The latest AI-writing scandal arrived, fittingly, with the Pope. On May 25, 2026, the Vatican published Pope Leo XIV’s first encyclical, Magnifica Humanitas, dated May 15, on safeguarding the human person in the age of artificial intelligence. Within days, people were feeding it to Pangram, the detector that has suddenly become the respectable instrument of literary and academic suspicion. Some portions were apparently flagged as AI-assisted, and the accusation was easy to understand: the Pope, or at least the Vatican, had used AI to write about AI.

My reaction was basically: OK, and then what? Encyclicals already pass through institutional drafting, staff work, consultation, revision, translation, and committee-like smoothing. If some Vatican official used Claude to turn Pope Leo’s ideas into prose, the relevant questions would still be whether the document is accurate, thoughtful, and worth reading. But I should admit I was wrong.

Subscribe now

In Part II of my AI series, I wrote that AI writing detectors were bad and would probably remain bad. Pangram changed my mind: it has third-party evidence behind it, claims a very low false-positive rate, and has become the detector people reach for when they suspect undisclosed AI writing. Kelsey Piper recently wrote about Pangram Labs’ claims that several prize-winning short stories were AI-generated or substantially AI-assisted, and The Atlantic’s Matteo Wong has now written about Pangram’s growing power in schools, publishing, journalism, and the AI-writing accusation economy.

A working detector makes the ethics more urgent because it tempts people to treat provenance as a verdict. That is why the joke about using Pangram to filter out human-written content is sharper than it first sounds: in plenty of settings, AI-assisted writing may be more readable and more useful than unaided human prose. If the data underneath are sound, the human sentence-level struggle contributes little.

I should also be upfront that I am an interested party here. I am that infamous AI professor who proudly writes with the help of AI. An essay arguing that AI detection can become status policing is conveniently a defense of my own practice, so readers should ask whether I am drawing the line in a way that flatters me.

The provenance spectrum

AI-writing ethics starts with the promise the writer made. A writer can make a promise to a teacher, editor, reader, institution, or recipient. The ethical question depends on that promise before it depends on a detector score.

Let’s start where the skeptics are right. A student assignment that explicitly bans AI is the obvious case. A creative writing contest that promises to recognize new human writers is another. A condolence note belongs in a different category from an exam, but if someone grieving expects words from you, outsourcing the emotional act to a machine feels like a betrayal.

Some decisions also require accountable human judgment. If I am deciding whether someone gets a scholarship or grant, provenance matters because the applicant is owed my judgment. AI can help organize evidence or check consistency, but the evaluative act has to remain mine.

Consequences matter too. The more influential the decision, and the more it depends on personal judgment, the stronger the case for knowing who or what made the call. Human discretion can also be worse than AI discretion: a committee can be biased or arbitrary, and a well-designed AI system may eventually make some decisions more consistent.

There is also a special rule for first-person claims. When I write “I think” or “I feel,” that conviction should actually be mine. AI can help me phrase it, pressure-test it, or make it less awkward. It cannot supply the conviction itself.

Research and journalism sit closer to the middle. A byline is a promise that the author stands behind the claims, evidence, and judgment calls. It has never meant that the author personally typed every sentence without help from search engines, copy editors, co-authors, translators, or now LLMs. If my name is on an argument, the argument has to be mine; the prose can be assisted.

Much technical writing belongs closer to the content-matters side. If I ask AI to describe a chart, write a methods paragraph, or translate a regression result into normal English, the important question is whether the output is correct. I still have to verify the numbers and own the final text. Accuracy and accountability carry the moral weight.

At the far end are administrative messages where almost nobody cares about the human act of writing. If a department asks you to send a polite note confirming a committee meeting, use AI freely. The relevant standard is whether the note is true and clear.

A single AI score cannot answer the ethical question. The same level of AI assistance can be harmless in an administrative email, useful in a technical report, questionable in a personal essay, and disqualifying in a no-AI classroom assignment. Context is the whole point, even when the underlying detection is accurate.

Detection has a spectrum too

The ethics of detection should follow the ethics of use. If a teacher has told students to write without AI on a particular assignment, a detector can be part of an academic-integrity process. A Pangram score should never be the only evidence, especially given the stakes for students.

Creative contests face a similar problem. Piper’s argument about the Commonwealth Short Story Prize should be taken seriously because fiction prizes are partly about human craft. If a prize is rewarding a human writer’s voice, a fully AI-generated submission violates the premise. The organizer can allow AI, ban AI, or create a separate category. Trust alone will not settle the problem.

Peer review is harder. Seth Lazar gave the strongest version of the pro-detection case in response to my earlier Pangram post: AI-generated research output can become a denial-of-service attack on peer review. The cost of producing plausible-looking papers collapses, while the obligation to read them remains expensive. In that context, a detector may help preserve scarce review capacity.

The peer-review case still depends on the goal. If the goal is to catch students violating an explicit rule, provenance is the target. If the goal is to protect reviewers from worthless submissions, provenance is only a proxy: a detector estimates the probability that a text is AI, never the probability that it is bad. The real target is bad work: hallucinated data, fake citations, nonexistent methods, and papers with no question worth answering. A detector might help triage some of that, but someone still has to check the actual claims.

My worry is that we will police the em dashes while ignoring the hallucinated data underneath them. That would be a very academic way to lose the plot: exquisite attention to the surface marker, little attention to whether the thing says anything true.

Why disclosure mostly fails

The obvious compromise is disclosure. Let people use AI, require them to say so, and let readers decide how much it matters. That sounds attractive because it treats AI assistance as information and lowers the moral temperature.

I argued in Part II of the AI series that disclosure norms collapse under the incentives they create. The detector half of that argument now needs revision because Pangram appears to work much better than I expected. The disclosure half still looks right to me.

The more ethically questionable the AI use is, the stronger the incentive to hide it. A student who used AI after promising to write unaided, a contest entrant who submitted machine-written fiction to a human-writing prize, or a researcher who used AI to paper over fake citations has every reason to stay silent.

The people most likely to disclose are the ones using AI in low-stakes ways: cleaning up a paragraph, translating a chart, or turning rough notes into readable prose they still own. Those are also the cases where disclosure matters least. The likely equilibrium is a world full of ritual acknowledgments about harmless AI assistance, while the genuinely deceptive cases remain hidden until someone investigates them.

Disclosure can still help when the disclosure itself explains the work, as it does here. Editors, teachers, prize committees, and people exercising institutional authority should also be clear about the rules they enforce. But if the whole system depends on honest confession, it will punish the conscientious and leave the strategic users alone.

The new grammar policing

I know the phrase “grammar Nazism” is abrasive, and I mean something specific by it. I was born in the Soviet Union, and Russian elite culture can be intensely sensitive to grammar, pronunciation, stress patterns, and the small status markers embedded in speech. In practice, grammar correction often doubled as social sorting: the wrong school, the wrong region, the wrong family background, or the wrong kind of education could leak through the way you spoke.

America has its own version of this. Academic English is full of status signals masquerading as standards. The right kind of fluency makes you sound smart before anyone checks whether you are right, and the wrong accent or idiom can mark you as unserious before your argument gets a hearing.

AI detection is turning this old habit into a new technical ritual. The same people who once policed grammar now police “AI tells”: em dashes, smooth transitions, generic metaphors, oddly balanced paragraphs, prose that seems a little too clean. Sometimes they are right. AI writing does have recognizable patterns, which is why I have a style guide full of them.

If someone reads a piece, learns something new, and then makes the conversation about one suspicious phrase, the Pangrammar instinct has wasted everyone’s time. A reader’s attention should go first to the claim, the evidence, and the payoff, with style policing saved for cases where the prose actually blocks understanding or signals deception.

This status dynamic is very familiar. The detector score gives a scientific-looking license to dismiss work without reading it carefully. The people who benefit are usually the incumbent writers and credentialed gatekeepers who can turn a judgment call into a score. The accusation becomes especially convenient against lower-status writers and people who do not write English well but can now use AI to translate, draft, and reach an English-language audience. Too polished looks fake. Too awkward looks low quality. Either way, the gatekeeper wins.

The moral contamination logic makes the problem worse. Once AI involvement is treated like impurity, any trace of assistance becomes enough to condemn the whole work. That is a strange standard for a world where human writing has always been socially produced by editors, reviewers, co-authors, translators, and the sentence you read yesterday.

The funniest possible equilibrium is already here. AI tools write prose that is too clear, detectors punish the clarity, and then new “humanizer” tools rewrite the prose to look more awkward. TIME recently described people inserting mistakes and oddities to avoid sounding AI-generated. This is Grammarly in reverse: make the writing worse so it looks more authentic.

What to do instead

I am arguing for detector modesty. Pangram should only be used where provenance is part of the agreement: exams with explicit no-AI rules, contests that promise human craft, or institutional settings where the source of the text is part of the job. The institutional rule should be written before the score is consulted: define what AI use would violate the promise and what appeal process follows a high score.

In many domains, the standard should be quite simple: if you put your name on the work, you own it. You own the facts, the claims, the errors, the taste, the structure, and the judgment. If AI helped you produce an accurate technical summary, good. If AI helped you produce nonsense faster, that is on you.

Because attention is scarce, people will still rely on shortcuts. They will trust names they know, journals they respect, editors with a track record, friends who have read the work, and institutions that have something to lose if they publish junk. That is imperfect and often unfair. Outsiders and newcomers pay a price when reputation becomes the filter. But at least reputation is accountable over time. If a journal, prize, professor, or writer keeps endorsing bad work, people can notice.

A Pangram score is different. It gives a quick guess about textual provenance and invites us to stop reading before we have asked what the text is doing. Pangram seems to work, so the question is no longer whether we can detect AI. The question is what we should do with that information. Use it when provenance is part of the bargain and the stakes justify an investigation. Treat it as a prompt for judgment, never as a substitute for judgment.

If the work is fake, wrong, plagiarized, emotionally fraudulent, or a violation of a clear rule, say that and act accordingly. If the work is accurate, useful, and owned by the person whose name is on it, the fact that Codex, Claude, or ChatGPT helped assemble the sentences is a weak basis for scandal. The scandal would be building a culture where everyone learns to make writing worse so it can pass as human.

One last disclosure, since the whole piece is about this question: the essay above was written entirely in Codex from several hours of my dictated thoughts, earlier posts, saved style instructions, and recent social media exchanges. The cover image and spectrum chart were also produced by Codex. This was not a one-shot prompt. We went through more than a dozen iterations, mostly refining the argument and the chart. Yes, I am using Codex more than Claude Code now. Yes, I read the draft before publishing but did not line-edit the prose at all. By my own chart, this essay sits on the content-matters side of the spectrum, and I stand behind it.

Share

Some good news first. I am opening Popular by Design to occasional guest contributors, with formal guidelines coming soon. The first two posts were a perfect reminder of why this is worth doing: asked whether intervention would cause a migration crisis in Cuba, and explained why mainstream parties seem to lose either way when they accommodate the far right. Gil’s piece asks what a Cuba intervention debate misses when it treats migration as an afterthought. Hanno’s piece asks what mainstream parties actually gain or lose when they borrow far-right immigration positions. Both pieces say something new and practical about the problems facing our politics.

Subscribe now

May was a lighter month for my own writing after the April sprint, which was probably healthy. I still managed to publish my Washington Post op-ed on why high-skilled immigration is a winning policy that needs a champion, plus a Popular by Design essay on why LinkedIn is doing what Bluesky was supposed to do. I also recently spent two days in DC at the Teaching Progress workshop, hosted by at the JHU Bloomberg Center, where faculty reworked syllabi around the question of how to teach about progress, especially now in the era of AI. I went in with my migration and conflict courses and came out with a clearer understanding of what I need to change. After a stretch of academic conferences that felt more draining than useful, this one felt like a breath of fresh air. I actually learned something new and met interesting people in a cool setting.

Here are the May links (linking does not imply endorsement):

• argues that Trump closed the representation gap on immigration. I agree with the broad thrust that Trump’s immigration politics are more popular than many smart people want to admit. My main criticism is that the usual Gallup increase/decrease item is a shaky way to diagnose representation: it mixes what people want from immigration, what politicians in power and opposition are doing, and thermostatic reactions to current events. I would also separate Trump’s popular moves on asylum and border security from his much less popular moves against legal immigration, green cards, travel, and high-skilled migration. Bundling all of that as “Trump’s immigration policy” hides too much of the real politics.

• at Cato reports on the DHS quitting granting green cards almost entirely. USCIS now says adjustment of status should be granted only in “extraordinary circumstances” and that most applicants should use consular processing abroad. It is still unclear if the memo will be implemented as written, including forcing many people with pending green-card applications to leave the United States and apply abroad. But crazy stuff nonetheless.

• Tara Watson, Matthew Wich, and Johnny Willing at Brookings give the broader talent-pipeline version of the same concern in a long report on how the Trump administration is eroding high-skilled immigration. They project a 29% decline in new F-1 student visa issuances in 2025 and estimate a green-card backlog of roughly 1.2 million. Their useful point is that the pipeline is sequential: international students become OPT workers, some become H-1B workers, and some eventually become permanent residents. Breaking one step does not stay confined to that step.

• The American Immigration Council released “Restoring Credibility and Humanity”, an enforcement framework built around compliance, safety, proportionality, and accountability. As I have been saying before, there is very little programmatic work on enforcement from the left, so this is exactly the kind of thing the pro-immigration side needs more of: a concrete alternative to the mass deportation agenda that doesn’t just brush off prevalent people’s concern about legality and order.

• uses recent CBO projections to show that aging is in the future for the United States too. In his zero-net-migration scenario, the 20-64 population falls by 20.5 million by 2056 while the 65-plus population rises by 21.9 million, and the ratio of working-age adults to people over 65 falls from 3.1 to 2.1. This is the immigration-demography link in its least sentimental form: even countries that avoid near-term depopulation still age, and aging changes what labor, care, and growth require.

• pushes back on a too-simple fertility narrative in “The (Fake) Long Decline of Fertility”. His target is the viral story that fertility has just been sliding down the same long trend line for centuries. Instead, he separates long-run forces like child mortality, norms, and the cost of children from the more recent post-2007 decline, where he emphasizes partnership, sex, contraception, and digital technology.

As before, if you want me to write more about one of these or other related topics (or now if you want to write something yourself!), let me know. And if you came across something I missed this month, send it my way.

Share

Today’s post is a guest essay by Hanno Hilbig, a political scientist at UC Davis who does some of the most interesting work on political behavior. I’m excited to publish his assessment of the accommodation literature, which has become especially controversial lately. Too often, the debate gets flattened into either “moving right on immigration never works” or “mainstream parties should do it.” Hanno’s piece gives the more nuanced useful answer: accommodation can move some voters, but it also creates hard trade-offs around credibility, base defection, and issue salience.

In May 2025, Britain’s Keir Starmer warned of an “island of strangers” and rolled out a sweeping immigration white paper. A few months earlier, Friedrich Merz had pushed a five-point migration motion through Germany’s parliament with the votes of the far-right Alternative for Germany (AfD). Both leaders reached for a familiar playbook: move closer to restrictive immigration voters, reduce the appeal of the radical right, and hope to gain back voters drifting away from the mainstream.

Subscribe now

The temptation is easy to understand, as mainstream parties often do not align with their own voters on the issue of immigration. As documents, large representation gaps exist on cultural issues across Europe, with immigration among the clearest examples. In many countries, majorities of voters favor lower immigration, while mainstream politicians, including those in center-right parties, hold more permissive views. Such a gap looks like an opening, and so far the radical right has often been the party to exploit this gap.

Whether this kind of mainstream “accommodation” wins voters back remains an open empirical question. The research has not settled it yet. Existing evidence suggests that restrictive shifts can attract some anti-immigration or radical-right voters. But those gains can be offset by losses among pro-immigration voters and the mainstream party’s own base. Accommodation can also make immigration more salient, look less credible than the radical right, and make once-fringe positions seem more normal.

Parties usually turn to accommodation only when they are already losing. That is part of why it is so hard to judge: the studies finding it rarely works are looking at parties that were sinking anyway. Accommodation can win some voters back, but only when the shift looks credible and the party keeps its own base. When it can’t, the party just ends up boosting the rival it set out to beat, which is why mainstream parties so often lose either way.

Accommodation as spatial voting

So what is accommodation? In practice, accommodation usually involves one or more elements: policy movement, cooperation with other parties on legislation, and rhetorical signals. Merz’s case involved substantive movement on migration and a controversy over the norm not to cooperate with the radical-right AfD party. Starmer’s case combined an official policy white paper with rhetoric that made immigration highly salient.

The basic case for accommodation stems from what political scientists call “spatial voting.” In the simplest version, it means that we can imagine that voters and parties sit on a certain policy line. Voters prefer the party closest to them. Parties can convert some voters by moving toward them but may lose voters that they are moving away from. If voters want more restrictive immigration policy than mainstream parties currently offer, this creates an opening for radical-right parties. In turn, spatial voting predicts that a mainstream party may win back some voters by moving toward those voters in policy terms. Since mainstream politicians are relatively distant from some voters in terms of immigration positions, this suggests an opportunity to regain voters by closing the representation gap.

But several processes are not captured in a simple spatial model. First, accommodation on issues like immigration can make these issues more salient or important to voters. This could then benefit radical-right parties who “own” the issue of immigration.

Second, accommodative positions by mainstream parties may not be seen as credible. Voters may consider such positions to be “cheap talk.” This could be the case when accommodation is proposed by some party officials but criticized by others. It could further occur when accommodative rhetoric directly contradicts a long track record of policy positions of the party that proposes it.

Third, accommodation may reduce the stigma around radical-right parties or illiberal political ideas. This can happen because mainstream cues signal that positions, parties, or forms of rhetoric previously treated as outside ordinary democratic competition are now acceptable options for ordinary voters. There is a tradeoff between short-term electoral gains and long-term normalization of radical positions.

What We Should Expect

My discussion focuses on programmatic accommodation: a mainstream party’s substantive movement toward more restrictive positions on immigration, asylum, border control, integration, or immigrant welfare.

The spatial-voting expectation. The pro-accommodation case starts with a “representation gap.” If mainstream parties are more liberal on immigration than many voters, radical-right parties can attract voters who are to the right of the mainstream. Recent work supports this possibility. Guenther (2025) argues that representation gaps are especially large on cultural issues and that right-wing populists fill those gaps. In Germany, the immigration gap is especially relevant to AfD support among voters who consider immigration important. Alizade (2025) makes a related point: concerns about immigrant crime can create a mismatch between left voters and left parties, pushing some voters toward the center right.

This logic gives accommodation its strongest theoretical foundation. If some voters are close to mainstream parties on most issues but more restrictive on immigration, a mainstream party might gain by reducing that distance. In this view, accommodation is a response to underrepresentation rather than an attempt to pander to radical-right voters.

Salience and issue ownership. Spatial voting is useful, but it leaves out several things that matter in campaigns. Movement on immigration can make immigration more central. That matters because radical-right parties often benefit when politics is organized around immigration, national identity, or crime.

Immigration was already highly salient when Starmer made his 2025 pivot. In May 2025, Ipsos found that 49% of Britons named immigration as one of the biggest issues facing the country, the highest level since the 2016 Brexit referendum. At the EU level, immigration was also among the top concerns in recent Eurobarometer data.

High salience does not necessarily doom accommodation. If immigration is already unavoidable, a mainstream party may prefer to compete on it rather than ignore it. But if accommodation makes immigration more salient, it can change the payoff of the strategy. Issue ownership means that voters see one party as more competent or authentic on an issue. If voters see the radical right as more credible on immigration, a mainstream party may end up making the radical right’s strongest issue more important.

Credibility and policy delivery. Accommodation also requires credibility. A mainstream party with a long record of pro-immigration policies and visible internal divisions on the issue may struggle to convince voters that its pivot is real and durable. The problem is especially severe for rhetorical accommodation. Voters may hear a restrictive speech and conclude that the party is merely opportunistic, especially if the party’s record points in a different direction.

A U.S. example is Kamala Harris’s 2024 pivot on border security. Harris emphasized border security and tougher asylum rules while still supporting comprehensive immigration reform. Substantively, this moved her campaign toward a more restrictive border position. Politically, it also created a credibility problem, because the shift came after years in which the Biden administration was associated with a more permissive border and asylum posture. Voters may therefore have perceived the move as late electoral repositioning rather than a durable commitment.

Policy change may be more credible than slogans, but delivery creates its own problems. Visible change can keep immigration in the news. Invisible change may not be noticed. Failed delivery can reinforce the perception that mainstream parties talk a lot but do little.

Where the Lost Voters Go

Similar vote movements can have different consequences across party systems. In a bloc-based proportional system, a social-democratic party might lose pro-immigration voters to Greens, liberals, or other left parties while gaining anti-immigration voters from the radical right. If these voters remain inside the left bloc, accommodation may improve the bloc’s chance of governing even if the main party loses some supporters. This is the logic behind the more favorable Danish case in Hjorth and Larsen (2022).

But this logic does not always apply. If alienated voters abstain or move across blocs, losses are not contained. In a majoritarian system, a voter leaving Labour for the Greens or Liberal Democrats may still hurt Labour’s ability to win seats, even if Labour gains some voters who favor more restrictive immigration policies. Radical-right parties can also respond. They can emphasize mainstream parties’ lack of credibility, shift the debate to stricter positions, or argue that their own ideas have been vindicated.

Accommodation may also change what counts as legitimate politics. The concern is not only that mainstream parties lose or gain voters in the short run. It is also that their cues reduce the stigma around radical-right parties or illiberal ideas. Evidence for this first-stage effect appears in several settings. Valentim et al. (2025) show that anti-immigrant statements by mainstream-right politicians erode anti-prejudice norms more than similar statements by radical-right politicians. Daur (2025) shows that mainstream legitimization can increase the perceived legitimacy of pariah parties. Ekholm et al. (2022) show that mainstream recognition can increase sympathy toward the Sweden Democrats.

A U.S. example is the erosion of the distinction between legal and illegal immigration. On some accounts, Trump did not create the underlying restrictionist constituency, but he recognized it and pushed it beyond prior boundaries. The result was not only a tougher stance on illegal immigration, but a broader weakening of the old norm that kept legal immigration politically protected. This is the normalization mechanism in a different setting: once the boundary of acceptable restrictionism moves, later political entrepreneurs can push farther at lower reputational cost.

So, the net effect is hard to predict. Accommodation can work through several channels at once. It can persuade voters who see the mainstream party’s shift as credible. It can make immigration more salient for voters who trust the radical right more on the issue. It can also fail if voters view the pivot as opportunistic.

What Does the Evidence Say?

Researchers have made progress on the relationship between accommodation and electoral returns. But the evidence is genuinely hard to read, for a simple reason: parties rarely accommodate at random. They usually move right on immigration when radical-right parties are already gaining, when immigration is already salient, or when public opinion has shifted. If radical-right support then changes, it is hard to know whether accommodation caused the change or simply followed the same trend.

The cross-national evidence is mixed and does not support a pro-accommodation story. Krause et al. (2023) find little evidence that mainstream accommodation reduces radical-right support, and some evidence that it may increase movement to the radical right. Dahlstrom and Sundell (2012) find, in observational evidence from Swedish municipalities, that tougher average mainstream positions are associated with higher Sweden Democrat support. Down and Han (2020) do not directly measure legitimacy, but their voter-level evidence is consistent with a legitimization channel: restrictive mainstream positioning is associated with more radical-right voting among voters who do not already see the radical right as competent on immigration.

Other studies are more favorable. Spoon and Klüver (2020) find that mainstream-left parties benefit electorally from going tougher on immigration, while mainstream-right accommodation is largely null. Thesen (2025) reanalyzes part of this literature with controls for immigration news and radical-right media visibility, and finds that once those media variables are included, accommodation appears to hinder radical-right success.1

The strongest causal evidence comes from designs that separate accommodation from the trends that usually precede it. Conjoint designs ask respondents to choose between profiles whose attributes are randomly varied. Survey experiments randomly assign information about party positions. Unexpected-event designs use real political shocks that occur during survey fieldwork. These designs are narrower than cross-national studies, but they are closer to the central question: whether a mainstream party’s restrictive movement changes vote choice, vote intention, or propensity to vote.

Chou et al. (2021) study German voters using a conjoint design that varies candidate positions on immigration and other issues. Restrictive mainstream positions can attract some AfD voters, which supports the spatial argument. But the same positions can alienate mainstream-party voters. The net party-level effect depends on whether gains among the target voters exceed losses elsewhere.

Hjorth and Larsen (2022) study the Danish Social Democrats after the 2019 election, when the party’s immigration position was plausibly ambiguous to voters. Their survey experiment told some respondents that the party had taken an accommodative immigration position and others that it had taken an adversarial one. The study finds that accommodation can attract anti-immigration and Danish People’s Party voters. Losses among pro-immigration voters were partly contained because many of those voters remained inside the left bloc. This is the clearest case in which accommodation looks electorally useful, but it also shows why institutional context matters.

Kollberg (2026) separates positional accommodation from populist rhetoric in a pre-registered factorial survey experiment. The central result is that position matters more than populist style. Accommodation increases propensity to vote among right-leaning voters but repels left and centrist voters. Again, the spatial mechanism is present, but so is the offsetting-loss mechanism.

An important recent non-experimental case is the preprint by Turnbull-Dugarte et al. (2025). They use the timing of British Election Study fieldwork around Starmer’s May 2025 speech. Some respondents were interviewed just before the speech and others just after it, so the episode works like an unexpected-event treatment. That makes the design unusually close to the question here: what happens after a real social-democratic accommodation episode? It is still not a pure test of policy delivery, since the treatment combines rhetoric, policy announcement, and media attention. The paper finds that exposure made Labour look more anti-immigration and right-leaning, reduced Labour support, and did not reduce Reform UK support.

