resources’.

Once on that page, scroll down.

You will see three relevant lesson plan links. ‘2023-2024 Grade 7 Benchmark Resources’ will take you to the newest lesson plans. ‘2023-2024 Grade 6 and 8 Benchmark Resources’ will take you to lessons for the civics and government benchmarks in middle school US history and world history. (PRE-2023) will obviously take you to what we have done previously.

Click on the ‘Grade 7 Benchmark Resources’ link. Below is what you will see. The first important link is to the new benchmarks and their clarifications. IF the Test Item Specifications are released, we will add them as well. Keep scrolling down.

You will notice a list of the new benchmarks. Click on the one you want. Please note that right now, lessons plans for every benchmark between SS.7.CG.1.1 through SS.7.CG.3.14, and then 4.1 and 4.2, are up, though that could change by the time you read this!

When you click on the one you want, you should see the following.

This is obviously far less than what we have on our old pages. Please note that we will be adding additional materials as we can. We are currently revising practice items and have new Dr. Fine related teacher content materials as well.

You have three options for these materials. As always, we have them in Word (so you can edit and modify!), PDF, and the new option of GoogleDocs. Let’s assume that you want the word version. Click on it. It will download a zip file.

Unzip the file, and you will see ALL lesson materials for that benchmark.

The procedure is the same for middle school US history and world history materials. Please note at this time ONLY THE WORLD HISTORY LESSONS ARE AVAILABLE!!!

For the K-5 Benchmarks, lesson plans for every civics and government benchmark through Grade 4, and Grade 5 2.1 through 2.6 are now available.

Please also be aware that we have not yet updated Civics360, though current materials on that site should be adaptable or useable until we can. We are targeting a mid-October relaunch, and REGISTRATION IS NO LONGER REQUIRED.

a presidential election with two incumbents?

Posted on by
Reply

Incumbent presidents have a substantial advantage for reelection. Only 10 have lost. Statistical models that attempt to control for other factors, such as the economy, typically give a presidential candidate a bonus of 4-8 percentage points just for being the incumbent. In 2009, David Mayhew noted that parties had won the White House only half the time when they didn’t run an incumbent for reelection, but two thirds of the time when they did. That disparity is a little smaller now that Trump has lost a reelection race, but it’s still substantial.

Incumbents also have a vast advantage in their own party’s nomination race, with only Franklin Pierce actually losing his party’s support while he was in the White House. (Several others have chosen not to run.)

However, it’s not clear what these patterns mean for 2024. After all, two leading candidates have recently been presidents of the United States. Whether one, both, or neither will get an incumbency advantage is hard to tell from the historical data. There have only been 59 presidential elections ever, and only 23 in the modern era since FDR beat Hoover. Many variables are relevant to the outcomes of these cases. It is therefore hard to detect which aspects of incumbency may matter–and certainly hard to extrapolate any patterns to our unprecedented situation.

(I think that only one president was unseated but got his original party’s nomination again in a subsequent year: that was Grover Cleveland, who won the rematch.)

Mayhew identified 12 possible explanations for the presidential incumbency advantage and collected the evidence that might bear on each explanation. See his table 3.

Possible explanations of presidential incumbency bias from Mayhew 2008.

Some of these factors presume that the incumbent typically has impressive “capabilities,” because a president has already managed to win a national election and has governed for up to four years. The challenger could be equally capable, but the odds are against it.

In this case, I personally think that Joe Biden has demonstrated strong capabilities, whereas Trump’s campaigns and his time in office demonstrate a blatant lack thereof. However, it doesn’t matter what I think, but what the electorate decides. According to today’s New York Times/Siena poll, 44 percent of Americans agree with me–they hold a very unfavorable view of Trump–but 42 percent strongly disapprove of Biden’s performance in office. What is unusual is the fact that both likely nominees have recently served as the president, which may make this factor a wash.

Voters might also simply prefer a candidate who already holds office. For instance, perhaps some voters are risk-averse: biased against changing horses. Or perhaps they think that any newcomer would need too much time to become effective. Risk-aversion might help Biden a bit in 2024, although some voters may feel that electing him in 2020 was the unwise change. Voters can presumably see either Trump or Biden as the status quo if they want to. Some Trump voters even believe that he is the rightful winner of the most recent election.

“Incumbent party fatigue” refers to the pattern that each party loses a small amount of support–on average–each year that it holds the White House. However, this trend does not doom incumbents, in part because too little “fatigue” typically occurs within four years. The candidates who tend to suffer are those who try to succeed a president of the same party.