Overall, the strongest causal designs suggest that accommodation can move some anti-immigration or radical-right voters, but these gains are often offset by losses among centrist and pro-immigration voters. Experiments also hold constant many parts of real campaigns: media attention, elite conflict, credibility, and radical-right responses. That makes the evidence useful but incomplete, and how much you trust it comes down to how much weight you put on a clean causal estimate versus the realism of an actual campaign.

Interlude: What Researchers Should Do Next

Before concluding, I would like to add a few thoughts on what ambitious grad students and other folks who care about this issue can do next. The main evidence gap is pretty clear at this point. We need more designs in which mainstream accommodation is the shock and both mainstream and radical-right electoral outcomes are observed in real elections. The key challenge is to separate accommodation from the political trends that usually precede it, such as rising immigration salience, radical-right growth, or shifts in public opinion.

Cross-national research will remain valuable, but even the best versions face the same problem. Parties accommodate when the political environment is already changing, and that endogeneity is difficult to solve with aggregate data alone.

One promising direction is subnational or candidate-level research. Close elections between candidates with clear programmatic differences on immigration, asylum, policing, or integration could provide useful evidence, especially when the office has visible policy authority. MPs, mayors, and local party branches also sometimes take positions that differ from their national party. These cases create more variation than national party-level studies, although national party brands may still drown out local differences.

Another promising direction is unexpected-event designs. Survey fieldwork sometimes overlaps with high-profile party pivots or policy shocks. These designs capture real-world accommodation without asking voters to imagine a hypothetical campaign. But they depend on the case that happens to occur. They cannot tell us what the effect would have been if the pivot had been more credible, or if issue salience had been held constant.

What Accommodation Can and Cannot Do for You

The spatial voting argument remains the strongest case for accommodation. If mainstream parties are more liberal on immigration than many voters (which they often are!), a more restrictive position can reduce the distance between the mainstream party and those voters. The evidence reviewed above suggests that this works. Accommodation can make mainstream parties more attractive to some anti-immigration voters. But a spatial gain among some voters is not the same as a positive net gain after factoring in the “side effects” of accommodation.

Recent arguments that “mimicking” the radical right is electorally self-defeating capture an important concern. Accommodation can make immigration more salient, activate an issue on which the radical right is often viewed as more credible, alienate centrist and pro-immigration supporters, appear opportunistic when it conflicts with a party’s prior record, and weaken the boundary between mainstream and radical-right positions.

But the broader conclusion that accommodation cannot work is too strong. The evidence points to a more conditional claim: accommodation can win some voters – but the shift has to be credible, voters must be persuadable, and losses elsewhere must be contained. The Danish case is the one most often associated with these favorable conditions.

The skeptical view is that these conditions are demanding and may often not hold. That skepticism is important, but it does not make the strategy irrational from the perspective of parties that choose it. Mainstream parties often accommodate from a position of weakness, after immigration is already salient and after some voters have already begun to defect. Mainstream parties may therefore choose accommodation even when they know it is risky. If defeat already looks likely, accommodation is less a path back to power than a bet to slow the bleeding. And that is the deeper bind: by the time accommodation looks tempting, a mainstream party is rarely choosing between winning and losing. It is choosing between two ways of losing, and betting that accommodation is the slower one.

Share

Subscribe now

Nearly 80 percent of American voters support high-skilled immigration, across party lines. That is more popular than nuclear power, building apartments, or deregulation. And yet not a single major political figure in either party is willing to champion it.

Republicans know the country needs foreign scientists and engineers. Most will say so privately. But they refuse to say it publicly, terrified of being outflanked on their right.

The left has a different version of the same cowardice: the abundance crowd wants more housing, more energy, more growth, but has quietly decided immigration is too dangerous to include in their pitch.

The piece proposes two concrete moves that do not require Congress: promote and expand the O-1A extraordinary-ability visa, which has no cap and no lottery, and fix how the government sets wages for foreign workers.

The Labor Department’s public comment window on the proposed prevailing wage rule closes May 26. If you work in research, tech, or immigration, submit a comment before it’s too late.

I’d love to hear what readers think I’m getting wrong or what’s missing from the argument. After all, if this were that easy and obvious, some Republican or Democrat would already be trying it much more forcefully.

P.S. As a fun tidbit for my readers, this is actually the first cold pitch I ever landed. That is, I just literally emailed a WaPo editor with the pitch and the piece, and they were like—yep, let’s publish it. This never happened before quite like that.

Probably non-coincidentally, this is also the most revised and re-revised 1,000 words of my career. Thanks to all the folks who have read and edited the piece for pushing me over and over again and making it all possible.

Share

I’m excited to share our first guest post on Popular by Design from . Gil does some of the best migration research around at the , grounded in hard demographic data, and it shows in his feel for what actually drives people to move. I found his read on the potential Cuban migration persuasive and constructive without being complacent. Follow Gil and his new work at Points of Entry.

Speaking on Face the Nation this past Sunday, former Secretary of Defense Robert Gates argued that the greatest national security risk posed by Cuba to the United States is another massive refugee outflow along the lines of the 1980 Mariel Boatlift. This sentiment has been approvingly echoed by established experts as an inevitable contingency the United States will have to deal with if it does intervene on the island.1

While no one can predict the future with certainty, this point seems intuitive at first glance. Cuba has a well-established history of weaponizing migration toward the United States, and the Syrian Civil War has ingrained the association between violent conflict and mass immigration in people’s minds.

Subscribe now

But on a closer look, the specifics that govern how large a Cuban departure wave can actually get all point in a generally conservative direction. Four features of the present moment should make us rethink the conventional wisdom and expectations around another Cuban migration crisis.

The people most likely to leave a crisis have largely already left. Migration waves recruit disproportionately from the young, the working-age, and the mobile, and Cuba has spent four years exporting precisely that cohort. The regime conceals the magnitude (it counts emigrants as residents until they have been gone two years) but an independent reconstruction by the demographer Juan Carlos Albizu-Campos puts the actual resident population near 8.62 million and total departures since 2021 at roughly 1.79 million, a figure consistent with Cuban arrival counts in the United States and elsewhere.

Composition matters more than the total here. An estimated 77 percent of recent Cuban migrants are between 15 and 49, a slight majority of them women. The Cubans who remain on the island today are older, sicker, poorer, and likelier to stay. A recent estimate I authored for the Niskanen Center caps a weaponized scenario at 60–75 percent of a Mariel-equivalent event on these grounds alone.

The regime has reduced incentives to encourage irregular migration. Cuba has a long record of wielding migration as a coercive instrument before, and part of the reason it worked according to political scientist Kelly Greenhill is because this type of coercion played on American humanitarian concerns and domestic political pressures over the treatment of Cuban migrants.

Whether you agree with the administration’s positions or not, it’s hard to argue that they are overly concerned with migrant wellbeing or popular approval. In our present state of affairs, any attempt to facilitate (let alone actively encourage) irregular migration would almost certainly be treated as casus belli by a White House seemingly itching for an excuse to intervene.

The calculus might change in the event of American military action on the island, but as I’ll argue in a forthcoming paper for Florida International University, we haven’t seen a migratory surge out of Venezuela following Operation Absolute Resolve, and Venezuela has a much larger population and geography that is better suited for mass migration.

Migrants would have only one feasible route, and the United States should be prepared for it. During both the 1980 Mariel boatlift and the 1994 Balsero crisis, the United States largely had to scramble to mount an adequate response under the pressure of a migrant wave that was already underway. In the present day, the Coast Guard appears to be running its standing maritime interdiction program (Operation Vigilant Sentry) at an elevated level. The most recent public disclosure from 2024 put our assets in the area at roughly eighty cutters against a baseline of thirteen. It’s possible, but unlikely, that the administration has wound this down given how telegraphed its willingness to confront Cuba has been so far.

While recent migration waves from Cuba have come by other means, at present the sea is the only theater where a mass migration event could take place. The 2021-2024 spike in Cuban encounters came from Cuban flights through countries like Nicaragua followed by travel on foot to the southern border. Nicaragua revoked visa-free access for Cubans in February 2026, and Mexico has stepped up interior enforcement to stop migrants from reaching the border.

Thus, migrants determined to leave Cuba for the U.S. will have to take their chances via the Florida Straits, where chances of interception are high and where their ability to leave at all will have to clear difficult hurdles like vessel quality and availability and favorable weather conditions. The fact that the dire conditions Cubans are currently experiencing on the island haven’t led to more irregular migration attempts suggests that Cubans are pricing in these factors in their decisions on whether to try to leave, which is the subject of the final constraint.

Cuban migrants have no reason to expect their journey will end well. Migration responds to expected outcomes, not to push factors alone, and the expected outcome for a Cuban reaching the United States in 2026 is an unpleasant detention and removal.

Every pathway that structured the last wave is closed: CHNV parole has been terminated, CBP One has been shut down, the Cuban Family Reunification Parole program was also terminated (but is in active litigation), a travel ban is in force, USCIS processing for Cuban nationals is paused, and the Cuban Adjustment Act is functionally inert.

A prospective migrant weighing the crossing confronts a harder route and a near-certain unfavorable result at the end of it, and that calculus dampens the pull even when the push is acute. The underlying factor that convinced many Cubans to undertake the treacherous maritime route to the United States during previous migrant crises was the “Wet foot, dry foot” policy, which guaranteed them a reasonable chance to pursue legal residency if they could reach the United States. No equivalent to this exists today, and there is no reason to expect one to appear in the near future.

What should the U.S. do? No one can be fully sure of whether a migrant wave will occur after a hypothetical conflict in Cuba or how large such a wave would be. But the administration can do two things now to ensure that minimal harm befalls ordinary Cubans seeking to leave the island if a military conflict does trigger a significant exodus.

The first is for White House officials and civil and military planners to craft a more detailed and long-term migration mitigation strategy on the island. The plan that exists now seems to be a holding pen: the head of SOUTHCOM has testified it would support DHS in a Cuban migrant crisis by standing up a camp at Guantánamo Bay. This answers where an outflow goes but not what happens once migrants get there. A camp without pre-set screening criteria, release standards, and a defined endpoint makes the base a ticking detention problem. Contingency planning here is prudent, and seemingly incomplete.

The second is that the administration, in partnership with Congress and with the help of the South Florida delegation in particular, should be negotiating third-country humanitarian arrangements for Cubans now, while flows are low, rather than improvising them under the pressure of a wave. Brazil, Uruguay, and Spain are already informally absorbing the redirected Cuban flow. Structured agreements that include specifics like resettlement slots, orderly processing, and burden-sharing in exchange for diplomatic and economic considerations would convert an unmanaged diversion into a managed one and free Washington from the hard choice of taking fleeing migrants ourselves or leaving them trapped on the island. The time to build that channel is before the boats are in the water.

Share

For a brief moment about a year ago, it really did look like Bluesky might work. Researchers and left-of-center intellectuals were flooding in, swapping starter packs, reassembling what felt like a nostalgic reunion of old Twitter. Then everyone arrived, and the center could not hold. It turns out that people can disagree even when they are all on the left, and that without strong social norms of free speech and civil conduct, humans—including university professors with allegedly high IQ—remain tribal and quick to pile on. Meanwhile, LinkedIn (yes, LinkedIn!) has quietly been doing the job Bluesky was supposed to do.1

Like most researchers and public intellectuals I know, I had a LinkedIn account for years that I barely used: I would accept the occasional connection request and otherwise ignore the platform. At some point I noticed that people were actually talking on it, often substantively, in a way I had previously expected only from old Twitter. So, for the last year or so, I've been cross-posting essentially the same content to Twitter, Bluesky, and LinkedIn. At this point, the pattern is consistent enough to feel like an A/B test. Pieces that read as pro-immigration get cheers on Bluesky and silence on X; pieces that read as "anti-immigration" get the reverse. Pieces that carry the most nuance get nothing on either platform. But LinkedIn has been the exception.

Subscribe now

It is not hard to specify what functional public-facing research discourse should look like in theory. You post something new and informative that you care about, like a working paper on a policy-relevant topic. People who find it interesting or useful say so and add nuance. People who disagree explain why in a respectful, substantive way. People who find it uninteresting ignore it. Nobody calls you names, impugns your motives, or turns a technical dispute into an accusation unrelated to your argument. If they do, they are in the minority, and they get called out.

These principles once described Academic Twitter. On Bluesky, they have mostly collapsed. Bluesky has fundamentally failed as a venue for public-facing research and, more importantly, as the channel through which serious research reaches policymakers, journalists, and the broader public. The platform that has quietly taken over that function is LinkedIn. If a lot of sane people left Bluesky or X for LinkedIn tomorrow, or at least started reposting their content there, the research internet would be better for it, and so would the rest of the internet.2

What the public square was actually for

Public-facing research had a simple social media function for most of the last decade. It was a cheap pipeline from research to policy, to journalism, and to the reading public. Old Twitter did this job unevenly but at scale through the famous posting-to-policy pipeline, as well put by . It was where a congressional staffer bumped into a political scientist’s thread and turned it into a briefing, where you learned about a working paper before it was published, where a fight over whether an RCT had identified the right thing played out in real time for an audience that partly understood and partly just absorbed the norms of how serious researchers talk.

When Elon Musk bought Twitter and rebranded it as X, the pipeline broke: the algorithm got more chaotic, external links got suppressed, spam and reply-guy slop proliferated, and blue-check monetization created a new economy of rage-farming accounts. The new regime did loosen some of Old Twitter’s speech restrictions around heterodox positions on public health, gender, and race, and for a minority of researchers on those topics, post-Musk X is genuinely freer than its predecessor. But the net effect was negative on balance, and a large plurality of social scientists, think-tank writers, and left-of-center journalists migrated to Bluesky over 2024-2025.

Why Bluesky failed

The short version is that most of what people used to blame on Twitter’s algorithm turned out to be problems with users and norms. Bluesky removed the engagement-maximizing amplification but, through aggressive self-selection during the Musk-era migration, kept a left-of-center user base disproportionately invested in policing speech and unwilling to treat ideological deviation as anything other than a moral failure. The predictable result is that any post touching a politicized topic (migration, AI, racial disparities, or whatever gets politicized next) draws volume rather than argument. For most researchers most of the time, the rational response is to say less rather than more, and the deterrent scales faster than the discourse.

This is not just something right-wingers complain about. As argues from the center-left perspective, Bluesky’s discursive norms are almost designed to repel outsiders. also argues that Bluesky functions less as a political movement than as a tribal affiliation with a narrow demographic profile; and observes that progressive commentators on Bluesky, having lost their mainstream audience, now spend much of their energy trying to cancel each other.

I have often experienced this all firsthand: academics respond to even my less controversial Bluesky posts by email, because they don’t want to risk getting piled on for saying anything in public. The numbers bear out this diagnosis: Bluesky peaked at around 40 million registered users by late 2025. Pew’s 2025 survey finds just 4 percent of U.S. adults have ever tried Bluesky.

Why LinkedIn

LinkedIn is not a perfect platform, and I will get to the downsides in a moment, but the case for it rests on a handful of things that have quietly become decisive for anyone doing public-facing research.

Scale and distribution. LinkedIn reports 1.2 billion registered users globally, roughly 30 times Bluesky’s total user base and orders of magnitude higher than its daily active count. A reasonable skeptic asks why an account number should translate into actual reach, since having a profile is not the same as reading a feed. Two things answer that. First, LinkedIn’s algorithm explicitly pushes posts with early engagement outward to your 2nd- and 3rd-degree connections, to followers of relevant topics and hashtags, and to professionals in the same industry, regardless of whether they follow you directly. A strong LinkedIn post travels to strangers in a way a Bluesky post simply cannot.

That said, LinkedIn engagement is notoriously hard to measure externally, because LinkedIn closed its public API in 2015 and restricts analytics data to approved Marketing Developer Platform partners, which is one reason the migration of research-adjacent discourse to LinkedIn has been largely invisible to researchers who expect to measure platforms the way they measured Twitter.

Audience mix. Think of LinkedIn as a kind of Switzerland of the internet: higher-status people from different camps show up under their real names and engage with each other because the professional cost of behaving badly is real. This is also the point most researchers have not fully absorbed, as I can attest, and it explains the apparent paradox of LinkedIn being the dominant platform for research-adjacent discourse without most researchers noticing. Bluesky is full of other researchers, the journalists who cover them, and activists.

The people on LinkedIn are the people we should be trying to reach: policymakers, congressional staffers, civil servants, industry analysts, foundation program officers, and journalists at general-interest outlets. A 2025 Avoq survey of DC policy insiders found that 81 percent of Democrats, 84 percent of Republicans, and 78 percent of MAGA-aligned respondents use LinkedIn. Good representative data on LinkedIn compared to other platforms is notoriously hard to find, but this looks like a bipartisan footprint no other platform comes close to matching. Researchers have not clocked the shift because the people reading their LinkedIn posts are not the people they spend time with online; they are the people their work is supposed to reach.

Format favors substance. LinkedIn’s format (longer posts, real names tied to real careers, a less snarky default register) does a lot of the work of civilizing discourse without needing heavy moderation, because when the poster is visibly accountable to an employer and a professional reputation, the median comment tone shifts correspondingly, and bad-faith quote-dunking becomes rarer. There are also no anonymous accounts and almost no sub-tweeting; the median post reads more like a memo than a hot take.

People can still disagree or criticize you heavily if you post something provocative, but they are much less likely to do it in a mindless or righteous way. In some ways, LinkedIn feels like an academic conference: people are civil, sometimes too nice, and not always willing to criticize a colleague openly. That conference-style politeness can smooth over real disagreement, but it is a much better failure mode than ad hominems and pile-ons.

Discussion that actually moves understanding. The clearest evidence I have for all of this is my own cross-posting experience. I have often shared the same piece, including the more controversial ones, simultaneously on Bluesky, X, and LinkedIn, and the pattern has been remarkably consistent. On Bluesky, the reaction is usually either silence or a small pile-on when the piece challenges prevailing consensus, and substantive engagement is rare. On X, responses are a mix of real engagement and the usual ratio of slop, bad-faith screenshotting, and reply guys.

On LinkedIn, the pushback I get is both the most civil and the most productive: named professionals who actually work on the topic, often from perspectives I don’t share, who write multi-paragraph responses that engage with the argument rather than perform outrage about it. This holds even for pieces and takes I expected to trigger the most hostility, because people disagreeing under their own name with their employer looking over their shoulder have strong incentives to be reasonable.

No, LinkedIn is still not perfect

A pragmatic case for LinkedIn has to be honest about what the platform does poorly. I’ve seen anonymous accounts on X complain that LinkedIn has been quietly suppressing links to right-coded publications for some time. I haven’t been able to verify this, but if you write from the political right, or many of your sources are outlets that LinkedIn’s filters treat as low-quality, your reach may be cut in ways that are not transparent. Even if this is true, for a left-leaning researcher writing about migration or AI, this is a non-issue, while for a right-of-center one writing about the same topics, Bluesky’s ideological environment is more hostile, and LinkedIn’s algorithm is not neutral either.

The interface is also genuinely clunky. The composer is awkward, threading is mediocre, search is bad, and basic features that X and Bluesky get right out of the box are either missing or buried. This is a fair complaint, but it is also a complaint that gets answered as the platform’s serious-discussion userbase grows: the more prominent intellectuals show up expecting a real public square, the more pressure LinkedIn faces to build the tools they need (I’m looking at you, ).

And yes, LinkedIn has its cringe: the AI-generated inspirational posts, the three-emoji bullet lists, the humblebrags about how humbling it was to be invited to speak somewhere, the AI-generated summaries of papers the poster obviously has not read. None of this is enjoyable, but it is easy to filter past, and the cost of ignoring the slop is low compared to the cost of staying on a platform that actively punishes substantive engagement.3

Come build the bridge

Every sane researcher who migrates to LinkedIn takes one more useful voice with them. The tipping threshold is probably not high: a few hundred visible researchers moving their primary public-facing work to LinkedIn, as many have already been doing quietly, would shift the center of gravity enough to make the migration self-sustaining.

If you are skeptical, you don’t have to abandon Bluesky (or X for that matter) tomorrow. Just start cross-posting whatever you’d normally write there onto LinkedIn too for a couple of months, and watch the difference in who shows up in your replies. The on-ramp is low-friction: post a short summary or excerpt of your latest piece with the link, tag a few people whose work it actually engages with, and see what comes back over a week or two. Based on two years of doing exactly this, my bet is that the LinkedIn version will draw substantive engagement from people whose opinions actually matter to your work, while the Bluesky version fades into silence or a small pile-on when you’re even slightly challenging the prevailing consensus. You won’t need another essay to convince you.

I am on LinkedIn here. Follow me, post your research and writing, and let the audience decide. The public square gets built wherever serious people choose to show up. Right now, that place is LinkedIn.

Share

Quick personal news before we get to the links. As this newsletter goes out, I'll be on my way to the Borders and Belonging conference at IAST in Toulouse, one of the more exciting migration gatherings in Europe this year. If you're there, come say hi. On the way, I've been trying to visit every Notre Dame cathedral I can reach, on the theory that this is what most of my non-American friends assume my employer looks like anyway.

At the actual University of Notre Dame, we just hosted the 2026 Midwest Migration Conference using a new workshop-format without presentation slides, gathering the new and existing work on the botched reception of Venezuelan asylum seekers in Chicago, the effects of anti-immigration policies, and the difference between various Chinese communities in Paris. Let me know if you want to participate next year.

Subscribe now

Looking further out, I’ll be speaking at the Progress Conference 2026 in Berkeley on October 8-10 about immigration as a key progress issue. It’s one of my favorite conferences out there, both because the attendees are unusually thoughtful and motivated to make things better despite their disagreements, and because it’s one of the few places where people treat policy design as a genuine problem to be solved rather than an afterthought. If you’re going, come find me. And if you don’t, try to sign up and come anyway.

April was busy. At Migration Policy Institute, Caitlyn Yates and I traced 15 years of UK immigration policy swings and the salience cycles that drive them: salience rises, labor visas get cut, shortages follow, and visas quietly reopen. At The Atlantic, I published “How the Left Accidentally Bolstered the Nativist Right” (their title, not mine), on what we lost when the “I only oppose illegal immigration” norm collapsed. In the Frankfurter Allgemeine Zeitung, I explained why the German migration debate has gone off the rails, specifically how “well-intentioned disinformation” from the center-left feeds the far right. I released a Keough School policy brief, “To Protect Humanitarian Immigration, Build Public Trust First”, on how the sequential logic of my case applies to refugee and asylum policy.1 And, thanks to The Chronicle of Higher Education, I now have “AI Is a Better Researcher Than You” as a headline attached to my name forever.

OK, let’s move on to something a bit awkward. After two rounds of online pile-ons this spring, I’ve received some earnest private feedback from colleagues, with some worried about my mental health and some others worried about me becoming a “right-wing crank” for lack of a better word, all suggesting I should dial back how much I post on social media (I’m talking about short-form takes rather than long-form essays). I want to take that feedback seriously, so I’d like to also hear from the people actually reading this newsletter and following me on social media. Conditional on being factually accurate (which I still strive for believe it or not):

Here are the April links (linking does not imply endorsement):

• launched In Development this month, a new magazine on what works and what doesn’t in global development. If you liked her job market papers roundup and her thinking on evidence-based policy, this is the natural next step. The inaugural piece, a guest essay by GiveDirectly co-founder Paul Niehaus on how evidence supported the organization’s journey to delivering $1 billion in unconditional cash transfers, is worth reading in full. Immigration is a huge part of global development, so I’m glad Lauren is doing this. There is also a rumor that their next piece will be exactly on immigration.

• Amy Nice and Paola Sapienza have a new Hoover Institution conversation on “Unlocking Global Talent: The J-1 and O-1A Visas”. It’s the clearest thing I’ve seen on the practical alternatives to the H-1B lottery for companies trying to hire international STEM talent. Amy is one of the sharpest practitioners working on legal immigration in Washington, and she’s unusually good at explaining the mechanics to non-specialists.

• at No Idle Sitting argues Trump’s H-1B lottery policy kind of works. The $100,000 fee is clumsy and imprecise, but it does effectively raise the bar on employer-sponsored immigration in roughly the direction selection-based systems should go. Take our wins where we can get them.

• We had some minor disagreements with lately, but when he’s right, he’s right. His new piece on the “culture crutch” is a banger. The problem with “culture” as an explanation is that it’s the most conceptually overstretched term still allowed in polite scholarly society. Whenever it’s actually useful, more precise concepts (norms, institutions, prices) usually do the work better.

• on selection, not origin, driving immigrant welfare use. When you control for pathway type, country-of-origin effects go away. The policy lever worth pulling is how people come (i.e., for needed work), not where they come from.

• My friends and co-authors James Dennison and Andrew Geddes have a new book forthcoming soon, What Europeans Think about Immigration and Why It Matters (Oxford). It maps decades of European survey data and offers a proper framework for why immigration becomes politically salient when it does. I pre-ordered mine. Anyone serious about European immigration politics should.

• at The Argument on whether a liberal society can do affirmative action, makes a point that generalizes well beyond DEI. As I’ve been beating this drum on immigration, the problem with unpopular policies isn’t usually that they’re sold poorly. It’s often that the policies themselves need to be different. Reframing only gets you so far when the underlying design is the thing voters are rejecting.

• at Stanford has a must-read piece on how AI is already 10x-ing academic research. His line I keep repeating: “stop waiting for permission. The tools are here.”

• at New Instruments asks who actually answered this survey in a world where people routinely delegate tasks to AI. Survey research has to shift from detecting AI-generated responses to redesigning instruments that make authentic participation easier than delegation. Yamil was creatively using AI in polls before it was cool, and he’s worth following if you want to make sense of where public opinion research is headed.

• Besides the aforementioned Progress Conference 2026, is still taking applications for the BBI fellowship (deadline: June 1). I’ll be advising and speaking on the human talent and potential track there. Highly recommended for any academic (or non-academic writer for that matter) seeking more serious public engagement.

As before, if you want me to write more about one of these or other related topics, let me know. Or if you agree that I should stay away from social media for some time before I turn into a complete crank, this is your chance to speak up, too :)

Share

Originally published in September 2025, this was the first long-form essay I wrote for Popular by Design, and the one that jump-started it. I only had a few hundred subscribers at the time. I still think it’s one of the best things I’ve written, since getting humanitarian immigration right is especially hard. I hope you enjoy it.

Despite all its supposed potential, immigration is deeply unpopular today. Refugee and asylum immigration is even more so, because humanitarian appeals don’t resonate much with voters. Most want to see clear benefits for their own country, not just compassion for strangers abroad. That’s why expanding refugee admissions is politically much harder than skilled or labor migration.1

The premise of this newsletter is that making meaningful progress on immigration requires more than just better messaging—it requires better policy. So I wanted to begin with one of the hardest cases and write about a possible solution for making humanitarian immigration more popular and sustainable. What I learned while working on this piece is that one doesn’t have to be a bleeding-heart liberal to support refugees.

Subscribe now

Enter programs for private or community sponsorship of refugees for permanent resettlement. The model was first launched in Canada in 1979 and is now being considered or piloted in other countries, including the United States. This policy innovation directly addresses a common skeptical retort in political debates about immigration and humanitarian obligations: “Why don’t you house them yourself?” This question, often used by skeptics to imply hypocrisy among pro-immigration advocates, points to the real and perceived costs of resettlement borne by taxpayers.

But the simple truth is that many people would gladly help refugees with their own money and resources—they just cannot legally do so. Outside of Canada, in most countries around the world—rich or poor, democracies or autocracies—only governments decide who gets to immigrate or resettle there and how, regardless of how generous their populations may be. This issue cuts across ideology—orthodox congregations can’t bring in culturally similar believers, while humanitarians can’t help families in danger even if they want to do it on their own dime.

Community sponsorship aims to change that. It gives willing individuals and private organizations a legal way to act on their motivations to help migrants, share the financial and social costs of resettlement, and show tangible benefits of migration to their communities. Just as importantly, unlike other pro-immigration policies, it creates a durable constituency of both conservative and liberal citizens with a direct stake in immigration and refugee protection. While this is a hard counterfactual to prove, I’m increasingly convinced that had Canada not pioneered sponsorship 45 years ago, it would have resettled far fewer refugees, and its immigration politics would be far more contentious.

There are many good overviews of what community sponsorship is, along with case studies and policy assessments. What I want to do here, in the spirit of Popular by Design’s mission, is something I haven’t seen anyone do yet: assess the potential of sponsorship to generate a more pro-immigration political consensus, both in theory and in practice, and with an open mind for its possible limits. Below, I outline the longest-standing Canadian PSR program, what we can learn from its successes and shortcomings, why it hasn’t spread more widely, the major criticisms, and the available polls on how this idea is received globally. I conclude with a discussion of America’s short-lived Welcome Corps sponsorship program (launched in 2023 but abruptly halted by the second Trump administration), and how better policy design could set it up for greater success if or (hopefully) when it resumes.

What is community sponsorship, and how does it operate?

Community sponsorship is a set of policies that let individuals, community groups, and nonprofit organizations sponsor specific refugees for resettlement in their country, in addition to or otherwise independently of traditional government resettlement. Sponsors cover housing and basic needs, provide social connections, and help with integration, for a defined period, typically twelve months after arrival.

Canada runs the longest-standing and most developed system. Since 1979, hundreds of thousands of regular Canadians have helped resettle around 400,000 privately sponsored refugees with the help of more than 200 local and faith-based groups, all in addition to government-assisted arrivals. In recent years, a slight majority of resettled refugees have come via private sponsorship, and federal targets now plan for more private than government-assisted admissions. Here is the basic breakdown of the current version of Canada’s Private Sponsorship of Refugees (PSR) program:

• Who can sponsor: Small groups of five or more Canadian citizens or permanent residents (“G5s”), Community Sponsors (local organizations such as cultural associations, schools, or municipalities), and Sponsorship Agreement Holders (“SAHs”) which are established charities, faith-based communities, or nonprofits previously approved by the government. SAHs also educate and support sponsors and the sponsored, and help resolve issues that arise.

• Who can be sponsored: Canadian sponsors may “name” a person abroad who meets Canada’s refugee definition. For sponsorships by G5s or Community Sponsors (but not SAHs), the person must also generally already be recognized as a refugee by UNHCR or a foreign state.2 Because global resettlement slots are scarce (UNHCR projects about 2.5 million refugees in need of resettlement in 2026, a fraction of the 30+ million recognized refugees worldwide), the eligible pool is rather constrained. In practice, the vast majority of named cases are distant relatives or close friends of people in Canada.

• What’s required from sponsors: Sponsors commit to 12 months of support: start-up funds, income support, housing, and hands-on help with school, work, and language. Government guidance suggests budgeting about 26,700 CAD for a family of three (minimum, varies by location and in-kind support).

• What happens to those sponsored: Resettled refugees arrive as permanent residents, receive federally funded interim health coverage, and after the sponsorship year, can access regular provincial benefits like all other residents.

• What the government still does: It sets and manages annual admissions targets (currently 21,000–26,000 for 2025, with new PSR applications paused until December 2025 to reduce backlogs), vets applications, conducts security and medical screening, issues visas and permanent residence, and monitors compliance across all resettlement streams. The federal and provincial governments are responsible for healthcare coverage from the time of arrival and for other benefits that accrue to permanent residents.

The system is now considered a global model that has inspired adaptations in at least 14 other countries while securing financial and other support pledges from dozens of organizations. In 2016, together with UNHCR and a range of non-profit partners, the Government of Canada launched the Global Refugee Sponsorship Initiative to promote community sponsorship as a complementary pathway for resettlement around the world. Since 2013, Canada has also been running a “mixed” stream, the Blended Visa Office-Referred (BVOR) program, where sponsors are matched to UNHCR-referred (rather than named) refugees and costs are shared with the government. Many countries have modeled their sponsorship schemes on either this matching approach or the traditional naming approach, with varying parameters.

In Australia, for example, the sponsorship programs can involve business support, but are explicitly counted within the same annual Humanitarian Program quota. In the United States, the Welcome Corps sponsors only provide the first 90 days of core services with arrivals entering as refugees and applying for permanent residency after one year. In Italy, the “Humanitarian Corridors” program allows only organizations (not individuals) to sponsor people on humanitarian visas, so there is no guaranteed permanent residency on arrival.

Why community sponsorship wins more support than resettlement or asylum

Although Canada’s program has occasionally been criticized over sponsor–refugee matching, long wait times, and tension with government quotas, it has not caused any significant right-wing backlash. The same is not true of humanitarian immigration generally—and asylum in particular—which often raises concerns about border chaos, arguably a major driver of recent populist resurgence worldwide. Even in Canada, the right of foreigners to claim asylum at the border is much more controversial than either government-assisted or privately sponsored resettlement or foreign aid.

The political promise of community sponsorship lies exactly in how it channels citizens’ both altruistic and somewhat parochial impulses—helping people you can identify with—into a structured way to resettle vulnerable populations from abroad while maximizing integration success and minimizing the concerns of skeptics. By providing individuals and organizations with a legal and effective way to help, community sponsorship makes larger refugee resettlement more politically durable in otherwise hostile anti-immigration environments.

First, it allows willing citizens to act upon their humanitarian beliefs beyond helping migrants who are already here or voting for a preferred party and leaving refugee protection solely to politicians and bureaucrats. The act of communal sponsorship builds lasting civic networks and constituencies of people invested in resettlement and immigrant success more generally. Research from Canada and other countries shows that sponsors overwhelmingly report positive experiences and stronger ties to their communities.

Second, it appeals to people’s conservative intuitions of localism, faith, and control, especially when “naming” the sponsored refugees is allowed. It is not a coincidence that Canada’s private sponsorship program roots lie in church-based aid and local civic voluntarism. Faith communities were already running settlement ministries and pressing the state to share responsibility, and then stepped in as enthusiastic yet “reluctant partners” during the late-1970s resettlement of Southeast Asian refugees. According to a recent survey of sponsorship organizations in Canada, 60% of them still belong to a religious organization, while 22% focus on another particular non-religious ethnic community or group.

Third, community sponsorship explicitly addresses common public fears. Because sponsors shoulder much of the cost and responsibility, perceived fiscal burdens are lower. Because sponsorship groups tend to be deeply involved in helping refugees they sponsor—finding housing, connecting newcomers to schools and jobs—social cohesion and integration outcomes should be stronger. While no randomized trials exist, observational studies generally find better integration outcomes in employment and income for privately sponsored individuals compared to government-assisted refugees, which is only partly explained by selection bias. A recent study by the Canadian government found that after one year 75% of privately sponsored refugees had employment earnings vs 37% of government-assisted, and social assistance receipt was 16% vs 93%, with advantages persisting over several years.

To my surprise, however, despite nearly half a century of Canada’s private sponsorship program and its recent global proliferation, direct public opinion evidence on the topic is scant. The only report I was able to find on public attitudes and sponsorship found high support but mostly relied on indirect or qualitative evidence (e.g., more positive general immigration attitudes among people who have participated or live in high-sponsorship areas). After further investigation, which took me much longer than I care to admit, I was able to locate a few relevant surveys that straightforwardly ask people about their support for sponsorship programs.

Here are the key reports and their highlights:

• In Canada, a vast majority are aware of the private resettlement program (which is impressive given the generally low political knowledge in public opinion). A clear majority—especially those who are aware—view it favorably. According to the 2018 and 2021 Environics surveys, about 3-7% say they have been directly involved, 15–25% say they personally know a sponsor, and about the same share say they would like to participate in the future. A 2017 McGill survey, which explicitly asked whether private sponsorship or government resettlement works better, found that significantly more respondents chose the former (41% vs. 6%, with the rest unsure).

• A 2021 Environics poll found that, among the small minority who view private sponsorship negatively (13-16%), reasons cluster around how the program is administered (taxpayer burden, insufficient resources) or unfavorable views of refugees (concerns about integration or competition for resources). Although these skeptics were not asked about other policies, it is reasonable to assume they have similar or stronger concerns about traditional government resettlement.

• In Germany, a 2016 More in Common survey conducted during the Syrian crisis found 45% in favor of introducing a sponsorship program, with about one-third opposed. These levels exceeded general positivity toward “refugees” at the time. Forty percent also reported donating or volunteering to help refugees, and 22% said they would be willing to participate in a sponsorship program.

• In the United Kingdom, a 2021 More in Common survey found 48% support and 34% opposition to accepting more (Afghan) refugees via community sponsorship. Net support was 14 points higher than for general resettlement, driven mainly by lower opposition among socially conservative and anti-immigration segments of the population.

• In Poland, a 2024 CMR Ipsos survey found 31-39% support for introducing a sponsorship program—the only case I saw where opposition exceeded support somewhat. Even so, community sponsorship was more popular than traditional government-led resettlement. An earlier survey from May 2022, fielded shortly after the start of the war in Ukraine by the same research team, reported much higher support numbers.

• In the United States, a 2023 YouGov survey at the launch of Welcome Corps showed 60% overall support, including 76% of Democrats and 53% of Republicans. Given heightened border salience and a thermostatic cooling on immigration during the Biden administration, these are notable figures. About one in four Americans also expressed interest in personally sponsoring a refugee in the coming years.

Bottom line: Community sponsorship is broadly popular—either absolute majorities or strong pluralities support it across various contexts—and it tends to outscore government-only resettlement and many other humanitarian policies.

Why hasn’t it caught on more? The major bottlenecks and limits of sponsorship

If community sponsorship works so well, why hasn’t it spread more widely? Despite my constant reminder to immigration advocates that humanitarian intentions are rarer than they assume, my sense is that the answer is probably not a lack of willing citizens. The Canadian surveys I described earlier show that a small but meaningful share of the public already participated (about 3-7%), or would like to participate (another 5–15%) if given the chance. This aligns well with my own surveys and incentivized experiments: while most people understandably prioritize their own or their country’s well-being, at least 10% in rich democracies display pronounced humanitarian motivations and are willing to benefit foreigners even at a personal cost. Even if we take a very conservative cap of 5% of the working-age population as the potential pool of sponsors, that is still a large number. Extrapolated to the United States and other rich democracies, this implies millions of potential sponsors. In short, public enthusiasm seems sufficient.

The bottleneck is the government’s resolve and capacity. Policy innovation in immigration is slow, especially when leaders want clear proof of success before scaling. Even Canada’s famous points-based skilled migration system took years to become a global practice. And sponsorship requires more than goodwill—it demands real administrative capacity.

Governments must vet sponsors, screen refugees, issue visas, arrange travel, monitor cases, and step in if failures occur. Many countries lack the bureaucratic infrastructure or trust in civil society to manage this. The start-up costs of building sponsor networks, training groups, and supporting them through the process are significant. Philanthropic seed funding has increased recently but remains modest, and officials rarely see enough upside to overcome inertia.

Even if some of these bottlenecks ease, community sponsorship is clearly not going to solve the world’s displacement crises alone. There are over 35 million refugees worldwide, with 2–3 million designated as urgent resettlement cases, and each year, only a fraction are resettled anywhere. If every wealthy nation decided to adopt the Canadian sponsorship model tomorrow, total numbers would still be in the hundreds of thousands per year, not millions.

Moreover, community sponsorship does not address the messy, politically toxic issue of spontaneous border crossings and asylum claims. Sponsorship is simply not designed for these scenarios—it is orderly and selective, which is the opposite of chaotic inflows. Some development economists—and now even The Economist—argue the asylum system is outdated and should be rebuilt around protection and legal work in proximate host countries, fewer camps, and more legal pathways with regional processing to deter dangerous journeys. In that reimagined setup, community sponsorship could serve as one of the channels to redirect some of the would-be asylum seekers into managed programs supported by citizens. But realizing this would require policy shifts far beyond sponsorship itself.

Interlude: the successful, yet short-lived, case of U.S. Welcome Corps

The recent two-year U.S. sponsorship experience illustrates both the appeal and fragility of sponsorship. The Welcome Corps, launched in 2023 as a pilot within the federal refugee admissions program, invited Americans to form groups and directly sponsor refugees for the first time. Several observers even called it a “revolution” in the U.S. refugee admissions or even immigration policy in general. The response was remarkable: more than 160,000 people across every state registered interest within two years. The most engaged states ranged from Minnesota and California to Texas and Indiana, showing geographic and political diversity.

Public opinion matched this enthusiasm. A YouGov poll found 60% of Americans supported the idea, including 76% of Democrats and 53% of Republicans. For a pro-immigration policy initiated by a Democratic administration to secure majority Republican support in 2023 was striking.

At the same time, the program did not generate any evident backlash. Some anti-immigration groups raised alarms about potential fraud and weaker vetting (which they do about pretty much all immigration programs), but a review by the Niskanen Center found those concerns unsubstantiated. Refugees underwent the same security screening as in other resettlement channels, and sponsors themselves were background-checked and trained. No major scandals occurred: refugees were vetted, sponsors were supported by intermediary nonprofits, and cases proceeded smoothly.

The U.S. case demonstrates the political potential of sponsorship: grassroots enthusiasm, broad partisan reach, and no visible backlash. It is not proof of long-term success, but it shows how strongly the model resonates with American civic culture. The Welcome Corps ended only because refugee admissions overall were paused by the second Trump administration in early 2025—not because of any explicit opposition to the program itself. If and when revived, it would likely continue to draw bipartisan interest.

Learning from Canada, the U.S. pilot, and other countries, we can try to identify some key design principles that make a community sponsorship program both more sustainable and scalable, from rigorous participant vetting to well-funded administration. I will write about these, as well as possible extensions to the program, in a separate post in the future. For now, I want to highlight two features that I find especially important for the program’s political success (notably absent in the initial version of the U.S. Welcome Corps program): naming and additionality.

Naming and additionality: the key sponsorship principles and the debates around them

As with any reasonable policy compromise, community sponsorship programs and their key principles have also been debated and criticized on both the left and the right. Let’s start with the aforementioned naming principle, which essentially allows sponsors in Canada to pick specific refugees (at least among those who qualify for resettlement by law). This principle raises obvious fairness questions: Are those refugees the neediest, or just the best connected? These concerns have led some left-leaning analysts to criticize the naming feature of private sponsorship as inequitable, since it tends to prioritize refugees who have family or friends abroad.

Although I have found relatively little explicit criticism from the Canadian right focused on the program itself, the concerns I did find are almost a mirror image. In particular, some worry that private sponsorship could become a sneaky backdoor for increasing lower-skilled immigration, in relative or absolute terms. Because sponsors usually name their relatives or co-ethnic friends, the program might be used to bring in people who would not qualify under stricter points-based streams. The most troubling aspect for these critics is that sponsorship leads to permanent resettlement, meaning those brought in—and their descendants—may draw on taxpayer-funded benefits if they contribute less in taxes than they consume. Given Sweden’s disappointing experience in improving fiscal outcomes for humanitarian migrants and their families despite strong integration efforts, this critique should not be easily dismissed.

As some rightly argue, however, one of the Canadian program’s strengths compared to its many offshoots is precisely that sponsors are allowed (though not required) to nominate specific refugees. Equity and human capital concerns aside, naming taps into the strongest motivations people have to sponsor in the first place. Individuals and groups are more committed when the person they welcome is not a stranger, but someone they already know, or someone with whom they feel a direct cultural or religious connection. Prior relationships often bring shared language and customs, which can ease integration. Besides, sponsors can also nominate people they do not already know, enabling creative uses such as campus sponsorship for refugee students or partnerships focused on sexual and gender minority refugees.

At the same time, matching-only streams like the aforementioned BVOR program have struggled to mobilize and retain large numbers of sponsors. After completing a matched case, many groups end up seeking channels that let them name specific people to help their relatives or friends. The U.S. Welcome Corps, for example, saw faster uptake after adding a possibility for naming in the second phase of the program, underscoring how the ability to nominate specific people can drive participation. In short, naming makes the program work politically by sustaining civic engagement over decades, even if it may complicate the purist ideals of impartial humanitarian protection or skill selection.

But the most serious structural critique of the program relates to the additionality principle or the lack thereof. Does sponsorship actually increase protection for vulnerable people, or does it substitute for government action? In 1979, when the program started during the Indochinese resettlement, the federal government made an explicit one-for-one pledge (one government-assisted admission for each privately sponsored case). The pledge was discontinued soon after as backlogs grew. Today, the government sets separate targets for government-assisted and privately sponsored streams, and allocations can shift between them from year to year.

This raises the familiar “crowd-out” concern: if volunteers sponsor 10,000 refugees, a cost-conscious government might reduce its own intake by a similar amount, yielding no net increase. The risk is debated and hard to prove, but in some years PSR admissions have exceeded GARs, which sponsors cited as contradicting their additionality expectations—even though additionality is not part of the PSR official program theory anymore.

From a political perspective, however, even a pure substitution arguably can have an upside: if taxpayers see that enthusiastic citizens are handling more refugees, it might reduce backlash and keep overall support higher than if the government tried to do it all. It also effectively addresses the most salient conservative critiques of the program. Still, for community sponsorship to reach its full potential, it ideally needs to complement, even imperfectly, not fully supplement, government resettlement.

Clear government commitments can prevent this—whether through pledges that private sponsorship will not reduce overall quotas, or even through formulas that increase official resettlement proportionally. Transparency is also essential: if citizens can see that their efforts genuinely expand the total number of refugees welcomed, more will step up. Creative mechanisms could reinforce this link, such as tying sponsor contributions directly to funding additional government-assisted arrivals. However it is achieved, additionality, even when it is only partial, is the key to unlocking sponsorship’s promise: mobilizing private compassion to help vulnerable populations beyond voting or charitable donations.

So, how can sponsorship make our immigration politics better?

Despite current challenges and limits, I believe that community sponsorship of refugees has a bright future. Its track record in Canada shows it can make refugee resettlement more popular and politically sustainable, even where traditional humanitarian policies face hostility. Programs that empower citizens to welcome refugees consistently score higher approval than almost any other immigration initiative. They harness grassroots goodwill that would otherwise go untapped. And they tangibly benefit not only refugees, who get a chance at a new life in a supportive environment, but also hosts, who often find new purpose and social ties, and their communities, which gain integrated workers amid population decline.

In an age of polarized politics, community sponsorship uniquely appeals to a broad demographic and manages to bring unlikely allies together—church groups and LGBT nonprofits, veteran groups and humanitarian agencies, liberals and conservatives, small towns and big cities. This coalition-building effect is invaluable for the long-term sustainability of refugee protection. It is much harder to demonize “refugees” in the abstract when your neighbors, co-workers, or your parents’ church are personally helping someone settle nearby.

In the more immediate future, we are likely to see continued scaling up country by country. The new Global Refugee Sponsorship Initiative has been advising governments, and roughly 14 countries have launched some version since 2016. Most remain pilot-sized and have resettled only a few thousand families, though separate family-reunion pathways for Ukrainians brought in tens of thousands under a similar sponsorship principle.

The real game-changer would be the U.S. fully embracing community sponsorship with naming alongside its government program. If the U.S. activated hundreds of thousands—if not millions—of willing sponsors, or even reached Canadian-level per capita rates, we could be talking about hundreds of thousands of refugees resettled annually via private means. Even if those numbers are aspirational, they illustrate significant untapped capacity. High-income countries collectively host only a small fraction of the world’s refugees today, but by empowering their own citizens to sponsor refugees, they could increase that share in a politically sustainable way.

Community sponsorship will not solve the refugee crisis on its own, and it will not replace the need for robust government action and international cooperation. But it will give tens of thousands of people a safe new home who would not have had one otherwise. In a world where so much of the immigration debate is abstract and mistrustful, community sponsorship offers a concrete and intuitively positive story: regular people working together on something compassionate and constructive, with visible results that many can admire even if they choose not to participate. That is a useful antidote to cynicism and a reason to think that, while community sponsorship may not transform global numbers overnight, it can improve our immigration politics over the long run, making it more open and humane by design.

Share

Many thanks to Gabriella D’Avino, Ania Kwadrans, Biftu Yousuf, and BBI fellows for their help and comments on this piece.

Select bibliography

• Global Refugee Sponsorship Initiative

• Fratzke, S., Kainz, L., Beirens, H., Dorst, E., & Bolter, J. (2019). Refugee sponsorship programmes: A global state of play and opportunities for investment. Migration Policy Institute.

• Labman, S., & Cameron, G. (Eds.). (2020). Strangers to neighbours: Refugee sponsorship in context (Vol. 3). McGill-Queen’s Press.

• Cameron, G. (2021). Send them here: Religion, politics, and refugee resettlement in North America (Vol. 5). McGill-Queen’s Press.

• Kaida, L., Hou, F., & Stick, M. (2020). The long-term economic integration of resettled refugees in Canada: A comparison of privately sponsored refugees and government-assisted refugees. Journal of Ethnic and Migration Studies, 46(9), 1687-1708.

• Labman, S. (2016). Private sponsorship: Complementary or conflicting interests?. Refuge, 32, 67.

• Manks, M., Monsef, M., & Wagner, D. (2022). Sponsorship in the Context of Complementary Pathways. Knowledge Briefs.

• Sustainable Practices of Integration (SPRING) podcast about Canada’s sponsorship.

• Private Refugee Sponsorship in Canada - 2021 Market Study, Environics Institute.

• Orth, T. (2023). Most Americans support “Welcome Corps,” Biden’s new refugee sponsorship program. YouGov.

Please like, share, comment, and subscribe. It helps grow the newsletter without a financial contribution on your part. Thank you for reading.

In Part I, I argued that AI can already do social science research better than most professors. In Part II, I engaged with over a thousand responses, conceding where critics were right, while standing by my main claim: the academic status quo was already broken, and AI is just forcing the reckoning.1 In this Part III, written collaboratively with AI and my peers over the last month, I move from diagnosis to what academics can and can't actually do about it.

The rather unlikely proximate cause of this third installment on AI was visiting the 2026 International Studies Association (ISA) Annual Convention in Columbus, Ohio—a preeminent multidisciplinary conference of the world’s leading international studies professionals. Or so I was told. What I actually witnessed were presentations so rough they would barely get a C in any of my classes: arguments with no thesis or coherence, grammar errors any spell-checker would catch, presenters reading off their slides as if encountering their own bad arguments for the first time. All without any AI involved, as far as I could tell, judging by the presence of typos and inconsistencies. These were not just grad students, but people with PhDs, tenure, and research budgets.

If AI slop is the crisis everyone warns about, I’d like to know what to call what I saw at ISA or most other big social science conferences, for that matter.2 The contrast was impossible to ignore: I was sitting through these presentations at precisely the moment I was receiving death threats and calls to fire me online for suggesting AI can do research better than most professors. That juxtaposition crystallized the argument for this piece.

Subscribe now

21. Most “slop” has always been and still is human slop.

My first thesis was the most provocative thing I’ve said, and I’ve adjusted it only slightly since then: agentic AI can already do most social science research tasks better than most professors globally. I still stand by it. In my recent interview with the Chronicle, they put it more bluntly: “AI Is a Better Researcher Than You.” If you still don’t believe that’s true, let’s talk in a few years.

But the flip side is just as important. If AI can produce better research output than professors, that’s also an indictment of the output those professors were and are still producing without AI.

“Slop” was Merriam-Webster’s 2025 Word of the Year, defined as low-quality digital content produced by AI. But the ISA conference was a reminder that the vast majority of slop has always been human slop. The academic journal system and big conferences in much of the humanities and social sciences were slop factories long before anyone had a ChatGPT subscription. Yes, I really mean that most research is slop.3

Some of it is also what the philosopher Harry Frankfurt would call “bullshit“: work that is indifferent to whether its claims are true, especially on politically charged topics like immigration, where researchers start with the left-wing conclusion and work backward. But slop is broader than bullshit. It also includes work that makes no claim at all, work that is supposed to have craft value, and simply fails. The researcher who finds a dataset before having a question, then dredges for significant results worth publishing, is producing slop. These researchers existed long before AI. They were just slower.