The explanations that Mayhew calls “strategic behavior” are of the following type. Perhaps the party of an incumbent president typically organizes to renominate him if he’s been doing well but moves against him if not–as was the case with LBJ in 1968. That would imply that incumbents who run for reelection have generally had successful first terms. And perhaps the opposition party is easily captured by marginal figures when its prospects already look weak against a popular incumbent (consider Barry Goldwater in 1964, or Michael Foot against Margaret Thatcher in 1983). However, Mayhew does not see a lot of power in this type of explanation.

Somewhat buried in his list are factors under the incumbent president’s direct control, such as the ability to allocate funds (e.g., to decide whether to locate the Space Command in Colorado or Alabama), where to deploy cabinet members, which bills to sign or veto, etc. Early in 2021, I predicted that the “$1.9 trillion American Rescue Plan Act of 2021 would prove popular by putting cash directly and quickly into people’s pockets,” and this would help Democrats in 2022. Democrats did somewhat better than most people expected that year, but there’s no evidence that nearly $2 trillion helped them electorally. (In other words, I was wrong.) The other side of the coin is that presidents must make difficult and unpopular decisions.

Overall, I think that Biden has an advantage in 2024 if the economy avoids a deep recession. I think a substantial Biden victory is possible if the economy starts to lift. I suspect that the 4-8 point incumbency advantage–which would typically give Biden a landslide–will not apply, because the electorate will split on the question of who is the rightful incumbent in the first place.

*Mayhew, David R. “Incumbency Advantage in U.S. Presidential Elections: The Historical Record.” Political Science Quarterly, vol. 123, no. 2, 2008, pp. 201–28. See also: 1984 all over again? The Reagan/Biden analogy; What kind of a claim is “Biden has an 87% chance of winning”? (on the metaphysics of probability)

highways on the sea: from Machado to Paolo Freire

Posted on by
Reply

This is a perfect short poem from Antonio Machado’s Proverbios y Cantares (1912):

     XLIV

Todo pasa y todo queda:
pero lo nuestro es pasar,
pasar haciendo caminos,
caminos sobre la mar

Everything passes and everything stays,
But ours is to cease to be. 
We make a highway as we pass,
A highway on the sea. 

Machado had already juxtaposed caminos (roads, paths, journeys) with el mar (the sea) in the second poem of the volume:

     II

¿Para qué llamar caminos 
a los surcos del azar?... 
Todo el que camina anda, 
como Jesús, sobre el mar. 

Why designate as highways
furrows left aimlessly? ...
Anything that travels moves,
like Jesus, on the sea.

The same pairing recurs in the most-quoted lyric of the whole book:

     XXIX

Caminante, son tus huellas 
el camino y nada más; 
caminante, no hay camino, 
se hace camino al andar. 
Al andar se hace camino, 
y al volver la vista atrás 
se ve la senda que nunca 
se ha de volver a pisar. 
Caminante, no hay camino, 
sino estellas sobre la mar.

Traveler, the highway
is your footprints, nothing more;
Traveler, there is no highway,
you make it as you walk.
As you walk, you make the highway—
and the path you see when you turn back
is the route where you'll never be.
Traveler, there is no highway,
save for stars upon the sea.

In 1987, the Brazilian educator Paolo Freire (then 66 years old) and the American organizer Myles Horton (82) interviewed each other at the Highlander Folk School in Tennessee, which Horton had led. Freire says, “Myles, I think we could start our conversation by saying something to each other about our very existence in the world.” A little later, he adds, “It’s very important for Brazilian readers to have information about Myles. About me, they have already, but about Myles they don’t have and it’s very, very important.”

Horton adds, “Yes, but the people in this country need the same thing about you.” He then proposes to talk “mainly [about] the things that would help people understand where I came from in terms of my ideas and my thinking, what they are rooted in. Is that the idea?” Freire replies, “Yes. Everything you recognize as something important. I think that even though we need to have some outline, I am sure that we make the road by walking. It has to do with this house [Highlander], with this experience here. You’re saying that in order to start, it should be necessary to start.”

The resulting book, We Make the Road by Walking (Horton & Freire 1990), explains in a footnote that Freire is adapting “a proverb by the Spanish poet Antonio Machado, in which one line reads “se hace camino al andar,” or “you make the way as you go.”