22. Academics were hallucinating and cheating before AI.

The concerns about hallucination and cheating that people raise about AI in academia describe problems that predate it by decades, if not centuries. A massive new project published in Nature this month (which I was a small part of) tested hundreds of social science papers: only about half of statistically significant claims replicated, with median effect sizes shrinking dramatically. This study confirms what the 2015 Open Science Collaboration study found in psychology alone, where roughly two-thirds of findings failed to replicate, and extends it across fields.

We didn’t call these “hallucinations” before AI. We called them “science.” If you think about it, hallucination and inspiration are actually not as different as they sound. Both involve generating combinations that go beyond the input. We call it hallucination when the result is wrong and a breakthrough when it’s right.

Meanwhile, academics routinely cite papers they haven’t read beyond the abstract.4 At least AI hallucination rates are tracked and improving. Human hallucination rates in academia are not tracked at all. We just call them “contributions to the literature.” And if you’re a peer reviewer, you don’t even have to hallucinate on your own: you just write “please cite me” and move on.

Some research is genuinely excellent. But before we worry about AI-assisted cheating, let’s reckon with the human kind for a moment. Diederik Stapel, Marc Hauser, Francesca Gino, Dan Ariely: the list of high-profile fraud cases keeps growing, and these are just the people who were caught. Aside from outright fabrication, p-hacking, HARKing (hypothesizing after results are known), and selective reporting were for years so common that they barely registered as misconduct. We’ve made real progress in understanding these practices, but they remain widespread enough to shape what gets published. Beyond all this, senior professors have always put their names on papers written primarily by graduate students, and entire books have been assembled by teams of research assistants working under a famous scholar’s name. None of this was considered cheating until AI made the process cheaper and more apparent.

23. The “stochastic parrot” metaphor describes humans better than AI.

One of the most influential slogans in the AI debate5 has always functioned as a thought-terminating cliche. As Cate Hall observed, it is a potent coinage: fun to say, conceptually efficient, and it has permanently colonized many people’s minds despite not being true of today’s models. A genuine linguistic work of art. It is also empirically false: every major frontier model since GPT-4 has been trained on non-textual input, and the original argument’s own logic requires text-only training to work.

Now consider something completely different yet still very familiar: the “In This House We Believe” yard signs. Science Is Real. Love Is Love. No Human Is Illegal, etc. You can believe every line and still have no coherent policy position on any of these issues. What does “no human is illegal” imply for enforcement policy: open borders, amnesty, something else? The sign doesn’t say, because saying would require confronting contradictions. It is a loyalty oath, not an argument.

But the deeper point is that this is exactly what AI gets criticized for: producing shallow, feel-good statements that communicate belonging rather than meaning. Turns out humans have been doing that with or without AI for as long as we’ve had lawns to put signs on.

The yard sign is not an outlier. It is how most people seem to engage with most issues: adopt the position of your group, repeat it, move on. The substance of what you argue matters far more than how you frame it, but framing is easier, which is why it dominates public discourse, from Twitter threads to conference presentations.

I saw this firsthand at ISA, where tenured peers presented work that amounted to the academic equivalent of a yard sign, complete with a few misidentified regressions to give it the veneer of science. The stochastic parrot critique was meant to diminish AI. It ended up being a better description of human intellectual life than anyone intended.

24. Yes, consenting adults can use AI for writing. Policing it doesn’t work.

In Part II, I noted that AI detectors are often not very useful and create more problems than they solve. But the deeper problem is the impulse behind them: the belief that AI involvement is inherently contaminating, regardless of what it produces. Quinn Que makes a fascinating case that the obsession with AI writing detectors is akin to enforcing a “one-drop rule,” the principle from 19th-century American racial classification: any trace of AI involvement contaminates the entire work, regardless of quality or the author’s intent.

I was initially skeptical of the analogy, but it’s quite right. In the view of the anti-AI activist, any word you have not written yourself is a moral pollution. Although using AI for writing is not technically “illegal,” there is basically a one-drop rule governing whether you are a legitimate writer or a fraud, a good person or a bad one. Hence, the condemnations and death threats to people like me who disclose their AI usage for “outsourcing their thinking to the machine.”

As I argued in Part I, much of the opposition to AI is status protection dressed up as principle. takes this further, arguing that the chatbot moral panic may have a source beyond the status project. This source is something closer to superstition (”chatbots are demonic”): the sense that AI-produced text is spiritually tainted, that there is something wrong or even evil about a machine that can write, regardless of what it writes.6

Even , who recently honorably disclosed her AI use and sparked among journalists the same conversation academics had been having on Bluesky a few months earlier, felt compelled to defend herself by insisting that “AI didn’t touch copy.” I admire her for speaking up. But why should the copy question even be an issue? If the work is good and the process is disclosed, the rest is aesthetic preference dressed up as ethics. Where does impurity start? Google search? Auto-correct? Spell check? Transcriptions?

And even setting all that aside: barring some Dune-style scenario of an AI catastrophe followed by humanity coordinating to ban the technology, widespread AI-assisted non-fiction writing is almost inevitable in equilibrium, given the existing incentives. The flip side worth mentioning is that even if you do manage to write well entirely on your own without “AI pollution,” no one is going to be rewarding you for that purity very soon, either.

25. Not using the latest AI tools in your research and writing is malpractice.

recently argued that large language models are underused in journalism. His point applies to academics, too: the purpose of research is the useful output, not the human-mediated process. The rigor should be in the thinking and the verification, not in whether a human or a machine typed the sentences or pressed enter when running the regressions in R. As Hollis Robbins put it well, professors should probably be testing AI models before breakfast (basically be like ).

The most mundane use of AI is catching errors. Consider the New York Times headline on April 3, 2026: “A North American Treaty Organization Without America?” The correct name is, of course, the North Atlantic Treaty Organization. The Times ran a correction the next day. Some speculated the error was caused by AI. We’ll never know, and frankly, it doesn’t matter. Any sufficiently smart LLM workflow would have caught the mistake in seconds. A simple automatic routine for all headlines—”Does this contain any errors? Verify against two separate sources”—would have saved the Times a news cycle of embarrassment.

But fact-checking is the floor, not the ceiling. The more interesting use is doing things that weren’t possible before. My now co-author, , recently had Codex build her an interactive website to help her actually understand a political science paper she was reviewing, then did the task herself the way the study’s participants had. AI doesn’t just compress the time it takes to produce an output. It lowers the cost of the kind of active engagement most researchers skip: building the thing yourself, stress-testing an argument, rerunning an analysis the way participants did. That’s what deeper understanding actually requires.

The same applies to academia. Half the ISA presentations I attended in Columbus could have been meaningfully improved by a cursory loop through ChatGPT: checking grammar, tightening arguments, catching logical gaps. These tools are free or nearly free. Choosing not to use them is a choice to deliver worse work than you’re capable of, especially if your calling is to inform the public. But journalists and academics alike should not use AI just to catch typos in headlines. They should use it to build interactive visualizations, stress-test arguments, and genuinely understand the complicated things they are writing about.

But what about writing itself? You may have seen a common response to anyone using AI in their writing: “Why should I read the article and not the prompt that caused it?” Sounds reasonable, right? Well, no, if you think about it a bit.

Why eat the meal instead of reading the grocery list? Why look at a chart instead of the ggplot script? Why read the book instead of the author’s notes? The prompt gives you no new knowledge. The output does. That’s the whole point. As I argued in Part II, different people with different expertise, data, and context (as manifested in their claude.md files) produce entirely different outputs from the same prompt. It’s a skill issue.

I get it: it can feel strange to spend more time reading something than the author spent producing it. But we’ve been here before. A chart that took seconds to generate in R can take minutes to read carefully. Nobody demands or fantasizes about seeing the code only instead of the chart. They evaluate the chart and what it says—the new knowledge not available before.

26. LLMs may indeed produce new knowledge.

As recently documented by , AI chatbots consistently pulled users toward expert consensus, the opposite of what social media does. The tools don’t just help you write better. Used actively, they help you think more carefully.7 This way, LLMs may even reverse the rise of populism. But fact-checking is not the same thing as producing new knowledge.

Before we ask whether LLMs can produce genuinely new ideas (there are good examples of that), we should ask how many humans do. Most academics spend their careers recombining existing ideas in minor variations, applying the same methods to slightly different datasets, producing incremental work that nobody outside their subdiscipline will ever read. Original thinking is extraordinarily difficult and rare in any generation. I don’t mean it as an insult to any of my colleagues (or myself). But the bar that LLMs need to clear is lower than we like to admit.8

My friend Robert Kubinec, a brilliant political methodologist and published fiction author9 at the University of South Carolina, seems to be pretty skeptical about much of the AI hype. He argues that “LLMs never create knowledge, [which] only exists in human brains. They can only compare one set of knowledge to another.” I respect Bob, so it’s OK if we disagree. My response: The philosophical question about self-awareness is real, but it’s orthogonal to the practical one. Whether or not the model “understands” anything, the output either contains new information useful to humans or it doesn’t.10

The most suggestive recent example is Claude Mythos, Anthropic’s new frontier model. In a few weeks of testing, it flagged thousands of previously unknown security vulnerabilities across major operating systems and browsers, including one that had gone undetected for 27 years. Whatever you want to call that, “compare one set of knowledge to another” does not quite cover it.

The best conceptual theories in social science are already recombinations of existing ideas. Anthony Downs built his rational choice theory by transplanting microeconomic utility maximization into voting and literally called his book An Economic Theory of Democracy. Baumgartner and Jones borrowed punctuated equilibrium from evolutionary biology and applied it to policymaking. Social network analysis imported graph theory wholesale from mathematics. Axelrod fused the prisoner’s dilemma with evolutionary fitness selection. Alexander Wendt went further and applied quantum theory to international relations in Quantum Mind and Social Science—a move many found (rightly) ridiculous, but which Cambridge University Press still published. Social science rarely invents from scratch. It translates across domains, and the translation is the theoretical contribution.

That is structurally identical to what LLMs do: recombine patterns and concepts across contexts. Sometimes the result is nonsense. Sometimes it is productive. The same is true of human recombination. Wendt’s quantum IR has been criticized as a mere metaphor masquerading as physics, but as mentioned above, Cambridge University Press published it. If that counts as knowledge production, it is hard to see why LLM-generated recombinations wouldn’t.

To his credit, Bob conceded the practical point even while holding the philosophical one: “the question is how to use this capability to advance knowledge.” Exactly. As I said in Part I, stop debating whether LLMs “truly understand” while the people with the most at stake are already using the tools to improve their work.

27. For critics, the mental model of an AI user is stuck in 2023, which is ages ago.

Remember how we got into this mess. Students got enthusiastic about using AI before their professors, using the very first public, free models like ChatGPT 3.5. The result is that many professors and intellectuals still picture the typical AI user as an undergraduate trying to cheat on an essay or their assistant failing by submitting AI slop. That may still describe a lot of casual AI use around the world. It does not describe how top researchers in most fields are working with these tools, or how anyone trying to use AI responsibly and mindfully actually operates.

put his finger on this the best: we underrate other people’s rationality, assuming they apply new tools much more mindlessly than they really do. There is a vast space between asking ChatGPT to write a whole paper and writing down every single word yourself. When you write “yourself,” you are already using shortcuts. You are not doing deep research on every reference you make. You Google a statistic and trust the headline number. You skim an abstract and cite the paper. If you have resources, you outsource verification to a research assistant or a fact checker. AI lets you do much of this in a more systematic, automated, and cheaper way.

Like many, rightfully believes that writing is thinking and that outsourcing the full writing process to AI leaves your mind empty. But, while much writing is thinking, thinking is not only writing. Making art is thinking. Talking is thinking, too. As Dina Pisareva argues, prompting is also thinking, if you have a good Socratic partner on the other side of the exchange.

Thompson also acknowledges that all writing has always involved reaching outside the writer’s mind for ideas, facts, editing, and fact-checking. The line between legitimate assistance and illegitimate outsourcing has always been blurry, and AI didn’t create that blur. It just made it impossible to ignore. Thompson’s own conclusion: “We should be honest and open about the blur rather than declare everybody with an open Claude window a part of the slopclass.”

Where does that leave disclosure? Many good folks and top academic journals suggest complete AI-use disclosure as a solution. But disclosure is not sustainable in a game-theoretic sense: disclosers bear reputational costs while secret users free-ride, so the equilibrium pushes toward non-disclosure. People like Megan McArdle and I are all living proof of this problem. And the better framing, as Ryan Briggs puts it, is that AI complements expertise: automated RAs checking your work, formalizing arguments as you go, gathering data on demand. It’s a multiplier that lets capable people think better and faster.

My sense is that the only case where disclosure is genuinely owed is when the audience has a reasonable expectation of fully human-produced work, and that expectation is part of what they’re paying for. The best analogy is a live concert in the narrow case of personal or creative writing: if you show up expecting a live performance and the artist is lip-syncing, that’s a legitimate grievance. More precisely, disclosure is owed when non-disclosure would mislead the audience about what they are getting. A journalist publishing under a byline is promising accountability and originality, not keystroke provenance. A memoirist is promising both. The test is what the author is putting their name on, not whether a tool was involved.

But science and journalism are not live shows. They’re about discovering and sharing new knowledge. Nobody discloses that they used spell-check, research assistants, or a Google search. The norms for creative writing and art will be different, rightly so, because audiences there are partly paying for the human experience of making the thing. Humanities will probably land somewhere in between through some lengthy and contentious process, with many friendships being broken. But for research and journalism, “disclose everything” doesn’t get us where we need to go. What readers should care about is accuracy and whether the author takes full responsibility for what’s on the page. Provenance is a much weaker signal than either of them.

28. Nobody knows anything, myself included. That’s OK—we’ll figure it out together.

I initially shied away from talking to journalists about AI because I’m not an expert on it in any meaningful sense. But I’m increasingly realizing that nobody is an expert on whatever this all is, not really, and waiting for one to appear is how you end up with no norms at all. I still shy away from giving any advice beyond installing Claude Code (or your agentic tool of choice) and talking to it to solve a problem you have at hand.

As usual, Arthur Spirling was blunt on this point: the AI-academic accounts that constantly offer “advice” to colleagues and PhD students, as if they were in the labs developing the models, are tedious. They’re spectators, like the rest of us.11 Even the people in the labs have no idea how good the models will get beyond a few months.

So we’re in an awkward position: somebody has to start establishing the norms of the new workflow environment, and the people doing it will inevitably be people who don’t fully understand what they’re building norms for. That’s weird and uncomfortable. It’s also how most other institutional transitions have probably worked, though at a much slower pace.

We went through a version of this that seems like ages ago, when professors had to accept that policing students’ AI use fully and meticulously was neither productive nor possible. The same is now true for researchers. You can audit the parts that actually matter, like the replication code, and LLMs already make it easier. Provenance is the one thing you can’t audit, and it’s the thing we shouldn’t be spending energy on anyway. You have to follow the incentives and build systems that ensure the tools are used well, not pretend they aren’t being used at all.

On a related note regarding teaching, I’ve been on leave, so I’ve shied away from advice here too. Given the many questions I’ve received, I’m more pessimistic about the effects of AI on teaching than on research. For instance, while I use AI openly and extensively in my research and writing, I plan to ban all electronic devices in my substantive classes and bring back in-person written and oral exams.

These are not contradictory positions. The classroom is precisely where students need to build the cognitive skills that make AI collaboration productive later. You cannot meaningfully direct an AI tool if you never learned to think without one. The skill-atrophy concern from Part II is most valid here: students need to internalize the fundamentals before they outsource them. Research is where you deploy the best tools at your disposal to produce a worthy outcome. Teaching is where you make sure the next generation can actually use those tools well.12

29. The best work happens when humans and AI collaborate.

I still see many folks pledging not to use AI in their writing. This is about as sensible as pledging not to accept help from research assistants or co-authors. So, I pledge the opposite: I will use the latest LLMs, and for that matter any other available tool or human co-author, to best improve my research or the way I communicate it. That way, if my name is on it, you can be sure it reflects my own best judgment.

Many academics, especially in the humanities, still seem to believe that it matters how much time you spend on doing your work. But the folk labor theory of value is wrong in economics, and it is wrong here as well. If you’re a weirdo who wants to spend years running all your robustness checks manually with matrices inverted by hand instead of running a few R commands or manually translating or transcribing original manuscripts, that’s on you. I didn’t care much if those international studies professors at ISA spent 10 or 100 hours on their human slop presentations. The work is either good or it isn’t.

That said, there are some norms worth establishing. When you write “I believe” or “I feel,” that first person should genuinely be you. The pronoun “I” carries an implicit promise of a human voice behind it. A factual claim, like the correct name of NATO, doesn’t care who typed it. But personal conviction does. Think of a live concert: the audience pays for a real performance, not a lip-sync. When you use “I,” the reader is entitled to expect that you mean it. That doesn’t require literally typing (or dictating) every word yourself. You can direct the model, work from your notes, and read it over carefully, but it does require that the conviction is yours.

This piece, for instance, was openly and proudly written using an iterative back-and-forth conversation with Claude Code, based on my original ideas, conversations with peers, and useful suggestions by human academic and non-academic colleagues.13 Part I experimented with AI capabilities with no human editing. Part II reflected with 100% human voice. Part III, which you’re reading now, combines my voice with AI capabilities, completing the circle. I’ll let you decide which one was best.

So, where do humans retain an edge? As Yiqing Xu argues, probably in open-ended environments where training data doesn’t yet exist and in tasks that require direct human interaction. This may change soon. But we chose this career because we find meaning in figuring things out, not just because we’re still better at it. That’s why I don’t see any contradiction in outsourcing most grunt work to AI (or RAs, if you hold traditional values) and focusing on what gives us the most meaning—coming up with important questions about the world and answering them with the best tools available.

30. The real risk in science is human slop at AI speed. We can still prevent it.

AI amplifies whatever you bring to it. Bring genuine curiosity and hard questions, and AI will help you produce something worth reading. Bring nothing, and you’ll produce nothing faster.

But there’s reason for optimism. The same Nature findings that revealed how broken reproducibility is also showed that journals with mandatory data and code sharing had meaningfully higher reproducibility rates. AI can already accelerate this: nothing but inertia precludes journal editors from instituting automated reproducibility checks and mandatory code verification as a condition of submission. These systems can be set up as quasi-deterministic, with very little room for error, and authors can always challenge an automatic desk rejection if they believe the in-house agentic workflow failed to determine how to reproduce their code.

More ambitiously, AI lowers the cost of doing the kind of large-scale, data-intensive work that used to require being Raj Chetty with a team of 30 research assistants and access to every administrative dataset in the country, or being Daron Acemoglu… with pencil and paper. A junior scholar with Codex can now attempt projects that would have been logistically impossible five years ago. The bar for what gets published should rise, because the bar for what can be produced already has.

The standard hasn’t changed: if you put your name on something, you stand by it. Judge the quality of the output, not the process. The conversation about AI slop is a distraction from the harder question, one we should have been asking long before the chatbots arrived: why have we been tolerating so much human slop in the first place?

***

My goal when I posted Part I was simple: to bring the conversation that was already happening behind closed doors and in DMs into the open. Still, I keep hearing the same thing from my colleagues: “Alex, we know you use AI, we do it too, but can’t you just be quiet about it?”

I understand the fear that AI will be used irresponsibly, producing more slop than it prevents. But if you think about it, “be quiet about it” is just advocating for universal hypocrisy as a professional norm. Part of what I wanted was for the uncertainty everyone is experiencing about their workflows and their futures to reach a more stable equilibrium. That requires honest conversation, not silence.

And it is already happening. Emily Oster recently shared Isaiah Andrews’ advice on AI for MIT PhD students in economics, calling it something that should be circulated to all PhD cohorts. As noted, the most important thing about it was not any particular advice per se but having high-profile faculty send a clear signal: this is something you need to take seriously. Even , one of the most thoughtful people in journalism, recently admitted the AI people have been right a lot.

The change is real. Academics are already half-awake, and they are not going back. Aspiring grad students and junior scholars should tread carefully but embrace learning about and using AI tools fully. The scholars, intellectuals, and writers who refuse to engage will not be rewarded for their purity. They will simply be outperformed by colleagues who bring the same intellectual rigor, the same curiosity, and better tools.

Share

Subscribe now

Here’s where I think the argument turns:

Many of my academic colleagues felt differently, however. To them, “I just oppose illegal immigration” was a socially acceptable way to express opposition to foreigners in general—xenophobia dressed in procedural language. I have been guilty of dismissing popular attitudes myself; when giving lectures about, say, the H-1B visa backlog or refugee processing times, I have found myself exasperated by audience members who stand up to ask why I haven’t gone out of my way to condemn illegal immigration.

Over the years, many scholars and advocates thus came to see the distinction as illegitimate. For some, the American immigration system is already so unfair and restrictive—fewer than 1 percent of people who want to immigrate can do so legally—that saying “Just follow the rules” can feel cruel. Others go further: Unauthorized border crossing is a victimless regulatory violation, they argue, and any law restricting people’s free movement is unjust.

I now think this dismissal was a mistake. The norm was imperfect and often insincere—but it was holding something together. Now that it’s gone, I’ll try to be a little less annoyed the next time someone in my audience stands up to ask why I haven’t condemned illegal immigration. Turns out that was the good version of the conversation after all. Read the full piece here (archive link).

Share

Some personal news first. I finally bought a Japanese toilet. A TOTO Nexus WASHLET+ S7A, to be specific. If you read my piece on Japan last year, you know I came back radicalized about how the Japanese make familiar things work better. The toilet is one of them. Heated seat, genuinely life-improving in ways it’s not possible to describe in polite company. It’s also arguably a progress and a policy design issue just like the shower heads: the technology has existed for decades, it’s demonstrably beneficial, and most Americans would love it if they tried it. The main barrier is outdated plumbing and electric codes, and the fact that nobody in power has bothered to update regulations that would make installation easier and cheaper. Sound familiar?

Subscribe now

Despite spending too much time on social media and receiving a fair share of death threats for my hot AI takes (I had to deactivate my Bluesky account), March was, by a wide margin, the most productive month in my entire academic career. Somehow, thanks to delegating grunt work to Claude Code, I’ve now actually been able to write more and better artisanal, hand-crafted human prose than ever before. Here is everything I published this month:

• On Popular by Design, I wrote “Academics Need to Wake Up on AI” and Part II (with Part III coming soon), “The Gay Marriage Playbook Won’t Work for Immigration”, “Public Engagement Is Good for Your Research”, and “What Is Populism Actually Good For?” (with Yaoyao Dai).

• I published guest posts on ’s Substack: “Spread the Word: Legal Immigration Is Incredibly Difficult” (with Michelangelo Landgrave). Marc Helbling and I wrote “How We Miscategorize the Categorizers” for Futures of Difference. and I co-authored “Why America Is So Much Better Than Europe at Immigration” at . Check those out if you haven’t yet.

• If you’re into the audio format, I also appeared on three podcasts (one still forthcoming): Opinion Science with Andy Luttrell and Money & Macro Talks, with more to come soon.

• And lest anyone wonder whether all this public writing comes at the expense of “real” research: I also published two peer-reviewed pre-registration reports: one with Yaoyao Dai, “What is Populism Good for?” in Research & Politics, on the mobilization effects of populism, and another (still in progress) on “Preventing Backlash by Shifting Issue Priorities: Immigration and Depopulation in Japan” in the Journal of Experimental Political Science (with Akira Igarashi, Rieko Kage, and Seiki Tanaka).

Before we get to the links, two quick questions for you all.

Drop your answers in the poll or the comments. The ideology question was inspired by interesting conversations I’ve been having with readers who assume very different things about where I sit politically. The branching-out question is genuine: some of my most popular posts this month were not about immigration at all, and I want to know whether that reflects what you actually want more of.

Here are the March links (linking does not imply endorsement):

• The Supreme Court will hear Trump v. Barbara this spring, challenging the executive order that would strip birthright citizenship from children whose parents lack permanent legal status. My colleague Amy Hsin at Notre Dame’s Keough School co-authored an amicus brief laying out the social science evidence against it. Among the numbers: Phillip Connor, Matt Hall, and Francesc Ortega estimate that birthright citizenship beneficiaries will contribute $7.7 trillion to the U.S. economy between 1975 and 2074. Revoking it could deny citizenship to 4.8 million U.S.-born children by 2045.

  • My other Notre Dame colleague Ashley Sanchez at The Conversation has a useful explainer on the Trump administration’s new interpretation of refugee detention rules. Legal refugees who haven’t yet received green cards now face prolonged detention under a reinterpreted DHS statute. Another case of the administration using administrative tools to restrict legal pathways without passing new legislation.

• (follow him!) at the Niskanen Center has a great piece in City Journal on assimilation. He looks at areas of the country with the best assimilation outcomes a century ago (measured by second-generation out-marriage) and asks what that tells us about the challenges we face today.

• My friend Hannah Postel (who should absolutely start a Substack) at Duke Sanford has new work in the Journal of Economic Perspectives on how U.S. immigration law shaped a century of Asian American immigration. Exclusion laws kept the Asian population below 1% of the U.S. total for nearly a hundred years. After the 1965 act abolished national-origin quotas, the Asia-born population grew by roughly 2,700%. The characteristics often attributed to Asian American "culture" like high educational achievement trace back to which entry pathways policy made available. A powerful case that policy design determines not just how many people come, but who comes.

• has a thoughtful piece on identity politics and the case for making identities broader in liberal democracies. Connects to the immigration debate in ways he doesn’t fully spell out but that readers of this newsletter will recognize: narrow identities are what make immigration politically toxic, and policies that encourage broader identification can defuse that.

• The Friedrich-Ebert-Stiftung published a brief on progressive immigration policy in Germany and the U.S., authored by Hannah Tyler, Stephanie von Meien, and Cristobal Ramon. I disagree with some of their points, but the general framing is close to what I’ve been arguing: humane and flexible policies that serve the national interest are good or at least better than the status quo. Restoring public confidence in the immigration system should be among the top priorities.

• tested whether AI can detect known errors in published papers. GPT (5.4) scored 5.9 out of 10 on errors in Moretti (2021), an AER paper with a huge number of identifiable problems. Human reviewers missed all of them the first time around. If journals adopted systematic AI checks, even an imperfect system would catch mistakes that currently sail through review.

• at WYSR has a piece comparing nanotech hype in the 2000s to AI hype today. The comparison is instructive. Nanotech was supposed to change everything, attracted massive federal funding, and produced genuine scientific advances that quietly found their way into useful products. But the transformative revolution didn’t arrive on the timeline the boosters promised.

• on the credential as a democratic institution. An argument about why expertise and credentials still matter in a populist moment. Connects to a question I keep coming back to: if populism is partly a rebellion against credentialed gatekeepers, how do you defend the value of expertise without sounding like you’re defending your own guild?

• has a new paper forthcoming in International Migration Review testing whether high skills insulate immigrants from ethnic bias. In a pre-registered experiment with British respondents, a Black South African doctor was rated less welcome than an identical White one (but not in the case of fast food workers). It's a well-designed study, but, with all due respect to Alexandre and his work, I worry this is also a common case of missing the forest for the trees.

  • The detected racial penalty is about 0.3 points on a 10-point scale, with race explaining less than 1% of variance. The skill premium, by contrast, is at least 7 times larger (2.3 points), explaining >20% of variance. If anything, this paper shows how remarkably little ethnicity moves the needle once you hold other factors constant. As I keep beating this drum, skilled immigration is popular regardless of ethnicity, and we never lose sight of that.

• at Field Notes on Progress has the best headline of the month: “The AI Revolution No One’s Talking About.”1 It’s about artificial insemination transforming dairy and beef farming. Really fascinating read, especially if you haven’t thought about the politics of cattle and dairy before.

• Published today: Madeleine Sumption's What Is Immigration Policy For? (Bristol University Press). I simply can’t recommend the book enough. Sumption directs the Migration Observatory at Oxford and knows more about what actually happens when immigration policies meet reality than almost anyone alive. The book explains why so many governments struggle to design immigration policies that people trust, and why many find current arrangements unsatisfying. The essential guide I wish every policymaker, advocate, and journalist would read before entering the immigration debate. Exactly in line with the spirit of Popular by Design, the book won't tell you what to think, but it will change how you think.

As before, if you want me to write more about one of these or other related topics, let me know!

Why does yelling about “corrupt elites” seem to work in politics? From Donald Trump to Viktor Orbán to Marine Le Pen, politicians who rail against the establishment and claim to speak for “the real people” keep winning elections. The populist playbook, us versus them, the pure people against the rotten elite, appears to be one of the most effective strategies in modern democratic politics. But what if it isn’t?

I’ve spent the better part of a decade studying populist rhetoric with my co-author Yaoyao Dai, now at the University of Pittsburgh. We just published our third and latest paper on the topic, and I thought this was a good moment to reflect on what our research program has found. The short version: populism’s power is real, but much more limited than most people assume. And the reasons why it works are not what you’d expect.

Subscribe now

What we mean by populism

Before getting into the findings, a quick definition. Political scientists generally follow Cas Mudde’s influential framework, which defines populism not as a full political program but as a simple worldview (or what Mudde calls a “thin ideology”). This worldview is based on three pillars: people-centrism (politics should reflect the will of “the people”), anti-pluralism (there is one authentic popular will, not many competing interests), and moralized anti-elitism (elites are not merely wrong but evil). This is what scholars call “thin” populism because it doesn’t tell you much about actual policy. A left-wing populist like Hugo Chávez and a right-wing populist like Trump share the same rhetorical structure, the people versus the elite, but disagree on virtually everything else.

This distinction between populism and its “host ideology” (the actual policy positions a politician holds) turns out to be crucial. Because when you peel them apart, something surprising emerges.1

When politicians gamble on populism

Our first paper, “When Do Politicians Use Populist Rhetoric?“ published in Political Communication in 2022, asked a deceptively simple question: if populist rhetoric is so effective, why don’t all politicians use it all the time?

To answer this, we built the most comprehensive corpus of U.S. presidential campaign speeches at the time: 4,314 speeches from 1952 to 2016. We used a novel text analysis method combining active learning and word embeddings to measure how much populist rhetoric each candidate employed across the campaign trail. I (Alex) should say, thanks to the prowess of Yaoyao, we did all that fancy text analysis stuff before it was cool and before LLMs were even around.

The pattern was striking. Candidates who were trailing in the polls consistently used more populist rhetoric, regardless of whether they were Republicans or Democrats, incumbents or challengers. Populism, we argued, is a gamble: a high-risk, high-variance strategy that trailing candidates adopt because conventional campaigning isn’t working. If you’re already behind, why not shake things up?

Think of it like a football team that’s losing in the fourth quarter. You start throwing long passes not because they have a higher expected value, but because safe plays guarantee you lose. Barry Goldwater, George McGovern, and Donald Trump (in 2016, when most polls had him behind) all fit this pattern. They reached for populist rhetoric when they had little to lose.2

The (in)effectiveness of populist rhetoric

But does the gamble actually pay off? Our second paper, published in Political Science Research and Methods in 2024, tested this directly with a survey experiment.

We presented U.S. respondents with pairs of realistic campaign messages from hypothetical primary candidates. The messages varied on two dimensions: populist features (people-centric language, anti-elite attacks, anti-pluralist framing) and substantive policy positions (on immigration and other issues). This design let us isolate the effect of populist rhetoric from the underlying policy content, something that is nearly impossible to do when observing real elections, where populism and policy positions come bundled together.

The result was unambiguous: none of the populist features had an independent effect on candidate choice. Not people-centrism, not anti-elitism, not anti-pluralism. Not individually, and not in combination. What did matter, enormously, were policy positions that aligned with voters’ own preferences. Voters chose candidates based on what they promised to do, not on how dramatically they framed the conflict between the people and the elite.3

This finding is consistent with other experimental work. When researchers across multiple countries carefully separate populist style from policy substance, the style itself contributes very little to voter decisions.

So: if populist rhetoric doesn’t actually persuade voters, why does it seem to work? Why do populist candidates keep winning?

What populism is actually good for

This puzzle motivated our newest paper, our first ever registered report (where scholars publicly specify their hypotheses before running their experiment), now published at Research & Politics. We hypothesized that populism’s real contribution might not be persuasion but mobilization: getting people who already agree with you to actually show up and vote.

Previous studies, including our own, used what’s called a “forced choice” conjoint experimental design: respondents had to pick one candidate or the other. But in real elections, people can also stay home. To capture this, we ran a large-scale, preregistered survey experiment that added an “abstain” option, a seemingly small change that turns out to matter a lot.

What did we find? First, the basic persuasion result replicated: policy positions still dwarfed populist rhetoric in driving vote choice. Having a policy-congruent candidate increased the probability of voting by a massive 27 percentage points. Populist rhetoric, by contrast, had no meaningful persuasion effect.

But here is the twist: populist rhetoric did have a small but statistically significant mobilization effect. Having at least one populist candidate in a race was associated with a ~1.5 percentage point decrease in abstention. The effect was concentrated among voters who already held populist attitudes and encountered a candidate whose policy positions they liked. In other words, populist rhetoric didn’t convert skeptics; it energized true believers to get off the couch.

Meanwhile, non-populist voters did not appear to punish their preferred candidates for using populist rhetoric. This asymmetry is key: populism is a low-cost mobilization tool. It fires up your base without alienating persuadable voters.

Are hypothetical but cleanly identified ~1.5 percentage points a lot? In most elections, no. But in a close race (and modern elections in the U.S. and Europe are often decided by razor-thin margins) even a small mobilization advantage can be decisive. This may help explain the apparent paradox: populist rhetoric doesn’t change many minds, but it doesn’t need to. It just needs to get a few more supporters to the polls.

The media amplification question

There’s one more possibility worth considering: the role of the media. Populist rhetoric is, almost by design, dramatic and newsworthy. When a candidate calls the entire political establishment corrupt and claims to be the voice of the forgotten people, that generates coverage, and coverage generates name recognition, which generates votes.

The most vivid illustration is Trump’s 2016 campaign, which received an estimated $2 billion in free media during the primaries alone, far more than any rival. Much of that coverage was driven by his populist style: the outrageous claims, the attacks on the “swamp,” the rallies designed for television. Journalists couldn’t look away. And there is some evidence in the growing media populism literature that this pattern generalizes beyond Trump, with populist candidates across countries receiving disproportionate media attention relative to their actual electoral standing. That’s what our Notre Dame colleague Marc Jacob recently found too, but in the case of negative politics and political insults more generally—it captures and generates attention.

If populist rhetoric’s main benefit is generating outsized media attention, which then translates into awareness and mobilization, then the mechanism isn’t really about what populism says to voters. It’s about what populism says to journalists. This is consistent with our finding that populism mobilizes rather than persuades. But the media-amplification hypothesis still needs direct testing, and ambitious PhD students should certainly take this on (unless we or our Claude Code get to it first).

What does this all mean?

So what is populism actually good for? Based on our and other recent research, we’d summarize it this way:

1. Populist rhetoric is a gamble, adopted primarily by candidates who are already losing. It’s a variance-increasing strategy, not a winning formula.

2. It doesn’t persuade. Voters care about policy positions, not populist framing. The “host ideology,” what you actually promise to do, matters far more than how dramatically you frame the people-versus-elite conflict.

3. It may slightly mobilize, particularly among voters who already hold populist attitudes and agree with the candidate on substance. This is a modest but potentially consequential effect in close elections.

4. Media amplification may be a key mechanism, turning populist drama into disproportionate coverage. But we need more direct evidence.

The biggest takeaway, both for those who fear populism and those who are tempted by it, is that substance matters more than style. Politicians who deliver tangible results, or credibly promise to, will outperform those who simply shout louder about corrupt elites. This is consistent with what I (Alex) argue in my recent book, In Our Interest, in the context of immigration: policies that are demonstrably beneficial do more to win and maintain public support than any amount of rhetorical framing.

Populism is not nothing. But it’s not the all-powerful electoral weapon it’s often made out to be. Don’t confuse volume for effectiveness. The people who keep winning elections on populist platforms are winning mostly because of what they promise and do, not because of how they talk about it.

Share

This piece is more personal than usual. After my recent posts on AI in academic writing, I received a wave of private messages from fellow academics who agreed with my hot takes but wouldn’t say so publicly. My first instinct was to write about self-censorship in the academy. But the problem runs deeper. Most academics don’t want to engage with the public at all. This piece is about why that’s self-defeating, and why many of my colleagues are getting it wrong.1

A few years ago, I gave a talk at a retiree center in Charlotte, North Carolina, about my research on public attitudes and making immigration popular. Before I could even start, an older woman in the back raised her hand. “Why,” she asked, “would we want to make immigration popular in the first place?” No academic colleague had ever asked me that question before. Although I wasn’t able to bring her to my side fully, it turned out to be one of the most productive conversations I’ve had about my research with anyone.

Subscribe now

I am increasingly convinced that for social scientists and academics, engaging with the public is not a distraction from research but a direct input to it. The audiences you encounter outside the seminar room, the questions journalists ask, and the pushback from readers who have no stake in your theoretical framework: these are all important data. They reveal blind spots that insular academic communities systematically miss. Public engagement also forces you to justify, in plain language, why your work matters, which turns out to be a surprisingly effective filter for figuring out whether it actually does.

The standard academic view treats public engagement as a trade-off: time spent writing for popular audiences is time not spent on “real” research. I will argue the opposite here. My own experience and the experience of researchers I admire suggest that talking to non-academic audiences, writing for the public, and presenting research to people who might genuinely disagree with you make your scholarship sharper and more honest. It does this by stress-testing our ideas against the one audience that academic peer review systematically excludes: the very people researchers claim to study.

What engaging the public taught me peer-review didn’t

One of my most cited findings on immigration opinion came not from a faculty seminar but from conversations with policymakers in Washington. They all kept telling me the same thing: even when polls show majority support for more liberal immigration policies, politicians still will not touch the issue. The anti-immigration side simply seems to care more. This observation never came up in the academic literature I had been reading, where the focus was almost entirely on why people oppose immigration, not on how much they care about it in the first place, including the pro-immigration side.

That disconnect led me to a paper documenting what I called the “issue importance asymmetry” in academic-speak, describing the simple fact that anti-immigration voters are consistently several times more likely to rank immigration as their most important political issue compared to pro-immigration voters. This holds across decades in the US, the UK, and Europe. It is one of the most consistent findings in immigration research. And it started with listening to people outside academia who were closer to the political reality than most of my colleagues.

This was not a one-off. People outside the university often see things that people inside it miss, not because they are smarter, but because they are working from a different set of assumptions. When the overwhelming majority of your colleagues share the same political priors, certain questions never get asked. I have written about how well-meaning colleagues have suggested I soften findings that might “feed the far right,” even when the results were solid. That kind of filtering is invisible inside the academy. It becomes very visible the moment you share the unfiltered version with a public audience and discover they find honesty more credible, not less.

My piece arguing that Western countries do not “need” immigration grew directly from this. Voters who heard experts claim economies would collapse without immigration and saw their country functioning just fine concluded the experts were dishonest. The reframing came from paying attention to what skeptics actually found persuasive, not from academic theory. Similarly, when I wrote about community sponsorship, I highlighted polling showing that 73 percent of Republicans supported the Welcome Corps, a U.S. pilot of the sponsorship program, because it taps into conservative values of localism and faith. Most immigration scholars had not even thought to test whether the right might support refugee resettlement, because the academic framing treated it as exclusively a humanitarian, left-wing cause.

The wisdom of an old Charlottean

Let me tell you more about that Charlotte talk. The audience was conservative and very old, and the woman who challenged my premise was not the only skeptic. Before I could start, an agitated man asked, as a gotcha question, whether I believed Americans had the right to secure their border. I said yes. He almost seemed disappointed I did not proclaim “no human is illegal” or something to that effect. He sat back and calmed down.

After the woman said we did not need any foreigners, I agreed that immigration is a challenging issue and asked whether she thought we should also stop, say, German engineers from coming. She thought for a few seconds and then said, “Of course not.” Within a few minutes, we had moved past the headline positions and were having a genuinely productive conversation about which specific immigration policies she did and did not support, and why. In the end, folks tried to listen to me despite all the hearing problems in the audience for the rest of my presentation.

No academic audience had ever forced me to defend the premise of my research in quite that way. It made me rethink some of the ways my colleagues and I frame our questions and answers. We often assume that the value of studying what makes immigration policies popular is self-evident. It is not, and discovering that in a room full of retirees was more useful than discovering it from a reviewer comment.

Engaging with the public also improved my writing, and more so than LLMs ever could. When you have to translate a complex finding into a sentence that a non-specialist can follow, you quickly discover whether you actually understand it yourself. The vagueness that academic peer reviewers sometimes wave through does not survive a comments section or even a relatively shallow journalist’s follow-up question.

When jargon replaces argument

This brings me to an uncomfortable observation about a certain kind of academic work that I think public engagement would cure. Some research, particularly in what is called “critical” or “postmodern” scholarship, has become so insulated from public scrutiny that it is almost impossible to explain what it is actually saying, or why it matters.

I recently attended a seminar by Charmaine Chua, a geographer now at Berkeley, who presented research from a forthcoming book based on fieldwork aboard a container ship. The underlying empirical work was genuinely fascinating, on top of her great photography: vivid, detailed observations about the enormous salary disparities among crew members based on national origin, and the daily mechanics of global shipping that most people never see.

But the framing was almost entirely directed at an audience of critical geographers and “abolitionists.” Every observation had to be routed through Marx or David Harvey. One framework had to be “connected” to another framework, which had to be “put in conversation” with a third. There is a real story here about global inequality and labor exploitation, and it was being buried under layers of disciplinary performance.

In fairness to Chua, she has also written for popular outlets like Boston Review and Jacobin, translating her shipping research into language that (at least highbrow, left-wing) non-academics can engage with. She is, in that sense, doing the kind of public-facing work I am arguing for here. But the gap between the seminar version and the public version was striking. Even though we may disagree politically, I suspect her public version was much better. And not just because it was more accessible, but because the discipline of writing for a general audience forced clearer thinking about what the research actually shows.

This is not an isolated case, and the problem is that the vast majority of critical and empirical scholars alike do not go beyond publishing their work in obscure journals nobody reads. When research is never exposed to audiences who might say “I don’t understand what you mean” or “why should I care?”, it can drift into a self-referential loop where the work exists primarily to satisfy disciplinary gatekeepers. Public engagement is a corrective. It forces you to answer the question that every taxpayer has the right to ask: What is this for?

But engagement is not activism

I want to draw a distinction here that often gets lost. Public engagement is not the same thing as political activism. Confusing the two has done real damage, particularly in fields like sociology and political science, where “activist scholarship” has become an identity rather than a practice.

Activist scholarship’s problem is not that scholars have political views. Everyone does after all. When the scholarship itself is oriented toward a predetermined political conclusion, it stops being scholarship in any meaningful sense. And in practice, activist scholarship has tilted overwhelmingly in one ideological direction, which has undermined the credibility of entire fields. This includes hard science, too. Even the scholars doing this work would benefit from making their research more accessible to audiences beyond their own political coalition, because accessibility invites challenge, and challenge is what separates inquiry from advocacy.

What I am describing is closer to what calls the “problem-solving” approach to social science. Samii argues that social scientists should orient their work toward clearly defined societal problems, using normative analysis to identify what needs fixing, observational research to understand why, and experimental methods to test what works. This is distinct from both “disinterested” puzzle-solving (which often produces technically impressive work that no one outside the discipline reads or needs) and from activist scholarship (which knows the answer before the question is asked). Problem-solving research takes sides on the problem, not on the politics. It asks: Does this policy work? How do we know? What should we try instead?

That framework describes what I try to do with my own research and public writing. I’m sure I have my own biases and blind spots, but my Substack is decidedly not an advocacy project. It is an attempt to make research that is often locked behind paywalls and disciplinary jargon available to the people, including policymakers, journalists, and voters, who could actually use it. And the process of doing that has made my research better, not worse, because it forces me to change my mind on issues every once in a while.

This does not have to be an individual effort. Some departments have made public engagement part of their institutional identity. George Mason’s economics department with probably the single highest concentration of influential bloggers is a good example: serious, well-published researchers who also shape public discourse on the issues they study even when they disagree (e.g., compare and on immigration). More social science departments, and especially public policy schools, should follow that model. The infrastructure for combining fundamental research with public influence already exists. Most places just choose not to use it.

Taxpayer-funded research belongs to the public

There is also a straightforward accountability argument for public engagement that I think deserves more weight than it usually gets. Most social science research in universities is funded, directly or indirectly, by taxpayers. The National Science Foundation, the National Institutes of Health, and state legislatures fund the grants, the labs, and the salaries. Taxpayers underwrite the whole enterprise.

That creates an obligation. Not an obligation to oversimplify, or to produce findings that voters will find convenient, but an obligation to make the work legible. If you cannot explain to a non-specialist why your research question matters and what you found, that is worth examining. Sometimes the explanation is genuinely difficult because the work is methodologically complex, and that is fine. But you should at least be able to explain why the methodological complexity is necessary and what it is in service of.

I think this test is actually useful as a self-check. If I am working on something and I find that I genuinely cannot explain to a thoughtful non-academic why it matters, that is a signal that I should reconsider either the framing or the project itself. Not everything that is publishable is important. And not everything that is important is inaccessible. The exercise of translation is also an exercise in self-honesty.

There is a more basic point here that often gets lost. Academics are not just academics. They are also citizens, presumably interested in contributing to the public good. It makes sense to do that using your expertise rather than compartmentalizing it. When I see colleagues who study migration and its political implications but never comment on the topic publicly, while sharing their hot political takes on Facebook anyway, it strikes me as a missed opportunity. The idea that you can wear a professor hat and a citizen hat and never connect them does not hold up for most social scientists. You are already a citizen with political views. You might as well be one with informed political views who shares the basis for them.

Yes, it costs something. But you should do it anyway.

Many academics have been told, by colleagues or even their dean, not to spend too much time on public engagement, or warned not to say something publicly that would embarrass their college. If this is advice against posting on social media without any serious research work behind it, that may be quite sound. After all, unless you are at a public policy school, even a piece in The New York Times will not count for much in your annual review, let alone toward tenure. So I do not want to pretend that public engagement is costless.

The most obvious cost is time. Writing a Substack post or giving a public talk takes hours that could be spent on a paper. For junior scholars without tenure, your promotion committee probably will not count your Boston Review essay or your popular podcast appearance. The incentive structure of academia still largely rewards journal publications, grant funding, and citations from other academics.

Then there is the social cost. Colleagues who view public engagement as unserious can be quietly dismissive. I have experienced this myself. Not as direct criticism, but as a certain subtle skepticism, a sense from some peers that time spent writing for the public is time not spent on “real” work. The signals are usually indirect: a raised eyebrow, a conspicuous lack of interest, the faint suggestion that popular writing is something you do instead of scholarship rather than alongside it.

And there is the online environment, which can be genuinely toxic. Platforms like Bluesky, in particular, have become what I can only describe as a corrupting influence on academic discourse. The incentive structure rewards performative outrage and virtue signaling over substance.

Academics who engage there often find themselves dragged into pile-ons that have nothing to do with the quality of their ideas and everything to do with whether they said something that violated the platform’s ever-shifting ideological consensus. Compare this to long-form platforms like Substack, where the incentive structure at least partially rewards depth and evidence. Not all public engagement is equal, and choosing the right venues matters.

Having said all that, do it anyway. The alternative is worse. recently argued that if you write publicly about contested topics, getting piled on is not a rare catastrophe but a predictable occupational hazard. The question is not whether it will happen but when. His advice is simple: accept the risk and keep writing. Do not let the possibility of a pile-on shape what you are willing to say.

The issue goes beyond managing pile-ons, though. has written persuasively about intellectual courage as a scarce resource. Her most striking observation is that academics regularly message her privately to say they agree with positions she has taken publicly but are unwilling to say so themselves. I have witnessed this firsthand recently with my hot takes on AI in academia:

The courage to state publicly what you believe privately, especially when it is unpopular within your professional community, is not a nice-to-have. It is an epistemic necessity. Truth emerges through open argument. If everyone self-censors, the entire discovery process breaks down. has made a related case that tenured academics possess extraordinary free-speech protections and have a responsibility to actually use them. His own experience suggests that speaking honestly and across ideological lines actually improved his professional relationships and research collaborations. The fear of consequences was larger than the consequences themselves.

I have found this to be true in my own experience as well. After my recent piece challenging pro-immigration misinformation from within the pro-immigration camp or prompting my AI-skeptical colleagues to lock themselves in a room with Claude Code, I received pushback from many quarters. But I was also struck by the number of academics, including left-of-center scholars, who publicly endorsed these pieces that challenged their own side’s orthodoxy. As I wrote at the time, tenured (and untenured) professors should do this more often.

DEI is what happens when no one talks to the public

Let’s talk about faculty hiring for a moment because this is somewhat personal. The standard explanation for why universities went so far off the rails on race-based hiring after 2020 is left-wing bias and self-censorship. People were truly afraid to speak up. There is truth to that. Even influential tenured Harvard professors like Steven Pinker and Jill Lepore found it difficult to challenge the new orthodoxies.

But the deeper problem was that academics simply did not talk to people outside their institutions. Many of the faculty and administrators who embraced racial balancing in hiring genuinely believed they were doing the right thing. They had spent years inside institutions where this logic was so normalized that it never occurred to them to ask whether the public supported it, whether it was legal, or whether systematically excluding qualified candidates on the basis of race and sex might be ethically wrong.

Had they asked, the answers would have been clear. Race-based affirmative action in hiring is extremely unpopular among the American public, and has been for decades. Taxpayers fund universities to advance science and the public good. Nobody is paying us to maintain a particular racial balance among the faculty.

The scale of what happened is now well documented. 's viral "Lost Generation" essay highlighted that white men went from 49 percent of tenure-track hires in 2014 to 27 percent by 2024. At UC Irvine, just three of 64 tenure-track hires in the humanities and social sciences since 2020 were white men (4.7 percent). of the National Association of Scholars obtained internal emails through hundreds of public records requests that showed the machinery plainly: at one NIH-funded program, an administrator wrote "I don't want to hire white men for sure." at the Washington Free Beacon documented similar patterns across the country. I can also tell you about my first-hand experiences of search committee members who invited me to give a job talk telling me candidly that it was not going to happen because of my racial background (of course, most would be smarter than to invite me or say anything at all).

All in all, if you were a white or Asian man on the academic job market in 2020 or 2021, especially one from abroad, your marginal chances of landing a tenure-track position in many fields approached zero, all else equal. The fact that the existing stock of senior faculty was predominantly white and male was no consolation to an ambitious but broke thirty-year-old finishing a PhD. So many brilliant scientists with enormous potential have either become adjuncts with no future or left academia if they were lucky. The harm to science in terms of discoveries delayed or never made is staggering.

So, the collapse of public trust in higher education over this period was quite predictable. Academics knew what was happening. Many disagreed privately. But almost nobody was talking or explaining it to the public or pointing out that these policies had no popular mandate. That silence left the field to culture warriors on both sides and made the eventual backlash worse than it needed to be. It also cost a generation of talented researchers their careers, which is not the kind of thing a healthy profession stays quiet about.

How to actually do it better

So, our grim collective action problems aside, if you are an academic considering more public engagement, here are three things I have found genuinely useful.

Have a functioning website. First, and foremost, for the love of all that is reasonable, have a website. An up-to-date, accessible academic website. I genuinely do not understand colleagues who don’t. The notion that good research will find its audience on its own is optimistic to the point of delusion in an age when people are bombarded with information from every direction.

If you have done the work, make it findable. Thanks to Claude Code, from now on, my own site will be available in a dozen global languages, because accessibility means nothing if it stops at the English-speaking world. This very piece will be available on it in all languages upon publication.