For activists, this phrase suggests that people can make new pathways by taking action, and perhaps that we should learn the trails that our elders have left for us. But I think Machado’s original point was apolitical. He meant that the stories we tell about ourselves are not permanent–or even important–and they vanish as we pass through them.

Translations by Peter Levine. Photo by Candie Carawan in Horton & Freire, We Make the Road by Walking (Temple University Press 1990). See also: Machado: Glory is never what I’ve sought; Lorca’s rivers

Is the USA the favorite destination for migrants?

Posted on by
Reply

When a conversation in this country turns critical about the USA, someone may interject that foreigners want to move here more than anywhere else. I have heard that point made quite a few times over the years.

It is not exactly false. The Gallup World Poll purports to be a representative global sample. It asks whether you would like to move to another country and, if so, which one you would most prefer. The USA is the most popular destination, at 21%, followed by Canada and Germany, at 6%, and France and Australia, at 5%.

But there are other ways to slice the data. One is to adjust for population size. The USA is almost 13 times as populous as Australia, but only 4.2 times as popular as the first choice for potential migrants. If you imagine that migrants could allocate themselves to the countries of their choice, then Australia would draw many more immigrants per capita than the USA. In fact, migration involves many factors apart from the potential migrants’ preferences, but 30% of Australian residents are actually foreign-born, versus 13.6 of US residents.

Likewise, 3.5 times more potential migrants would go to the USA than to Canada, but the US is 8.7 times more populous than Canada. And in reality, 23 percent of Canadians are foreign-born, versus 13.6 percent of US residents.

Another way to analyze the data is to combine countries. Gallup does not provide statistics on the popularity of smaller European countries, such as Belgium, whose actual immigrant population is 17.2% (i.e., higher than that of the USA). But if you combine the EU countries that Gallup lists, they total 20%, just shy of the USA’s 21%. I imagine that adding the other EU countries would make Europe more popular than the United States in aggregate. To be sure, some European migration is among the EU countries, but why shouldn’t that count?

Another interesting tidbit is that 10% of US and Canadian residents (combined) would prefer to migrate away from their countries. That is actually a higher rate than in Asia.

People who consider migrating tend to want to move to places where there is capital, demand for labor, and no war. Some countries that meet these standards rank unusually low as the preferred destinations for migrants, Japan being the most prominent example. But most of America’s counterparts would attract at least as many immigrants on a per-capita basis as we do, and many actually have more foreign-born residents.

Meanwhile, the countries that draw the most migrants are mostly not democratic, social-democratic, or liberal. Not counting micro-states, six of the eight sovereign nations that have the highest proportions of migrants are monarchies in the Persian Gulf (the other two being Luxembourg and Singapore). They are not necessarily preferred destinations–they register on the global poll but fall behind NATO members and Australia–but they do meet the criteria of capital and domestic peace, and they let people in.

See also: American exceptionalism; American exceptionalism, revisited; and anxieties about American exceptionalism

Reformation propaganda (note #2 from the Levine library)

Posted on by
Reply

This post is one of a series about books that my father left to me and that now line my office shelves at Tufts. More on how that happened is here.

In a folio volume of almost 2,000 dense pages, informally known as Foxe’s Book of Martyrs, John Foxe describes the persecutions of true-believing Christians since Roman times. Most of the book is devoted to its own period, when the persecutors were Catholics. This narrative helped create a sense of Protestantism as distinct from Roman Catholicism, and of England as a Protestant nation. It influenced and supplied some illustrations for my book The Anachronist.

Foxe’s Booke was a massive undertaking. It required the vast collaborative labor of collecting the names and stories of tortured and executed Protestants from across Europe. Foxe published numerous editions as new names arrived, and many copies were sold.

I have the 1576 edition, a great block of Gothic text with numerous engravings, most of which depict Protestants being tried or killed by Catholics. Here, for example, is the image on p. 1468, which shows “The talk between M[r] Bradford, and two Spanish Friers.”

John Bradford (1510–1555) was a Protestant clergyman who was burned by the government of Mary Tudor. The image shows him in his cell in the Tower of London, being interrogated by Spaniards. Queen Mary–“Bloody Mary” to Protestants–had married King Phillip II of Spain and brought England back into the Roman Church. However, by the time Foxe published an edition of his Booke in England, the Protestant Queen Elizabeth was “our gracious Lady now reigning,” and Spain was the hated enemy. This is an image of foreign treachery as well as Roman Catholic intolerance.

I thought of it last spring when I saw a painting in the Carthusian monastery of Granada, Spain. Painted in the early 1600s, it shows Thomas Cromwell condemning four Catholic clergy to death for their faith. He is likely sitting in an imagined Tower of London. That is where John Bradford met the Spanish friars under Queen Mary and where Cromwell had his own head chopped off (meriting a heroic account in Foxe’s book).

Juan Sánchez Cotán, Historia de los mártires de Inglaterra. Tres priores y un monje de Santa Brígida juzgados por C[r]onwel, Granada, Spain, via villadeorgaz.es

This painting hangs in the refectory, where the monks would dine, as part of a series entitled “The History of the Martyrs of England” by Juan Sánchez Cotán. Its didactic purpose was to remind the Spanish friars of Granada that Protestant Englishmen were their persecutors and foes.

The two images have some iconographic and stylistic similarities, although the “Gaoler” in the English print looks Mannerist (with his elongated body), and the Spanish painting is baroque.

Most of the specific stories that both sides collected were probably true. Each side interpreted the persecutions of their own co-religionists as clear evidence that their enemy was cruel. There is perhaps a lesson here about selection bias …

See also Coryat’s Crudities (note #1 from the Levine library); twenty-five thousand books to Bosnia; and my father’s books are going to James Madison’s desk at Montpelier.

1984 all over again? The Reagan/Biden analogy

Posted on by
Reply

The Biden team is preparing for a Reagan-style victory lap in the event that the economy looks healthy in 2024. I hope they are also prepping for more difficult circumstances, since the economic forecast is highly uncertain. However, for what it’s worth, Morgan Stanley “now projects 1.9% GDP growth for the first half of this year.”

These are my prior assumptions:

  1. A president has limited and ambiguous impact on macroeconomic trends. Too many other factors matter, from the Federal Reserve and Congress to the business cycle, and from wars and technological developments to the budgets of 50 states.
  2. Although many people vote on other grounds, enough marginal voters are influenced by recent changes in the economy that those trends predict the results of presidential elections.
  3. People cite the economic and electoral fortunes of each presidential administration to support their ideological positions. For instance, the apparent successes of FDR and Reagan were used to vindicate New Deal liberalism and neoliberalism, respectively.
  4. Economic trends influence public opinion, but they do not determine the outcome of the debate about ideology. Eisenhower, Clinton, and Obama are examples of presidents who saw healthy net economic growth and were reelected, yet their administrations did not move public opinion as FDR’s and Reagan’s did. The big shifts in opinion seem to reflect changes in the Zeitgeist, not just electoral and economic developments.

Applying #1 means that Biden’s policies will not affect the economy much in 2024, even though Morgan Stanley believes that he is responsible for the upturn. According to #2, Biden will win pretty easily if growth is robust but could lose to Trump or another Republican if it stalls.

Per #3, if the economy grows and Biden wins, the President and many Democrats will argue that the reason was his industrial policy, which is aggressive, green, and pro-equity. I favor the policy, so I will be hoping that this argument sticks, even though I believe that macroeconomic trends are out of his hands.

If the case for Bidenomics does persuade, it will be thanks to the Zeitgeist. To make that explanation a little less mystical, I would focus on two factors.

First, the previously dominant neoliberal view really is fading now, much as late-Victorian laissez-faire was faltering by 1932 and New Deal liberalism had run its course by 1980. By a certain point, established theories don’t offer plausible solutions to the problems of the day, but alternatives do. By the time FDR took office, his home state of New York and several others had already moved in the direction of the New Deal; Roosevelt took the opportunity to bring their ideas to Washington. Likewise for Reagan in 1980–and for Biden in 2020. Indeed, Trump is no neoliberal, and there will probably be no national candidate who runs on cutting taxes and spending.

Second, the demographic basis of politics has shifted. I avoid crude materialistic and class-based predictions, yet the people who have the most to gain from low taxes and light regulation are business owners and investors. They are now outweighed in the GOP (which they once controlled) by working-class voters. They are represented in the Democratic Party, but outweighed there, too, by diverse lower-income voters, public sector workers, and salaried professionals.

The tectonic plates are shifting, and it’s at such moments that presidential administrations become examples or even metaphors for fundamental change. We had a “new deal for America” and then saw “morning in America” once the actual New Deal state faltered. The opportunity to make another such shift is a good reason to run a celebratory campaign in 2024–if (but only if) the economy holds up.

See also: federal spending for both climate and democracy