Present your research to people who might disagree with you. This sounds obvious, but surprisingly few people do it, so I cannot recommend this enough. Go to a retiree center or a community forum. Or Bluesky if you’re writing on AI or LGBT issues. The audiences in these spaces are far more politically and demographically diverse than those of any college seminar. They will ask you questions that your colleagues never would, and those questions will reveal whether your argument actually holds up outside the assumptions of your discipline. The woman in Charlotte who asked me why we would want to make immigration popular taught me more in five minutes than many peer review reports have.

Write for the public. Start a blog or a newsletter. It doesn’t have to be a Substack, even though all cool people in academia are increasingly here now. The discipline of writing regularly for a non-academic audience changes how you think. It improves your prose, which then improves your academic papers. It forces clarity. And it opens you up to feedback from people with real-world experience in the things you study. Some of the most useful responses I have received to my Substack have come from readers who challenged my research claims based on their own experience: voters, immigrants, local officials, business owners, and even anonymous randos from the internet. That is a form of peer review that the academy does not provide.

Give pre-recorded interviews and do science podcasts. Popular science and policy podcast hosts ask different questions than academics do. They want to know what your findings mean for people who are not specialists. They push you to be concrete and specific. And they often identify angles that you, embedded in your own literature, have overlooked. They are not interested in any gotchas, so they would send you questions in advance. I have had those folks ask me questions that opened entirely new lines of inquiry, things no academic colleague had thought to raise because everyone in the field took the same assumptions for granted.

What not to do, or do with caution

Do not confuse social media arguments with public engagement. Getting into reply threads on X or Bluesky can feel like engaging with the public, but the incentive structure on those platforms rewards dunks and outrage, not depth. A 280-character exchange rarely changes anyone’s mind or improves your thinking. Long-form writing, in-person talks, and substantive interviews are where the real feedback loop happens. Use social media to share your work and find your audience, not to conduct your debates. And yes, I know I should follow this advice myself more.

Do not wing it on unfamiliar topics. The fastest way to undermine your credibility as a public-facing academic is to opine confidently on something you have not studied. One bad appearance on a topic outside your expertise can overshadow years of careful work within it. If you are asked about something adjacent, either redirect to what you actually know or say “I don’t know enough about that to give you a useful answer.” That sentence, rarely heard from both pundits and academics, tends to earn more respect than a half-informed hot take.

I have personally been asked on multiple occasions to come on news shows and talk about the US-Mexico border crisis, but I politely declined because it is not my area of expertise. Similarly, I now mostly refuse to talk to journalists about AI despite my recent notoriety on the topic, because I am a novice.2 Knowing when to say “this is not my lane” is itself a form of intellectual honesty that builds credibility over time.

Be generally selective with media requests, especially live interviews. If a journalist you know and respect reaches out on a topic you have actually studied, you should absolutely talk to them. Just understand that you will spend a few hours prepping and talking, and you may not be acknowledged or, worse, may be misinterpreted when their piece comes out.

For live interviews in particular, the risks are higher: the time you are given is limited, and you may not know what you will be asked. If someone you have not heard of reaches out, or the topic is adjacent to your expertise rather than central to it, the answer should be a polite no in most cases. Unless you do want to become the maligned stereotypical “talking head,” of course.

I will write more on this soon, but my sense is that with the help of agentic AI tools, scientists and experts should increasingly be able to produce better popular pieces on their own topics than generalist journalists can.

What’s lost when researchers stay silent

The stakes of this argument go beyond individual careers. When researchers with genuine expertise refuse to engage with the public, they leave a vacuum. And that vacuum gets filled by journalists and pundits without relevant training and advocates with axes to grind, and eventually by politicians who find it convenient to misrepresent what the evidence shows. The result is a public discourse about science topics that is poorer, more polarized, and more detached from evidence than it needs to be.

I have written at length about how “highbrow misinformation” develops when academic research gets filtered through advocacy groups and media outlets that strip away caveats and complexity. One way to fight this is to cut out the intermediaries. Not by replacing them entirely, but by making sure that the original researchers are also in the room, in the comments section, on the newsletter, explaining what their findings do and do not show.

The false trade-off between “serious scholarship” and public engagement has real consequences. It keeps good research invisible and lets bad arguments go unchallenged. It deprives researchers themselves of the feedback that would make their work better. If you are a scientist sitting on findings that matter, and you are not making them accessible to the people they are about, you are leaving value on the table for your field and for the people your research claims to serve.

Share

A few years ago, while presenting my research on immigration attitudes to a room of policy advocates, I was politely but firmly told that studying what makes immigration policies more or less popular was not really necessary. Immigration support, my influential interlocutor explained, was already growing steadily, much like support for same-sex marriage. We just needed to keep telling people that immigration is good, correct the misinformation spread by bad actors, and wait for the generational tide to carry us forward. Why bother designing policies for popularity when popularity was already arriving on its own?

I have seen and heard versions of this argument more times than I can count. The comparison between immigration and same-sex marriage has become something like conventional wisdom among progressive advocates, a comforting story about the arc of public opinion bending toward openness. And it is not hard to see why the analogy is tempting. Both causes involve expanding rights and freedoms, face opposition rooted in cultural anxieties, and have seen meaningful shifts in public attitudes over recent decades.

Subscribe now

The post-2024 reckoning over progressive immigration strategy has only reinforced the comparison. As the Trump administration’s enforcement measures take hold, public opinion is swinging back in a more pro-immigration direction. To many advocates, this looks like the tide turning, much as it did for gay marriage, and it would seem to validate the theory that immigration supporters should continue to focus on messaging. But support for same-sex marriage rose steadily for two decades and then locked in place (with some minor fluctuations): once Obergefell v. Hodges settled the legal question and millions of Americans came to know gay and lesbian people in their own lives, there was no mechanism to reverse the shift.

Immigration opinion, as I will try to convince you in this piece, does not work like this. The gay marriage analogy is wrong in ways that matter enormously for strategy. And the longer immigration advocates cling to it, the longer they will delay the kind of work that could actually make progress possible.

The triumph that became a template

The success of the same-sex marriage movement in the United States is genuinely extraordinary. In 1996, when Gallup first asked Americans whether marriages between same-sex couples should be legally valid, just 27 percent said yes. By 2015, when the Supreme Court decided Obergefell v. Hodges, that number had crossed 60 percent. Today, it sits around 69 to 71 percent. This is one of the fastest and most dramatic opinion shifts in the history of American polling.

The movement achieved this through a combination of moral clarity, personal storytelling, and strategic litigation. Advocates refused to settle for civil unions. They framed their cause around love, commitment, and family, values that resonated across ideological lines. And crucially, as more gay and lesbian Americans came out to their families and communities, abstract opposition gave way to personal connection. It was, by almost any measure, a masterclass in social change.

It was also, as has argued, a deeply unusual case that progressives mistakenly adopted as a universal template. The gay marriage playbook (refuse compromise, frame opposition as bigotry, deny trade-offs, and wait for opinion to catch up) was then applied to issues ranging from health care to policing to immigration. Other like and have noted that even within the LGBTQ movement itself, the playbook has not transferred well from marriage equality to the more contested terrain of gender identity and trans issues. has recently argued that the structural conditions that made the “it gets better” trajectory work for gay rights (demographic scale, random distribution across families, the powerful contact effect of coming out) simply do not hold for every cause.

Immigration is one of those causes. And the mismatch runs deeper than most advocates realize.

Why the analogy breaks down

To be fair, there are similarities between immigration and same-sex marriage. Both involve, at some level, expanding personal freedoms and reducing legal discrimination based on characteristics largely beyond an individual’s control. Both efforts ask a majority to accept people whom some portion of the public views with suspicion or hostility. And in both cases, opponents have relied on fear-based messaging that exaggerates threats and dehumanizes the people in question. These parallels explain why thoughtful advocates reach for the comparison. But the structural differences are profound and show why a strategy built for one cause will fail the other.

Ingroup versus outgroup. Gay and lesbian Americans are, by definition, members of the national community. They are someone’s child, sibling, coworker, or neighbor. The success of the marriage equality movement depended heavily on this fact: the most powerful engine of attitude change was personal contact with people who were already part of the social fabric. Roughly 84 percent of Americans report knowing a gay or lesbian person personally, a figure made possible by the fact that LGB individuals constitute 8 to 10 percent of the population and are distributed randomly across families, communities, and political affiliations. The question was never whether they belonged, but whether they would be fully recognized.

Immigrants, and especially prospective immigrants who have not yet arrived, are outsiders seeking entry. While many Americans do know immigrants personally, the people whose admission is under debate are often thousands of miles away, invisible to the voters deciding their fate. The emotional and political dynamics are fundamentally different. You cannot “come out” as a future immigrant to your family at Thanksgiving dinner in America.

Already here versus seeking entry. The marriage equality movement asked the public to recognize a reality that already existed. Gay and lesbian couples were already living together, raising children, building lives. Legal recognition was about catching the law up with the facts. Immigration, by contrast, is primarily about regulating flows: how many people to admit, under what conditions, through what channels. The people whose fate hangs in the balance often have no presence in the host country and no voice in its politics. This is not a matter of recognizing what is, but of deciding what will be. That is a categorically harder sell, because the beneficiaries of more open policies are largely absent from the political conversation.

Symbolic recognition versus material trade-offs. Same-sex marriage was, for most Americans, essentially costless. Extending marriage rights to gay couples imposed no burden on straight couples’ marriages, finances, or daily lives. There are no material constraints on the number of marriage licenses. Granting more marriage licenses does not reduce the value of existing marriage licenses. This is a crucial and underappreciated feature of the issue, and one that Johnson identifies as the key reason the playbook fails when applied elsewhere.

Unlike marriage licenses, immigration involves real and perceived costs to people you care about the most: competition for jobs, pressure on public services, cultural change, and housing demand. Whether or not these costs are overstated in the aggregate (and economists generally agree that they are), they are not evenly distributed, and they are not imaginary to the communities that experience them most acutely. A strategy that worked for a costless cause will not work for one where trade-offs are genuine and felt.

Courts versus legislatures. Obergefell settled the marriage question through the judiciary. A single Supreme Court ruling made same-sex marriage the law of the land, regardless of what any state legislature thought. This created a kind of finality that is enormously powerful for social movements: once the ruling came down, the debate was effectively over, and the remaining task was cultural adjustment rather than ongoing political combat.

Immigration policy has no equivalent shortcut. While courts can and do adjudicate individual immigration cases, block executive overreach, and shape enforcement at the margins, the fundamental architecture of immigration (visa categories, numerical caps, enforcement priorities, funding levels) is set (or at least supposed to be set) by legislation. There is no Obergefell for immigration. Each policy change requires building and sustaining legislative coalitions, which means contending with the very public opinion dynamics that advocates hope to bypass through persuasion.

Salience and who has a voice. For LGBT Americans, marriage equality was intensely personal, arguably the most important political issue in their lives. This asymmetry in passion was a strategic asset: advocates cared more than opponents and organized accordingly.

The “other side” of the marriage debate, socially conservative voters, was not uniformly passionate about preventing it. Anti-gay-marriage organizations were vocal and well-funded, but their intensity was not shared by the broader base they claimed to represent. In 2004, a constitutional ban on same-sex marriage ranked 21st out of 22 national priorities in Pew polling. By 2014, more than a third of same-sex marriage opponents told PRRI the issue was not that important to them personally, and large majorities on both sides saw legalization as inevitable. Many rank-and-file opponents simply had other priorities and came to see the fight as not worth the political cost.

Immigration presents the opposite dynamic. The people who stand to benefit most from more open policies, potential immigrants abroad, have no vote, no voice, and no political power in the receiving country. Meanwhile, those who perceive themselves as bearing the costs of immigration often care about the issue intensely and have proven willing to organize politically around it, from Brexit to Trump’s 2024 campaign. The passion asymmetry runs the opposite way.

The limits of persuasion

None of this means that persuasion is useless. Alexander Coppock’s careful experimental work has shown that providing people with information about policy issues shifts attitudes by about five percentage points on average, and that this shift occurs roughly equally across the political spectrum. There is no “backlash” effect from trying to inform people. Similarly, David Broockman and Joshua Kalla’s deep canvassing experiments have demonstrated that non-judgmental, narrative-based conversations can reduce exclusionary attitudes toward immigrants already in the country, a meaningful and durable effect, even if modest in magnitude.

But there are reasons to believe that persuasion alone, no matter how sophisticated, cannot solve the immigration puzzle. First, immigration is a domain where counter-messaging is powerful and abundant. Anti-immigration advocates, from populist politicians to media figures to viral social media accounts, often care more about the issue than pro-immigration forces do, and they have a structural advantage: concrete stories of harm are more emotionally compelling than abstract statistics about aggregate benefits. For every careful study showing that immigrants contribute more in taxes than they consume in services, there is a vivid news segment about a local community overwhelmed by a sudden influx. Coppock’s own findings suggest that if persuasion moves people roughly equally in both directions, the side with more motivated and more prolific messengers may well have the edge.

Second, persuasion’s political relevance is limited by a basic fact about democracy: people do not directly set immigration policy. Even if a well-designed campaign moved public opinion several points in a more favorable direction, this would not automatically translate into legislative change. Immigration policy is shaped by legislative coalitions, interest groups, bureaucratic capacity, executive priorities, and, critically, how salient the issue is to voters at election time. Public opinion is merely one input, not a mandate. This is quite different from gay marriage, where opinion change combined with judicial action to produce a fait accompli.

Third, and perhaps most fundamentally, the trajectory of immigration opinion looks nothing like the steady upward march of support for same-sex marriage. Immigration attitudes are thermostatic: they react to the policy environment rather than following a secular trend. When a government is seen as having lost control of immigration, public opinion turns sharply restrictionist. When enforcement tightens, opinion softens. Gallup’s data illustrates this vividly: the share of Americans who said immigration should be decreased surged to 55 percent in 2024, then fell to 30 percent by 2025 as border crossings declined under the new administration’s enforcement measures. This is not an arc bending toward openness. Rather, it is a thermostat that adjusts up and down in response to perceived conditions. You cannot persuade your way past a thermostat.

What playbook would actually work

If the gay marriage playbook is the wrong model, what is the right one? A better analogy might be vaccination. Vaccines are among the most demonstrably beneficial interventions in human history, and yet persuasion alone has never been sufficient to achieve the uptake that public health requires. Anti-vaccine sentiment persists despite overwhelming evidence of efficacy, because persuasion, however well-founded, cannot single-handedly overcome entrenched suspicion, motivated counter-messaging, and the human tendency to weigh vivid anecdotes over aggregate data.

What actually works is not just telling people vaccines are safe and effective, but designing systems (school enrollment requirements, workplace policies, accessible distribution networks) that make vaccination the easy default. The product had to be genuinely good and the policy architecture had to make participation straightforward. Persuasion played a supporting role, but it was not the main mover.

Immigration needs a similar shift in thinking. Rather than pouring resources into campaigns designed to convince the public that all immigration is beneficial, a claim that is, at best, an oversimplification, advocates should focus on working with governments and policymakers to design immigration policies that are genuinely and visibly beneficial to receiving countries and communities. This is the difference between persuasion and what I have called making immigration popular by design.

What does this look like in practice? First, it means pushing for specific, well-designed programs rather than abstract openness. New visa categories for high-skilled workers who demonstrably fill acute labor market gaps. Labor mobility partnerships that connect migrant workers with employers in sectors facing genuine shortages, with built-in mechanisms for oversight and accountability. Private refugee sponsorship programs that give communities a direct stake in successful integration, turning residents from passive spectators of government policy into active participants with skin in the game. Administrative reforms that make the system faster, more predictable, and more transparent, so that the legal pathway is not so dysfunctional that circumventing it becomes the rational choice.

The common thread is specificity. The gay marriage movement had the luxury of a single, clear demand: let us marry. Immigration reform has no equivalent single slogan because immigration is not a single thing. It is dozens of distinct policy channels (family reunification, employment visas, refugee resettlement, student migration, seasonal labor, asylum), each with its own logic, constituency, and set of trade-offs. Treating immigration as a single cause that just needs its “marriage equality moment” obscures the reality that different policies enjoy vastly different levels of public support. Skilled worker visas are broadly popular. Large-scale low-skilled immigration is not. Pretending otherwise is self-deception.

Second, it means being honest about trade-offs. The gay marriage movement could afford to be maximalist because the cause was genuinely costless. Immigration is not costless, or at least it is not perceived that way, which in a democracy amounts to much the same thing. Advocates who dismiss public concerns about rapid demographic change, labor market competition, or the strain on local services are being strategically obtuse. The path to more open immigration runs through demonstrating that specific policies produce specific benefits, not through insisting that opposition is merely a product of ignorance or bigotry that better messaging will cure.

Third, as often emphasized by and folks from , it means engaging with enforcement rather than treating it as the enemy. One underappreciated reason that support for same-sex marriage proved so durable is that the reform did not require the public to trust the government to manage a complex system. Marriage equality was self-executing: once legal, couples could simply marry. Immigration reform, by contrast, requires public trust that the government can administer new policies, that new visa holders will actually leave when their terms expire, that employers will be held accountable, and that the system will function as designed. Advocates who treat enforcement as inherently hostile to immigrant rights undermine the very trust that makes more open policies politically possible. The countries that have managed to sustain relatively open immigration (Canada, Australia, and, until recently, Germany) have done so in part by maintaining credible enforcement alongside expansion.

Finally, it means working with the thermostatic nature of public opinion rather than against it. If public attitudes soften when people feel the system is under control and harden when they feel it is not, then the most pro-immigration thing a government can do is create an immigration system that visibly works. This is counterintuitive for many advocates, who see enforcement and restriction as the problem rather than part of the solution. But the evidence is clear: the way to expand immigration over time is not to win an argument, but to build a system that earns and sustains public confidence.

The arc of progress is not automatic

The comparison between immigration and same-sex marriage flatters advocates by suggesting that history is already on their side. It implies that the hard work is simply to keep pushing the same message until the laggards catch up. This is comforting. It is also dangerous, because it discourages the much harder work of policy design, coalition-building, and institutional reform that real progress requires.

The gay marriage movement won an extraordinary victory under conditions that do not apply to immigration: an ingroup seeking recognition rather than an outgroup seeking entry, a costless reform rather than one involving real trade-offs, a judicial pathway to finality rather than an endless legislative grind, and a passionate constituency with direct political voice rather than voiceless people abroad.

Immigration advocates do not need a better version of the marriage equality playbook. They need a different playbook entirely, one built around political compromise and designing policies that earn public support by deserving it. The arc of progress on immigration is not automatic. It has to be achieved by hard work.

Share

Disclosure for my Bluesky friends: This post is 100% human-written (assisted with human-verified AI summaries of your critiques of my earlier post). Human mistakes are possible.

Earlier this week I argued that academics need to wake up on AI and offered ten theses on how agentic AI is changing social science research. The post went viral, especially after I revealed it was fully generated and posted by AI, leading to news features, over a million views, and over a thousand (often angry but also enthusiastic) responses across various social media platforms.

Subscribe now

In hindsight, I should have done a few things differently. First, revealing that Claude wrote the original post (even based on my earlier social media writing) as a cheeky follow-up was a mistake. It distracted from the substance and gave critics an easy reason to dismiss the arguments. Rather, I should have been upfront about my basic setup.1 Second, I should have been clear that current agentic AI tools are better at doing most social science research tasks than professors globally.2 This doesn’t mean you necessarily get replaced, but it does mean the nature of your work will change. Third, the AI-generated post had minor, somewhat weird stylistic errors that better human-augmented editing would have caught.

In this respect, I want to highlight Solomon Messing and Joshua Tucker’s Brookings piece, published shortly after mine, which makes a much more persuasive version of many of my arguments—with concrete use examples, no unnecessary provocation or AI-generated text, and a more constructive vision for the future. If my post was too abrasive for you, read theirs instead.

I was deliberately provocative, and I stand by that choice. It backfired in some sense but worked in another sense: dozens if not hundreds of academics are now trying agentic AI tools who would not have otherwise yet. After reading most responses, I certainly changed my mind on a few things, but I am still convinced of my core claim that, because of already existing AI tools, our research workflow will have to change whether you like it or not. Here are ten more theses that came from my reflections.

11. Qualitative research and novel data collection will increase in relative value.

The strongest substantive critique of Part I was that it conflated “research” with the specific tasks AI handles well—literature reviews, data analysis, conceptual synthesis. Several respondents rightly pointed out that AI cannot conduct ethnographic fieldwork, interview detainees in illegal prisons, or spend years building trust with a community. They are absolutely right. My theses were primarily about (the currently dominant) quantitative and conceptual work in social science, and I should have been clearer about that scope.

But the implication is not that qualitative researchers should relax. It is that the relative value of original data collection—fieldwork, interviews, archival work, participant observation—is about to rise. If AI can synthesize existing literature and run standard regressions, then the premium shifts to the things AI cannot do: generating new data that did not previously exist, especially from hard-to-reach contexts. Qualitative researchers and field experimentalists should see this as an opportunity to do more great work they have comparative advantage in instead of transcribing their interviews or compiling literature reviews.

12. Due to “jaggedness,” AI opinions are polarized by beliefs in their utility.

describes AI’s capabilities as a “jagged frontier”—superhuman at some tasks, embarrassingly bad at others, in ways that do not map to human intuition. AI can write a serviceable literature review but struggle with a basic visual puzzle. It can synthesize findings across 500 papers but hallucinate a co-author’s first name.

This jaggedness explains why the AI debate in academia is so polarized. Critics point to the troughs; enthusiasts point to the peaks. Both are right about their corner of the frontier. The overlap with the qualitative-quantitative divide in social science is hard to ignore: researchers whose work involves the tasks AI handles well (data analysis, literature synthesis, pattern recognition) tend to be more positive, while those whose work involves the tasks AI handles poorly (fieldwork, interviews, archival interpretation) tend to be more skeptical.

But I noticed something beyond mere disagreement. Bluesky users who despise AI viscerally were often the first to deny basic, easily verifiable facts—for instance, that it can produce slide decks well. Very few respondents acknowledged that AI capabilities for research are real but worried about their consequences. People either dislike AI and deny any productive use, or like it and exaggerate its utility. Some of this is motivated reasoning—the explicit existential threat of a computer doing things better than you. But I suspect even more of it is simply ignorance about “the other.”

Contact theory is real. If you believe that Claude Code is evil or incompetent, I dare you to install it and use it to organize your research folders or create slide decks for your upcoming conferences. Earlier I encouraged folks to “spend a week with Claude Code.” It should have been “spend a day” (which should be enough).

13. User expertise still vastly determines output quality.

Perhaps not surprisingly, much of the criticism on Bluesky still assumes that using AI means copying and pasting from a chatbot. That is just not how agentic AI works. Agentic AI operates autonomously within your file system, reads and writes code, consults documentation, and executes multi-step research workflows—all guided by detailed instructions you build over time.

One related, common form of AI denial also assumes that because the tool is accessible, anyone could produce the same output. That is like arguing that because everyone has access to a stove, everyone can cook a good meal. There are obvious differences in cooking skills, recipes, and the quality of ingredients.

But the question is not whether AI is better than most professors at doing most important research tasks (I still stand by the assertion that it is), but whether good researchers with AI are better than good researchers without AI (they absolutely are). Honestly, I would take well-prompted AI slop over Bluesky slop (hundreds of anonymous users responding ai/dr whenever they see the feared AI keyword regardless of any substance) any day of the week.

14. Publication lag makes AI capability critiques obsolete by the time they come out.

Here is a problem that almost nobody in the debate acknowledges: academic and book publication timelines are structurally incompatible with AI’s rate of improvement. When someone cites a 2025 paper (initiated in 2024) documenting GPT-4’s hallucination rate to argue against using AI in March 2026, they are citing evidence about a system that no longer exists. It is like citing a 2005 study on flip phone limitations to argue against smartphones. That’s probably why the new “AI Con” book is so bad—clearly outdated before it even hit shelves.

I am not dismissing all of this research itself. The studies are often methodologically sound. But the evidence base expires faster than it can be published, reviewed, and cited. Messing and Tucker’s Brookings piece, published in March 2026 (and reviewed “rapidly” in only two weeks), already documents capabilities that would have seemed speculative six months earlier. By the time a peer-reviewed paper on current AI limitations appears in a journal, the limitations it documents will likely be fixed. This is not a comfortable situation for academics who are trained to rely on published evidence. But it is the situation we are in.

15. Most papers are already mostly read by AI, not humans.

It is an open secret in academia that most published papers are never cited or read by anyone beyond the authors, reviewers, and sometimes the editor. With the coming proliferation of AI-written papers—whether complete slop or not—it will become impossible for researchers to keep up even with their own niche fields. I like to think I am aware of all the new literature on immigration attitudes, but I am probably missing 80% of what gets produced outside the US, Europe, and top disciplinary journals.

This means that academics should accept their primary audience is increasingly LLMs. has been talking about writing for LLMs for some time, but with the ascent of agentic tools, this applies to most academics too—including qualitative researchers whose work itself cannot be automated. I do not have a firm sense of what authors should do about it, but ensuring that a machine-readable version of your paper exists (ideally in .md format) seems like a good first step.

16. AI exposes what was already broken in academia and beyond.

Related, a large number of responses to Part I amounted to: “If AI can do your research, your research was never good.” I agree (LOL)—but that is an indictment of much of social science, not a defense against AI or a smart attack against me personally. The replication crisis, citation padding, p-hacking, and the production of papers no one reads were all pre-existing conditions.

Human-generated academic slop was always pervasive; AI just makes it visible. put this more bluntly in his restack: academic institutions hoard human capital, the tenure system rewards collective navel-gazing over public impact, and most professors could be more useful doing something else. That is a harsh framing. But if only a small percentage of published papers have genuine value, the system AI is disrupting was not exactly thriving.

17. Skill atrophy is a real risk, especially for the future generation of scholars.

This brings us to what I consider another strong reaction to my initial post: that outsourcing cognitive processes like “evaluating sources” and “coding data” damages the researcher’s own understanding. Many folks rightly worry about “reducing complex, thought-driven processes to a series of discrete tasks to be outsourced, when there’s so much that goes on cognitively both between and after the steps.” Messing and Tucker flag the same risk under “skill atrophy.”

I take this seriously, and I concede the risk is real—especially for students and trainees who have not yet internalized the cognitive skills that AI might short-circuit. The researchers who worry about skill atrophy are right that something is lost. But they underestimate what is gained: the ability to operate at a higher level of abstraction, to test more hypotheses, to iterate faster. For established researchers, the risk of atrophy is low because the skills already exist. For students and future researchers, we urgently need to figure something out in updating our grad school curriculum.

18. AI writing detectors and disclosure norms do not work.

AI writing detection tools were bad, are still bad, and will probably remain bad. The original Claude-produced post passed every major AI detector as “100% human” without any elaborate prompting to avoid this on my side. Many critics of my initial post said they immediately “sensed” it was AI-written. But they said this after I revealed the workflow—a textbook case of confirmation bias. Before the reveal, nobody flagged it. In fact, someone even complained I didn’t use AI to write a post boosting AI.3

The more important point is about disclosure incentives. Messing and Tucker recommend standardizing AI usage declarations across fields. I respect their reasoning and the call for standardization (instead of the chaos that we have now), but I disagree that any expansive AI declaration standard can have any merits given the current incentive structure.

Do not get me wrong—people in positions of authority like journal editors should be transparent about their workflow. But for regular authors, voluntary disclosure creates a system where honest users get punished and dishonest users face no consequences. I disclosed my AI workflow and received threats, professional attacks, and calls to fire me. The rational incentive is to lie. “AI usage acknowledgments” sound reasonable, but they collapse on contact with the actual social dynamics of academic life in 2026. Until the professional costs of disclosure drop, mandatory acknowledgment norms will select for dishonesty.

There is also a deeper problem: disclosure norms get the accountability question backwards. For some, AI disclosure can even function as a cop-out—”I used AI, so it’s on you now to figure out if it is slop.” But authors should stand by the final product regardless of how it was produced. If AI introduces an error, that is the author’s responsibility. What matters is whether the work is correct and valuable, not whether a human or a machine typed the sentences.

19. Academic Bluesky is not a serious venue for this debate.

I have to address this because it colored everything that followed. Bluesky generated almost as many reactions as Twitter, but they were overwhelmingly hostile in the least productive way possible. The most common response was some version of “If you didn’t write it, why should I read it?” or “ai/dr.” Many included curses, accusations of being paid by AI companies (?), and calls to not cite my earlier published work (??) or even to fire me (???) with people tagging my employer to replace me with AI since I’m claiming it’s so good.

My original post was provocative. But I did not attack anyone personally. I made arguments about AI and academia, based on my own experience in the field, which you may agree or disagree with. For that, academics on Bluesky responded with professional threats, ad hominem, and coordinated pile-ons. I have thick skin and employment security. I can absorb this.

But most people who might share heterodox views on AI in academia do not have that luxury. They are graduate students, contingent faculty, and junior researchers (in fact, I was one myself just a couple of months ago!) who watch what happened to me and draw the obvious conclusion: keep your mouth shut. That is the real cost of pile-on culture—not to people like me, but to the open exchange of ideas that academia is supposed to protect. And while I appreciated all the sympathetic folks who reached out in DMs, I wish you would speak out publicly. That is the only way this unfortunate dynamic can change.

20. Research can lack “soul” and still serve the public.

Max Kagan articulated and addressed a common concern from Bluesky folks that resonates with me too: the idea that research produced by or with AI lacks something essential—call it soul, craft, or authentic intellectual engagement. The process of struggling with a question, sitting with ambiguity, and slowly building an argument is personally transformative for many scholars. There is a reason people pursue PhDs despite terrible labor market prospects: the work itself is meaningful. When AI compresses that process into hours, something genuinely valuable is lost.

I feel the pull of this. But I am not sure it survives contact with the question of who pays for it. Most academic research is publicly funded. Taxpayers do not fund universities so that professors can self-actualize. They fund universities to produce knowledge that benefits society. If AI-assisted research produces more and better knowledge faster, the public interest argument for embracing it is hard to resist—even if the private experience of research becomes less romantic.

***

All in all, the discourse around Part I was messy. But it was also productive. It probably encouraged a few dozen if not more academics to try agentic AI tools for the first time, so I take it as a win. The strongest substantive objections—hallucination, skill atrophy, qualitative research gaps—forced me to think more carefully about both the risks and the opportunities. The Bluesky pile-on artists just showed yet again that it is not a serious platform for an open exchange of ideas. But the intensity of the reaction only confirms the stakes. People do not argue this fiercely about things that do not matter. Academics are waking up—some enthusiastically, some kicking and screaming. Either way, they are waking up.

Update: I’ve written a Part II with reflections on the responses to this piece and a Part III with ideas about what academics can do next.

Please like, share, comment, and subscribe. It helps grow the newsletter without a financial contribution on your part. Thank you for reading.

This piece is inspired by a wave of recent AI-related writing from people I respect: , , , , , , , Claude (yes!) Blattman, Kevin Bryan, , , Sean Westwood, and many others. So here, I’m continuing the tradition of writing the takes that are upsetting but needed.

I study immigration and public opinion, not AI. But I’ve spent the last few months watching AI transform my own research workflow, and I have some things to say to my colleagues. For the first time in my life, I genuinely do not know what academia will look like in five years.1 Even if progress stalls completely and we are stuck with the current models forever, the changes already in motion will transform my field of academic research and publishing beyond recognition. The status quo is unsustainable. It may take time, because academia is the most dispositionally conservative institution on the planet. But it will change.

Subscribe now

Here are ten theses for my colleagues, most of whom still seem oblivious.

1. AI can already do social science research better than most professors.

This is not hyperbole. Tibor Rutar recently described generating a full research paper using AI prompts alone, producing work he considers publishable in first-quartile journals. Paul Novosad reportedly accomplished similar results in 2-3 hours. claims that Claude can produce a publishable-quality political theory paper in under two hours with minimal feedback. Scott Cunningham estimates that manuscript creation now basically costs roughly $100 in editing services plus a Claude subscription.

And this goes well beyond crunching numbers or running pre-existing Stata code. Yes, what I’m claiming here is that LLMs produce excellent literature reviews and generate fruitful recombinations of existing ideas. Let’s be honest: academics haven’t been particularly great at writing either, and AI can make your ideas far more accessible to the people who actually need them. But effective use requires investment: describes building a ~200-line instruction file encoding his research workflow, judgment calls, and behavioral guardrails. This takes a skill.

2. The academic paper is a dead format walking.

Sean Westwood put it bluntly: “AI does lit reviews better. AI will do peer review. Users will skim AI summaries. The real science is the question, the pre-analysis plan, and the analysis. The 30-page paper is just vestigial wrapping paper.” He got roasted on Bluesky for saying this. But he’s absolutely right, and the backlash proves his point: the field can’t even discuss the obvious without circling the wagons. Arthur Spirling is also right that we need conversations about what a paper is, what “review” means, and the correct role of generative AI. Perhaps it’d be a good thing if AI finally pushes us to move on from a system where universities spend taxpayer money to pay commercial publishers to very slowly produce paywalled PDFs2 with outdated results of publicly funded research.

3. The commercial journal system may not survive this.

Cunningham’s latest piece models the math. If manuscript creation drops to a couple of hours and ~$100, submissions could increase fivefold while journal slots stay fixed. Desk rejection rates would go from ~50% to ~90%. The revenue model collapses. Peer review, already strained, becomes impossible at scale. Kevin Munger makes the case for submission fees, paid reviewers, post-publication review, and LLM-assisted screening. The question is whether journals adapt or get bypassed. My bet is most get bypassed.

4. Academics hold AI to absurd double standards.

Hallucinating content is concerning, and researchers should always verify their sources. But just like with self-driving cars, we need a reference point: human writers have been superficially citing papers based on the abstract for ages. Journals already publish studies with data errors, p-hacked results, and non-replicable findings at alarming rates. One estimate puts the share of genuinely useful published papers at around 4%. An LLM that occasionally hallucinates a citation is competing against a system that routinely produces junk science dressed in enough jargon to pass review. If we applied the same skepticism to human-produced research that we apply to AI outputs, we’d shut down half the journals tomorrow.

5. Junior scholars face the biggest disruption and opportunity.

This is probably bad news for junior academics trying to advance their careers in the middle of this shake-up. Jason Fletcher argues that the strategic logic of tenure hasn’t changed—survive the gate first—but AI fundamentally alters how you get there. Teaching prep costs drop. Data cleaning and debugging get delegated to AI. The bottleneck shifts from execution to verification and original thinking.

Gauti Eggertsson observes that the returns on conceptual thinking and original ideas are now relatively higher compared to technical grunt work. A junior scholar with good ideas and Claude Code can now produce research at a pace that would have required a full lab a few years ago. But so can everyone else, and the evaluation criteria haven’t caught up.3

6. I don’t envision a research assistant role in my workflow anymore.

I still think it’s invaluable to have mentees and co-authors. But their role is changing fast. I’m not going to hire someone to clean data, run regressions, or draft literature reviews when AI does all of it faster and at negligible cost. What I want from collaborators is original thinking, domain expertise, and intellectual challenge. This is a genuine loss for the traditional apprenticeship model, and I don’t have a clean answer for how to replace it. Fletcher’s complementary framework—AI produces initial analyses, human researchers independently replicate from scratch—points in a promising direction. But it’s clear that the trend for increased co-authorship in social sciences, for instance, may reverse very soon.

7. Much of the opposition to AI is status protection dressed up as principle.

I recently wondered on Twitter how much of the distaste for AI telltale signs is basically a new version of grammar policing—people enforcing status markers through language gatekeeping. Kevin Bryan said it plainly: “I get the desire for artisanal, hand-crafted research, with the matrices hand-inverted. But our job is to move the frontier of knowledge, not self-actualization.”

Dan Williams has written persuasively about how highbrow misinformation flourishes inside institutions where nearly everyone shares the same biases. I think something similar is happening with AI denial. Many academics—especially those concentrated on Bluesky4 and, I suspect, those who are completely offline—are in complete denial about what’s already happening. Chris Blattman went from a Claude Code skeptic to building an entire AI workflow toolkit in a matter of weeks. Robert Wright recently hosted Alex Hanna and Emily Bender arguing that LLMs are useless. Smart people claiming that a tool millions find useful is fundamentally broken. This smug attitude is exactly why populists are winning, and it applies to AI denial just as much as to politics.

8. The productive worries are about security and verification.

My challenge for anyone who dismisses AI capabilities: spend one week alone in a room with Claude Code or Codex. Not the chatbot—the agent. Most people still think of AI as a search engine that sometimes makes stuff up. They have no idea what agentic AI systems can do.

Focusing on whether LLMs “truly understand” or produce “real” knowledge is a philosophical indulgence that takes away from the things worth worrying about. How do we verify AI-generated claims at scale? How do we prevent p-hacking? (Andy Hall’s team found that AI agents are surprisingly resistant to sycophantic p-hacking—but can be jailbroken with modest effort.) How do we protect sensitive data when AI tools access institutional repositories? How do we ensure that online survey respondents are real? These are solvable engineering and institutional design problems, the kind that Hollis Robbins calls “last mile” challenges—things that live in the edges of expertise, in the contextual and the unsettled. Debating whether Claude is “really” intelligent is like debating whether a calculator “really” does math while your competitor finishes the problem set.

9. We are about to get much better science.

There are some silver linings, however. On my own turf, immigration: we can now automatically catalogue policy and opinion changes across countries and suggest fixes in real time. We can build algorithms to better match refugees and migrants to destination communities. We can make sure research and evidence are accessible to policymakers and voters who never read an academic journal.

More concretely, Yamil Velez and Patrick Liu have been building AI-generated experimental designs since 2022; tailored Qualtrics experiments can now be created in 15 minutes via prompts. Velez’s work points to something even bigger: AI doesn’t just speed up existing survey methods, it makes entirely new forms of interactive, adaptive surveys possible—designs that would have been impractical to program manually. David Yanagizawa-Drott has taken things further still, launching a project to produce 1,000 economics papers with AI—not as a stunt, but as a stress test of what happens when the cost of generating research drops to near zero.

Non-native English speakers also stand to benefit enormously: researchers in Cairo, Sao Paulo, and Jakarta can now produce prose that reads as well as anything coming out of Cambridge or Stanford. Eggertsson suspects AI will erode the monopoly that top US schools have long enjoyed, since their advantage rested partly on knowledge transmission that is now nearly instantaneous. If you care about democratizing science, this matters more than most of the things universities spend money on.

10. Apart from the doomsday scenarios, AI is genuinely exciting.

Yes, there are real risks. Job displacement for some academics (and most other folks) is not hypothetical. The alignment and safety concerns are genuine, even if unlikely to play out in the worst-case scenarios. I take those seriously and I fear our uncertain future somewhat.

But here’s what I keep coming back to: AI is useful and fun. My sense is the “agentic AI is making us dumb” crowd is probably right about some things. But I’ve also noticed my procrastination bar going up. Instead of doomscrolling, I now slack off by trying side projects in Claude Code. May be the most productive form of non-work there is. I’ve been vibecoding a few pretty exciting projects over the past few weeks. Stay tuned.

The wise Yiqing Xu advises that we should all pause for a month to reassess and redesign our workflow, then resume. I agree. The payoff will be large. Lock yourself in a room with Claude Code and see what happens.

P.S. This post was entirely generated and posted on Substack by agentic AI using my new Claude Code (Opus 4.6) workflow. Make of that what you will.

P.P.S. That is, entirely generated based on my artisanal, hand-crafted human social media posts and thoughts on the topic. So who wrote it, really? You tell me.

We hit 1,000 subscribers this month! Thank you for reading, sharing, and arguing with me in the comments. I started Popular by Design because I think the immigration debate deserves more honesty and less tribalism—and it turns out at least 1,000 of you agree (or at least enjoy disagreeing).

Subscribe now

A few updates: I’m hiring a postdoctoral researcher to join me at Notre Dame’s Keough School to work on politically sustainable immigration. If you know someone working on immigration, public opinion, or policy design who might be a good fit, please send them my way. Immigration folks with a PhD (an unfortunate formal requirement by the admin) are more than welcome to apply. I’m also starting as an associate editor at the Journal of Ethnic and Migration Studies (JEMS)—one of the top journals in the field. Excited to help shape what kind of research gets published and how we evaluate it (at least before AI will change it all).

Since joining ’s ABI fellowship, this month I published more pieces than ever before. Hopefully, I will be able to continue the streak at least until the end of my sabbatical. First, “Reflections on the Uncomfortable Truths about Immigration”—my attempt to address the most frequently asked questions about my earlier “Uncomfortable Truths” post. Second, “Student Migration Is Popular... Until It Isn’t”, on what went wrong in Canada, Australia, and elsewhere when universities started selling immigration status instead of education. Third, “Western Countries Do Not Need Immigration”—a deliberately provocative title for an argument that cuts in a surprising direction.

Before we get to the links, I have a favor to ask. I’m curious who’s actually reading this newsletter. We’re now roughly at the population of a small medieval village, which seems appropriate for some basic ideological census:

OK, here are the February links (linking does not imply endorsement):

• I normally don’t follow the Olympics, but the restrictionists’ reaction to Alysa Liu has been fascinating to watch. Liberals thought MAGA celebrating her Olympic gold was a gotcha, but it’s actually a concession I’m happy to accept: no country needs immigration, but the countries that choose it wisely end up stronger. An American daughter of a Chinese dissident draped in the flag on the Olympic podium while the authoritarian government that persecuted her father watches—that’s a recruitment ad for every talented person in the world.

• has a thoughtful piece on what a liberal immigration enforcement regime might look like. Most voters aren’t categorical restrictionists—they can support freer immigration if they believe the system is orderly and serves the national interest. The uncomfortable truth for many liberals: public opinion does support deporting most undocumented immigrants, not just criminals.

  • also compiled a great selection of the latest immigration enforcement research—my reading list is now complete.

  • I often disagree with on how to interpret public opinion on immigration, but his work this month on asking people concrete questions about enforcement is valuable. Abstract attitudes are one thing; specific policy preferences are another.

  • But here’s the uncomfortable truth for conservatives. published new polling on immigration attitudes among Trump voters. Some of their findings surprised even me: 90% of Trump voters agree that “properly controlled immigration can be good for America,” and 70% want it to be easier to immigrate legally (but harder to come illegally). More room for agreement than usually assumed.

• has a fascinating conversation with Ruud Koopmans on the differences between immigrant integration in Europe and the US. I’ve been beating this drum for a while: immigration is not one thing that has uniform effects—policies and contexts matter enormously.

  • makes a related argument I’ve been pushing for years: you can’t just import conclusions from European immigration debates to the US, or vice versa. Showing a Danish crime chart says nothing about how well America integrates migrants.

• explains how Trump botched immigration. What we’re seeing is textbook thermostatic reaction to government overreach—voters still support deportation in the abstract but strongly disapprove of how it’s actually being carried out.

  • Case in point: the administration tried suspending TSA PreCheck and Global Entry, then caved within 24 hours after backlash. It’s honestly a bit funny, if not sad, that the trusted traveler programs are the thing that finally breaks it for normal folks who happened to vote Republican.

• has several related pieces worth reading. First, shares new data on the trans rights backlash—I haven’t seen much from social scientists on what’s driving it. Is it thermostatic? Elite cue-taking? Something else? One of the most dramatic opinion shifts in recent memory, and the field has sat it out.

  • Second, on why thoughtless moderation is a mistake—voters aren’t dumb or unnecessarily cruel. Being thoughtlessly harsh on trans rights or immigration doesn’t win elections.

  • has a great conversation at The Argument about the end of conservatism. His point on immigration: the current conundrums have nothing to do with big ideological debates about what we owe each other and foreigners. Trump’s “own, concrete choices, not just the spirit of nationalism, have led his administration to become very unpopular.”

• has a beautiful piece on Tokyo as a megacity at human scale. The “city of doorways, not vistas” framing is great. After living there, I came away with a similar sense that the intimacy is mostly about zoning and land use, not some mysterious cultural essence. Where I’d push further is toward what happens when demographics undermine that vibrancy—outside greater Tokyo, depopulation is devastating.

• has a sobering piece on why comparativists should speak up. As someone who ended up studying US politics from a comparative perspective, I’m struck by how often the “it can’t happen here” instinct comes from people who simply haven’t looked at how things have unfolded elsewhere.

• came out against white culture. It’s kinda funny since this is pretty much exactly what the critical studies people have been saying all along, just without the unnecessary jargon. But well…sometimes it takes a different messenger.

• pushes back on the idea that immigration will fix fiscal problems. at Cato responds with a different set of assumptions that flip the results dramatically. My sense is that fiscal impacts are genuinely hard to estimate, and reasonable people can disagree on assumptions. But I’ve definitely learned from both Cato’s and Manhattan Institute’s work on this, and I wish more of our immigration debates were this technical rather than vibes-based.

• and at EIG found major data errors in George Borjas’s paper making the case for a $100,000 H-1B fee. Ozimek’s broader point is one I keep returning to: you have to do the policy right. Many people who think they understand high-skilled immigration actually don’t understand the economics or the mechanism design.

• won his tariff case at the Supreme Court—a 6-3 decision challenging presidential tariff power. A Soviet-born constitutional scholar turning ideas into action by defending limits on executive authority. Worth reading his initial thoughts at Reason and his inspirational profile at IHS.

• has a new paper using survey evidence with a conjoint experiment to test how state-level immigrant integration policy features affect perceptions of fairness and support. Important new evidence that US attitudes are much more inclusive than conventional debates suggest—but this support is conditional on policy design. People are more supportive of integration when it includes social support and clear eligibility criteria, which they view as fair. Another reminder that how you design and implement policies probably matters more than how you talk about it.

As before, if you want me to write more about one of these or other related topics, let me know!

Subscribe now

Many folks told me my latest post challenging the pro-immigration orthodoxy was like a breath of fresh air. To keep with the theme of radical honesty, I believe we also need to reflect on whether countries need foreigners in the first place.

Let’s be honest with ourselves: no Western country will collapse without immigration. The United States is a powerful, functioning state. So are Japan, Germany, France, and most of Europe. The lights will stay on. The trains will run. The US Super Bowl will be just fine without foreigners singing in Spanish.1

Pro-immigration advocates who claim otherwise—who insist that immigration is “the only politically feasible solution to population collapse”, that normal GDP growth will be “impossible” without sustained immigration flows, that no one else will take care of your elderly parents—are overstating the case. And in doing so, they are losing credibility with the very people they need to persuade. When you tell someone their country cannot survive without immigration and they look around and see it surviving just fine, you have not made an argument. You have made yourself easy to dismiss.

So let me start where immigration skeptics start and explain why no country needs immigration. But in the spirit of radical honesty, I’d ask my restrictionist friends to return the favor—and follow their own argument to its logical conclusion.

The case for no immigration, taken seriously

One of the common good-faith conservative arguments against immigration is not about its effect on crime or culture—it is about dependency. As critics at outlets like The American Conservative have argued, wealthy countries have become “addicted to cheap labor.” If an economy cannot function without constantly importing foreign workers, maybe the economy is broken, not understaffed. Maybe the answer is automation, higher wages, and policies that get native-born men—millions of whom have dropped out of the labor force—back into productive work. Maybe immigration is a crutch that lets governments avoid harder structural reforms. This is a serious argument, and it deserves a serious answer.

But first, it deserves an honest concession: you can build a rich, functional country without much immigration. Consider Japan. In the early 1970s, Japan had a population of more than 100 million people and virtually no immigration. Over the next two decades, it built the world’s second-largest economy through domestic investment, export-driven manufacturing, and a disciplined, highly educated workforce. By 1995, Japan’s GDP per capita was among the highest in the world. No immigrants needed.

Or consider Sweden and Canada in 1900—two countries with almost identical populations of roughly five million people. Sweden was actually losing people: between 1850 and 1930, about 1.3 million Swedes—a third of the population—emigrated to the United States. Yet Sweden went on to build one of the world’s most admired welfare states. It industrialized, innovated, and became synonymous with quality of life—all without relying on large-scale immigration until the very end of the twentieth century.

Sweden also tells a different cautionary story. After decades of prosperity built on a homogeneous welfare state, Sweden began accepting large numbers of humanitarian migrants in the 1990s and 2000s. The result has been among the worst integration outcomes in the OECD: foreign-born residents face an employment gap of over 20 percentage points compared to natives, non-European immigrants earn 20 to 30 percent less even after decades in the country, and someone has to pay for it. Not surprisingly, the Sweden Democrats—an anti-immigration party—rose to become the second-largest party in parliament in one of the most cosmopolitan countries in the world. If you want an argument that not all immigration is beneficial, Sweden hands it to you on a silver platter.

So the restrictionist premise is correct. You can have a prosperous, well-governed country without letting foreigners in. Japan proved it. Even when you do embrace immigration, it can go badly if you do it wrong. Sweden proved it. Immigration skeptics are not crazy. On the basic facts, they have a point.

The question is what happens next

Japan’s population peaked at 128 million in 2008 and has been falling ever since. Today it stands at about 123 million. By 2070, demographers project it will drop below 90 million. Japan’s economy, once the world’s second-largest, slipped to fourth in 2023, overtaken by Germany—partly due to currency effects, but also reflecting decades of stagnation that economists consistently connect to demographic decline.

Canada took a different path. Starting from the same base of five million as Sweden in 1900, Canada chose relative openness. It built an immigration system—imperfect, sometimes messy, as I have written about—but one that consistently welcomed newcomers. Today, Canada’s population has grown to over 41 million, more than four times Sweden’s. The exact surplus to native-born Canadians from all that immigration can be debated and is likely modest per capita. But without relatively open immigration, Canada would be a much smaller, less influential country than it is today—and heading down the same demographic path as Japan.

Now consider the United States. When Matt Yglesias proposed “One Billion Americans,” many on the right thought he wasn’t serious. But in 1800, the United States had just 5.3 million people—smaller than Sweden is today. If someone had argued then for “100 million Americans,” they would have sounded equally delusional. The country got there—and then tripled that number—largely through immigration. According to the National Academies, most Americans today descend from immigrants who arrived after the nation’s founding. Without those arrivals, the United States would not have had the population to industrialize, settle a continent, or become the dominant power of the twentieth century. The notion that America can simply close the door and remain what it is—that is the truly radical position.

Stasis is not stability

Here is what immigration skeptics get wrong: they confuse the absence of collapse with the presence of thriving. Countries without immigration don’t stay the same. They age and, now they also shrink. They lose fiscal capacity. They still get “woke.” And, yes, they quietly start opening the very door they swore they would keep shut.

Japan is the clearest case. The country that proved you don’t need immigration has 2.57 million foreign workers—a record high, nearly triple the number from a decade ago, and growing at double-digit rates every year. Japan recently scrapped its controversial Technical Intern Training Program and replaced it with a new system designed to attract and retain skilled foreign workers, setting a target to admit even more. The government does not call this “immigration,” of course—Japan has never been comfortable with the word.2 But whatever you call it, the country that needed no one is now competing globally for foreign labor.

Every retiree in Japan is now supported by roughly two working-age people, and that ratio is projected to worsen to less than 1.5 by 2060. Hospitals need nurses. Construction sites need workers. Someone has to care for the elderly, staff convenience stores, and contribute to the pension system. The Japanese government looked at the math and decided that ideological purity was a luxury it could no longer afford. Despite all the amusing hysteria about having a new ultraconservative prime minister, Sanae Takaichi, her government is planning to admit over 1.2 million foreign workers under new visa programs—because the math doesn’t care about your ideology.

The question is whether we want immigration

The word “need” has been doing enormous work in this debate, and it is time to retire it. No country “needs” immigration in the same way that no country “needs” international trade. Or universities. Or highways. A nation can exist without any of these things. North Korea basically exists without trade. Some countries have gutted their university systems and survived. You could stop building roads tomorrow and the state would endure—at least for a while.

But no serious person argues against trade by saying “we don’t need it.” The question is whether trade makes you better off. The same logic applies to immigration. The question is not whether your country can survive without it. The question is whether you want growth, innovation, fiscal solvency, and demographic vitality—or whether you prefer to manage decline.

Here is what strikes me most about the “we don’t need immigration” position: even if you accept every conservative premise—enforce the border strictly, be very selective, prioritize fiscal impact, demand almost complete assimilation, put the national interest first—you do not land on zero immigration. You still land on a lot of foreigners coming every single day for life.

And when restrictionists call for “zero immigration” or a moratorium or a “pause until we figure out what’s going on”—what does that actually mean in practice? Does it mean telling your buddy he can’t bring his wife home from Canada? Would you look a fellow American in the eye and say the government forbids him from living with the person he married? Would you tell a hospital already short on nurses that they need to start rationing care for seniors because hiring a qualified foreign nurse is off the table? Because that is who we are actually talking about.

The reality of migration on restrictionist terms

Immigration in practice is not just dangerous men sneaking across the border or businesses “importing” cheap labor in droves—it is an American citizen waiting years to reunite with a spouse, a rural hospital trying to stay open, a university lab trying to keep its best researcher. Even Donald Trump sponsored a foreign spouse—twice. If the policy you are proposing would not have let the president bring his own wife to the country, maybe it is time to revisit the policy.

If you believe in national strength, you should want the world’s best doctors, engineers, and researchers competing to come to your country. If you believe in family values, it is worth asking why the U.S. makes it agonizingly difficult for American citizens—including white, native-born Americans—to bring their foreign-born spouses home. If you believe in fiscal responsibility, the actuarial case for working-age immigrants paying into Social Security is straightforward. If you believe in national sovereignty, you should want a legal immigration system that works, so that people have lawful alternatives to crossing the border illegally.

Even Richard Hanania, who is hardly a bleeding heart progressive, has argued that opposing high-skill immigration is flatly irrational—pointing out that 46 percent of Fortune 500 companies were founded by immigrants or their children, and that restricting elite talent harms the country far more than any conceivable benefit from keeping people out. This is the argument that follows from taking national interest seriously. This is the logic of any competitive sports team: you want the best players regardless of where they come from. National strength works the same way. If you are serious about greatness, you recruit talent—you don’t turn it away.

I would like to hear an immigration restrictionist describe, concretely, the immigration policy they would actually be happy with. Not “less immigration” or “mass deportation now” as a slogan—a specific system. Who gets in? Through what channels? With what requirements? My prediction is that any honest answer to that question looks a lot like substantial, well-designed immigration—a points-based system, employer sponsorship, family reunification for immediate relatives, and yes, some humanitarian admissions. In other words, something not so different from what most mainstream economists and policy analysts recommend already.

The debate was never really about whether to have immigration. It was about how much, what kind, and how well-managed. That is a reasonable debate worth having—and one that pro-immigration advocates should welcome rather than fear.

No country “needs” immigration, but smart countries can choose it

No country will collapse without immigration. But the countries that chose it—thoughtfully, selectively, with an eye toward visible public benefit—grew larger, richer, and more dynamic. The countries that avoided it are now scrambling to reverse course before the math and the demographic reality catches up with them.

Immigration is not a necessity. It is an advantage—and right now, it is an advantage that is unusually easy to take. Hundreds of millions of people around the world want to move to wealthy democracies. That will not always be the case. Global population is projected to start declining within a few decades, and when it does, the competition for immigrants will get much fiercer. Countries that build good immigration systems now will have a head start. Countries that wait may find there is no one left to recruit.

The restrictionists are right that no country needs immigration. But they are wrong about what follows. What follows is not a reason for complacency—it is a reason for ambition. The smart move, on their own terms and by their own metrics, is to build an immigration system that actually works. Not because the country will collapse without one, but because the country that gets this right will be larger, richer, and stronger than the one that doesn’t.

Surely, some people would choose to be poorer and smaller rather than accept any immigration at all. I do not think most people would—not even most hard conservatives. And if you do, you don’t have to be a “white nationalist” to make that choice unlike what some critics say.

But please be straight with the rest of us: admit that you are fine with a shrinking economy, that you want native-born Americans picking strawberries at $50 an hour rather than learning a skill, and that you would rather manage decline than compete for the world’s best talent. That is a coherent position. It is just not a popular one—and the radical honesty I am asking for here should apply equally to the cosmopolitan left that pretends countries will collapse without immigration and the nationalist right that pretends they will be just fine without it. The real work is in the details—and proposals like, for example, this one show that serious restrictionists and other reformers may be closer to agreement than either side admits.

I'm happy to share my forthcoming piece in the International Higher Education journal, reprinted with permission. International student migration has long been one of the most popular forms of mobility across borders—yet some countries have moved sharply to restrict it. This article applies the lessons on "persuasion by better policy design" from In Our Interest to explain why student migration usually works politically, and how bad policies can destroy that consensus.

Subscribe now

A cycle of backlash and counterbacklash to immigration is reshaping global politics. Yet one form of immigration has long enjoyed remarkably broad support: international student migration. Students pay tuition, fill classrooms, boost local economies, and many stay to become skilled workers. In most democracies, the public has viewed international students favorably—even when attitudes toward immigration in general have soured.

That consensus, however, is now fraying. Canada, the United Kingdom, and Australia—three of the world’s top destinations for international students—have all moved to restrict student immigration in the past several years. What happened? And what does it tell us about how democracies can manage immigration more effectively?

Research across democracies shows that voters largely care about their compatriots and prefer immigration policies that benefit their countries. Public support for immigration rises when policies are “demonstrably beneficial”—when ordinary citizens can see, in practical terms, how immigration serves the national interest.

Most voters are neither unconditionally hostile nor unconditionally welcoming toward immigration. The vast majority hold conditional preferences, supporting it when they believe the system is working and opposing it when they do not. But persuasion through better messaging alone does not cut it—what wins voters’ trust are better policies.

Why student migration (mostly) works

International student migration is a powerful illustration of this framework. Student migration is overwhelmingly popular. Its popularity stems from the fact that students bring money into publicly financed universities, reinvigorate the communities where they study, and are expected to be skilled after they graduate. Interestingly, the most prominent concern people have about international students is not about their impact on the host country but about the possibility that students may return home rather than staying to contribute.

In the United States, international students contribute over $40 billion to the economy annually. In the UK, Canada, and Australia, international tuition effectively subsidizes the cost of education for domestic students. Beyond revenue, those who stay after graduation contribute to innovation and fill labor shortages. Those who return home create lasting networks and spread the host country’s culture. This combination of economic contribution and institutional orderliness—students come through a legal channel with clear gatekeeping by universities—makes student migration intuitively appealing across the political spectrum, much like skilled work migration more broadly.

When the consensus breaks

Canada offers the most dramatic cautionary tale. Its international student population roughly tripled in a decade, exceeding one million by 2023. Much of this growth was driven not by selective universities but by colleges—including many that Canada’s own immigration minister labeled “diploma mills”—that enrolled students in low-quality programs where the primary value was a post-graduation work permit and pathway to permanent residence, not the education itself.

When the product being sold becomes immigration status rather than education, the demonstrable benefits of student migration evaporate. Students were paying high fees for programs with minimal instruction, living in overcrowded housing in suburbs like Brampton and Surrey, and working multiple part-time jobs with poor employment prospects. Local communities bore visible costs—housing pressure, strained infrastructure—without seeing corresponding benefits.

Public support for immigration—previously a Canadian point of pride—plummeted in what observers described as the sharpest shift in Canadian immigration attitudes in the history of the country’s polling. As a result, the Canadian government decided to impose a cap on new study permits in 2024, which helped but didn’t fully resolve the situation or fully win people’s trust back.

Similar dynamics unfolded in the United Kingdom and Australia, where rapid growth in student numbers—amplified by dependant visa surge in the UK and a poorly regulated vocational education sector in Australia—eroded public trust in the student migration system. In both countries, governments moved to tighten restrictions, and the political debate shifted from whether international students were welcome to whether the system was out of control

In all three countries, the backlash follows a pattern consistent with the demonstrably beneficial framework. Student migration became politically toxic not because voters suddenly turned against education or foreign students, but because policy design failures—diploma mills in Canada, the dependant loophole in the UK, the unregulated vocational sector in Australia—severed the link between student migration and visible public benefits. When students come for education and stay for skills, it works. When the education system becomes a backdoor immigration channel, trust collapses.

The curious case of Germany

Germany offers a striking contrast—but perhaps a fragile one. Its public universities provide effectively free higher education to all students, including those from outside the EU—a taxpayer subsidy that might seem like a political flashpoint. Yet international students in Germany have so far generated comparatively little controversy.

German universities still maintain rigorous academic standards with no large, poorly regulated private college sector gaming the system—though a growing private sector increasingly serving international students bears monitoring. The post-study pathway ties continued residence to securing qualified employment. The 2023 Skilled Immigration Act even expanded work opportunities for foreign graduates—framed not as an immigration concession but as an economic competitiveness strategy to address Germany’s well-documented skilled worker shortage. And because domestic students also pay no tuition, international students are not perceived as receiving a special deal.

Germany’s stability, however, should not be mistaken for inevitability. If German universities—or a parallel private sector—were to begin using degree programs primarily as immigration pathways for foreigners, outside of democratic oversight and labor market alignment, the same erosion of trust could follow. The AfD’s rising anti-immigration platform has not yet targeted the free tuition consensus for foreign students, but that does not mean it won’t—especially if policy failures give it an opening. The lesson is not that Germany has found a permanent solution but that its system currently maintains the conditions under which student migration remains demonstrably beneficial: genuine educational quality, labor market linkage, gradual growth, and a framing that emphasizes mutual benefit.

What this means for higher education

For higher education professionals, the central lesson is not to take the popularity of international students for granted. The public support that student migration has traditionally enjoyed is conditional—it depends on the system working to benefit citizens, alongside students, as advertised. When universities or governments prioritize enrollment numbers and revenue over educational quality and labor market alignment, or take the role of immigration admissions, they undermine the very foundation of that support.

The backlash in Canada, the UK, and Australia is not evidence that student migration is inherently unpopular or that xenophobia permeates everywhere. It is evidence that badly designed student migration policies become unpopular—a distinction with enormous practical implications.

Obviously, replicating Germany’s tuition model is not feasible in most countries, where international student fees effectively subsidize domestic education. But governments do have design levers available: robust accreditation that shuts down programs functioning primarily as immigration pathways, post-study work rights tied to qualified employment rather than granted automatically, and enrollment growth calibrated to housing and labor market capacity. None of these require eliminating the revenue benefits of international students. They require making sure the revenue model does not swallow the educational one.

International student migration can remain popular by design, but only if higher education systems and governments do the work of maintaining quality, transparency, and visible benefit. The countries that get this right will attract global talent, strengthen their universities, and build durable public support. The countries that do not will discover—as Canada, the UK, and Australia already have—that even the most popular form of immigration can become toxic when voters conclude the system is no longer working in their interest.

The response to my recent piece on “highbrow” misinformation has been overwhelming. The piece argued that pro-immigration advocates, academics, and fact-checkers routinely make claims about immigration that are technically defensible but often misleading. I got a fair share of support and hate emails from across the political spectrum—which, I suppose, is one way to know you’ve touched a nerve. But I was also particularly heartened by the public approval the piece received from academics, including left-of-center scholars for whom endorsing a piece that challenges pro-immigration orthodoxy carries real reputational costs. Tenured (and untenured) professors should do this more often.1

Subscribe now

What I learned

Here is what I learned from the comments and reactions across platforms. First, the piece resonated with lots of folks who haven’t thought about immigration before at all. The reason for that is that the pattern of strategic half-truths and noble lies that and others describe as “highbrow misinformation” is hardly unique to the immigration debate. Commenters pointed out eerie parallels in gun policy, climate, public health, and more.

Second, it’s hard to please everyone—but I was struck by how the piece also resonated with a broad spectrum of ideologies. Some of the most thoughtful responses came from people who disagree with each other on nearly everything else. The post was not meant to convince everyone. Different readers will find different claims more or less persuasive, and that is fine. After all, my goal was never to smear any particular individual or organization—it was to call out and do something about the misinformation dynamic that erodes public trust across the board.

What I should have done differently

I stand behind what I wrote. One thing I do wish I had done differently, however, is less throat-clearing. While some of it was probably necessary—and I say this as someone who already cut the throat-clearing in half from the original draft—it was still not sufficient to prevent people from misinterpreting or outright yelling at me. Quite a few readers ignored most of it, found the part they objected to, and ignored the caveats anyway. So it goes.

More importantly, while it is simply not possible to cover all myths and misinformation instances in a single piece, I wish I had given at least a few more specific examples besides the Oxford literature review stating, with a broad stroke of the brush, that “immigrants commit fewer crimes worldwide.” So let me use this space to do what I should have done in the original, since it was probably the most common complaint among those on the pro-immigration side.

So, let’s take crime again and consider how confidently some prominent voices state things that are, at best, misleading oversimplifications.2 For example, Hein de Haas, a prominent left-of-center academic, in the PR materials for his widely read book How Migration Really Works, summarizes: “There is no evidence that immigration leads to more crime. In fact, crime rates have dropped as immigration has increased.” I genuinely admire de Haas’s original research (like this paper on the effectiveness of immigration policies)—but this kind of confident, sweeping summary is precisely the problem.

While immigration does not generally increase crime, context matters enormously: in the United States, immigrants commit far less crime per capita than native-born citizens, but this is not universal. In several European countries, including Sweden, I wrote at length that foreign-born individuals are disproportionately represented in the prison population, particularly where rapid immigration of young, unskilled males intersects with labor market discrimination. These sorts of sloppy generalizations happen even to the best of us, but in a high-profile book marketed to the general public, it becomes highbrow misinformation.

Here is another representative example of fighting “lowbrow” misinformation with “highbrow” misinformation. When trying to “debunk” another random thing that Trump said at a rally, FactCheck.org quoted Swedish criminologist Jerzy Sarnecki describing claims linking immigration to rising crime in Sweden as “lies”—while acknowledging that Sweden’s large refugee intake creates “various types of strains.” But Sarnecki maintained that the increase in lethal violence “has nothing to do with the recent large refugee wave,” despite the fact that the Swedish government reports finding overrepresentation of foreign-born individuals in crime statistics.

The same pattern also often applies to the mainstream description of immigration’s fiscal impacts—the blanket claim that “immigrants are net contributors” depends enormously on the skill and age composition of immigrant flows, the generosity of the social welfare system, and the time horizon you choose. Saying “immigrants are net contributors” without these qualifications is not just incomplete. In many European welfare states with large humanitarian intakes, it is simply not true. With those additional examples on the table, let me turn to what I learned from the comments themselves.

Comment highlights

More generally, the comments section on the original piece was among the most substantive I’ve seen when it comes to a public piece about immigration issues—over a hundred responses, many of them long and thoughtful. The piece was discussed on Substack, Twitter, LinkedIn, Bluesky, and Reddit—with strikingly different reactions depending on the platform. Here are a few that stood out, along with my brief reactions.

On Substack, drew a sharp parallel to gun policy, where “gun violence” is routinely defined to include suicides, which inflates the correlation with gun ownership and, as he put it, “gets in the way of building consensus for policies that would actually have an impact on gun homicides, such as safe storage laws and owner liability.” This is a perfect example of how strategically inclusive definitions—a form of highbrow misinformation—can undermine precisely the policies their proponents claim to support. As I noted in my original piece, has also written on similar dynamics within the climate debate.

User offered a succinct list of things “everyone can agree on”: high-skilled immigration is almost always good; the Right’s discussion on immigration tilts towards xenophobia; the Left often gaslights normal people on immigration by framing it as overwhelmingly good; low-skilled immigration can be good but tilts bad in Europe due to generous welfare systems; immigrants who commit violent crimes should be deported if possible. I thought this was a good consensus summary—though I imagine most people would still disagree with at least one or two of these points depending on where they come from politically. Which is precisely the point: even a reasonable attempt at common ground will leave some people unsatisfied.

agreed that pro-immigration types should be more ambitious and not knee-jerk defend current policies—but disagreed on the value of acknowledging tradeoffs. His argument: nobody in politics ever talks about tradeoffs for their preferred policies, because that’s “political suicide.” This is probably the strongest critique of my piece, and it deserves a serious answer.

Hanania is right that politicians rarely volunteer the downsides of their own agenda. But I think the relevant audience for my call for honesty is not necessarily politicians—it’s researchers, advocates, and communicators who shape the information environment that politicians respond to. And the cost of not being honest is compounding. As one commenter put it, telling only half the story year after year eventually backfires, because people experiencing the downside of tradeoffs are not blind. Ignoring their experience doesn’t make it disappear; it just makes the messenger look dishonest.

Besides, it can also leave the messengers themselves misinformed. I have met immigration scholars—people who study this for a living—who had no idea that foreign-born individuals are significantly overrepresented in crime statistics across several European countries. I’m not sure I fully understood it myself until halfway through graduate school. If the experts don’t know the basic facts, the information environment has a problem that goes beyond spin.

On the brighter side, , a self-described “open borders guy,” acknowledged that it’s “not exactly a secret in working-class America” that competition with employers who hire undocumented workers at low rates puts legal businesses at a disadvantage. He referenced roofers, restaurants, and housing competition.

What followed was remarkable. One commenter called him “the first pro-open borders person I’ve ever come across who openly acknowledges that working-class people are economically hurt by low wage competition from immigrants.” Mitchell fired back: “Telling people that they’re bigots because they actually trust their M1A1 Eyeballs is profoundly counter-productive.”

The thread got heated from there—but the core exchange is telling. When open-borders advocates struggle to say what Mitchell said, something has gone wrong with how one side of this debate communicates. Just like I want immigration-skeptical folks to acknowledge the trade-offs of restricting immigration, I also want all of us in the pro-immigration space to follow Russ and be able to admit at least one thing our side is getting wrong. Anything. Really. Please!

On Twitter, the piece reached its widest audience. quote-tweeted it approvingly—”Unusual honesty from immigration academic reveals how elite misinformation on sacred topics works”—and that post alone reached over 300,000 views. Philosopher Nevin Climenhaga found the concept of “highbrow misinformation” helpful and shared a related formulation from philosopher Rishi Joshi, who defends immigration restrictions: “Immigrants don’t come from immigrant-land.”

On the critical side, user Dion, among many other folks I respect like and Stan Veuger, argued the piece “would have been more convincing if you cited examples of people expressing the views you criticize”—a fair point I’m trying to address above.

On Bluesky, the reaction was more revealing. A handful of replies to my own post ranged from substantive critique—user named Charles raised an interesting charge of inconsistency in how I treated normative vs. empirical claims—to dismissive ad hominem of me being a “white man” that Bluesky’s own system flagged as “rude.” One reply even argued that the piece was itself an example of “highbrow misinformation” for not providing exact figures of how many academics believe or say certain things.

But the most telling feature was the silence. Bluesky was the only platform where the piece did not travel much beyond my own followers—no organic sharing, no discussion threads. When a piece generating hundreds of substantive comments elsewhere barely registers in one space, that says something about the information environment there.

On LinkedIn, the reception was more measured and constructive. I particularly liked Justin Schon’s point that there exists an asymmetry where “the burden of proof seems to fall on people to prove positive effects” of immigration, while negative claims face lower evidentiary standards. I think he’s right—but part of what I was trying to show is that the asymmetry can run in both directions depending on the audience.

FAQ

Are well-intentioned, misleading claims and omitted regression tables really misinformation?

Some commenters, including those who generally agreed with the piece, pushed back on how I treat the term “misinformation.” In the original piece, I relied heavily on Dan Williams’s concept of “highbrow misinformation”: claims that aren’t technically false but are strategically framed to mislead by omitting important context or presenting contested findings as settled.

’s recent piece on misinformation research makes a related but distinct point: that the entire field of misinformation studies has become overly procedural, measuring outputs (claims fact-checked, labels applied) rather than outcomes (beliefs changed, harms reduced). As Garimella notes, determining what’s “misleading” ultimately requires political rather than scientific judgments—which is why the infrastructure of fact-checking tends to focus on some types of misinformation more than others. This resonates with what I was trying to get at.

What strikes me is the dynamic we seem to be in: “it’s not misinformation unless it comes from the right.” As we see from the factcheck.org example, the infrastructure of content moderation and media literacy is overwhelmingly aimed at one direction. But as I tried to show in my piece, highbrow misinformation—the kind that comes from elites, academics, and well-meaning advocates—can be just as damaging to public trust, and it is far less scrutinized.

Is it really helpful to post this now, given everything that is going on?

I believe it is! There is never perfect timing for anything, but I should also note that I’ve been working on and sitting on this piece for quite some time—the first draft was finished in December.

As has argued, the real scarcity in our intellectual environment isn’t information or good analysis, but courage. She describes academics who privately agree with heterodox positions but won’t say so publicly because the career calculus makes silence rational. That dynamic helps explain what I found: not a conspiracy of lies, but a slow accumulation of strategic silences that leaves the public conversation distorted.

I can’t control how people use my piece. What I can do is make sure that what I say is accurate to the best of my knowledge. If someone cites it—and some people did, in fact—by saying that “this liberal professor acknowledges that immigration is not good,” I don’t think that’s persuading moderates to become anti-immigration. But it does increase the chance that some of them will actually read the piece and get exposed to the genuine pro-immigration arguments I make—like the evidence on increased productivity, the benefits of skilled immigration, and the case for demonstrably beneficial policies that can actually win public support.

More broadly, several commenters—from very different ideological starting points—raised the question of whether intellectual honesty is even a viable strategy in a polarized information environment. If one side tells the truth about tradeoffs and the other doesn’t, does the honest side lose? I think this is one of the most important questions in public discourse right now, and I don’t think the answer is as bleak as some fear.

The whole reason the piece resonated is that people are hungry for honest analysis. Two-sided arguments are usually more persuasive, not less—especially when audiences are already skeptical. And the cost of dishonesty is compounding: every time an advocate makes a claim that voters can see through, the credibility of the entire pro-immigration project erodes a little more.

Was I using hyperbolic language?

Perhaps, but I don’t see anything that I have gotten factually wrong. “What elites don’t want you to know” might have a populist flair, and I can acknowledge that. It was a deliberate choice to signal that this piece was not going to be a typical academic exercise in hedging (which many people still accused me of anyway). But the substance behind the rhetoric stands: the examples I gave are real, the research I cited is accurate, and the pattern I described—strategic omission and overclaiming by pro-immigration advocates—is well-documented.

If anyone can point to a specific factual error, I’m genuinely interested. So far, the pushback has been more about framing and tone than about the underlying claims.

Why did you like the comment or repost from someone I don’t like?

Likes are not full endorsements. As someone noted on Reddit, I liked most of the main-branch comments—even ones where I disagreed and pushed back on the message. The reason is simple: I appreciate thoughtful, respectful responses that engage with the substance of what I wrote. The bar is not “I agree with everything this person says or has ever said.” The bar is: “Did this person take the time to write something that wasn’t just a knee-jerk reaction—not just ‘immigration good’ or ‘immigration bad’?” If so, they got a like. I think that’s a reasonable standard, and I intend to keep it.

Aren’t you strawmanning pro-immigration advocates? I haven’t heard anyone even say “Immigration is ALWAYS good for everyone”

Folks, the subheads were not supposed to be literal things that people say! Of course, nobody walks around saying “immigration is good for everyone, everywhere, all at once”—but a great many people act as though or imply that’s their position when they dismiss every piece of evidence that complicates the picture.

It’s been interesting to observe the split reaction. Some people—mostly on the left—said I was strawmanning them, exaggerating the problem, or outright lying. Some other people—mostly in the center—said they feel seen and that everything I wrote is basically a truism. Both reactions happened in response to the exact same points. This meme post from captures this dynamic pretty well:

I think what is going on is that many people fail ’s “ideological Turing test“—they cannot accurately describe how the other side sees their arguments, even at a basic level. When centrist readers tell me that everything in my piece is obvious, and left-leaning readers tell me I’m making things up, the most parsimonious explanation is not that one group is right and the other is wrong. It’s that they live in different information environments—and the people who think I’m strawmanning might benefit from spending more time in the environments where these “strawmen” are, in fact, the conventional wisdom.

Coming next

In follow-up posts, I plan to engage with the most important—and possibly good-faith—criticisms from both the left and the right. Here is a preview of what I’m considering:

Selection by origin (a right-leaning criticism): Some commenters asked why I didn’t address “the elephant in the room”: the argument that immigrants from certain countries of origin have inherently lower potential—and that origin-based selection would be the best immigration policy. I take this argument seriously enough to engage with it rather than dismiss it as simply racist.

The strongest versions of it—for example, ’s work on national IQ—raise real empirical questions that deserve honest answers. I don’t think origin-based selection (as opposed to individual assessments) makes sense in 2026, for reasons that have less to do with political philosophy and more to do with data availability and the logic of liberal democracy. More on that soon.

Why immigration is not about humanitarianism (a left-leaning criticism): This was among the most passionate pushback I received. One reader argued: “You complain that the argument of ‘opposition to immigration is just racism’ is normative and not empirical, and in the same piece, you ‘debunk’ the ‘myth’ that immigration is about helping the vulnerable and treat it as an empirical claim when it obviously is not.”

I understand the sentiment. But I think this conflates what immigration should be about with what it is about as a matter of policy design and public support. The framing that immigration is fundamentally about humanitarianism is not just a normative preference—it is commonly deployed as a factual description of what immigration systems do and why they exist. And it is empirically wrong: the vast majority of cross-border movement is economic, and the vast majority of public opinion on immigration is shaped by perceived national interest, not humanitarian concern.

If you think there is anything I forgot to answer or mention, or you feel there are some factual errors or omissions, do let me know in the comments. As before, if you want me to write more about one of these or other related topics, I’m all ears.

Subscribe now