The Long 20th Century & Deep Economic History
17 tier-5 · 22 tier-4
This is DeLong's home turf - the spine of *Slouching Towards Utopia* and his 2026 Hicks Memorial Lectures. The throughline is a quantitative framework for very-long-run growth: his "index H" of deployed technological capability, the Malthusian trap that swallowed pre-1500 progress into population rather than living standards, and the discontinuous 1870/1875 break onto Kuznets's ~2%/year path where roughly one-fifth of the economy is leveled and rebuilt every generation. He insists civilization runs on a division of labor far exceeding any legitimate authority's reach (the "Globalization Problem"), that the agrarian age was a "society of domination," and that humanity's real superpower was never individual genius but the collective "anthology intelligence" built from cumulative culture. Stage theories, deep prehistory, and the contingency of the modern growth miracle recur throughout.
TIER 4
Apr 13, 2025
A book-draft meditation on the 1848 chapter of Harold James's Seven Crashes, reading the 1840s polycrisis (famine, disease, financial collapse, revolution) as the first push into modern globalization. DeLong's key point: 1848 failed yet did not produce reactionary clock-rollback—instead elites (Louis Napoleon, Bismarck) reinvented governance and its relationship to commercial prosperity, an institutional-adaptation thread he wishes he'd had for Slouching Towards Utopia. Substantive economic-history material tied to his long-20th-century framework, though short and paywall-cut.
economic history1848 revolutionsglobalizationSlouching Towards UtopiaHarold James
TIER 4
May 2, 2025
DeLong reflects on his Berkeley American economic history course, arguing there is no single Grand Narrative for the U.S. economy but thirteen distinct episodes/facets (frontier conquest, slavery, mass production, New Deal, Silicon Valley, neoliberal collapse, etc.) that should function as a 'filing cabinet' of historical analogies. Drawing on Dan Davies and Machiavelli, he frames history's value as a cheap library of mental models to riffle through, and asks how to design an exam that trains students to build that analogy-generating index. A useful pedagogical and methodological reflection on American economic exceptionalism and the practical use of history.
economic historyAmerican exceptionalismpedagogyhistorical analogyteaching
TIER 4
Jun 17, 2025
A draft lecture on what 'utopia' could even mean as a target for economic growth, surveying ancient archetypes—Sparte (order/discipline), Arkadia (pastoral sufficiency), Sybaris (sensual abundance)—as benchmarks for the good life, and asking whether prosperity bridges to or merely sharpens our appetite for new discontents. A thoughtful history-of-economic-thought piece, though paywall-truncated mid-survey and explicitly marked as an unrevised draft.
economic historyhistory of economic thoughtutopiaeconomic growthlecture notes
TIER 4
Jun 21, 2025
A complete lecture essay tracing how post-1500 merchant and industrial capitalism transformed slavery into racialized, market-driven plantation brutality, and how 'race' was invented as the ideological device to reconcile slavery with Enlightenment rights. DeLong argues the prime beneficiaries were middle-class consumers of cheap sugar, cotton, and tobacco, and closes with Claudia Goldin's striking point that the Civil War cost more than enough to have bought every enslaved person's freedom plus land. Self-contained and substantive.
economic historyslaveryracecapitalismCivil War
TIER 4
Jun 24, 2025
DeLong engages Marcella Alsan's instrumental-variable paper showing the tsetse fly, by killing cattle and blocking plow agriculture, drove centuries of African institutional divergence (less centralization, more slavery, lower population density). He finds the thesis persuasive but raises the methodological problem that 'history can only be truly causally explained once,' situating it against the Bantu expansion and the deep history of African cattle-herding. Substantive economic-history discussion, though cut off at the paywall.
economic historyAfricatsetse flyMarcella Alsaninstitutions
TIER 4
Jul 3, 2025
Drawing on a McKinsey Global Institute study showing fewer than 100 'standout' firms drove two-thirds of productivity growth across Germany, the UK, and the US, DeLong argues that technological dynamism is led by a handful of risk-taking firms rather than gentle diffusion, and that the US wins by reallocating resources toward the vanguard. He builds a policy case for subsidizing standout firms' expansion, grounded in the firm as imperfect satisficing bureaucracy (Kodak, Boeing) rather than rational profit-maximizer. A substantive industrial-policy argument with original framing of the firm's five roles.
productivityindustrial policySchumpeterian growthfirm theoryinnovation
TIER 5
Jul 13, 2025
A lecture making 'property' and 'exchange' strange: the belief that something stays 'mine' when I'm not guarding it, and the leaps from possession to reciprocal gift-exchange to one-shot market trade to a fluctuating-price market economy, are contingent socially-constructed institutions rather than natural propensities. DeLong threads Doug Jones on handaxes, Adam Smith's truck-and-barter, and the decentralized-knowledge case for markets (valid only for rival/excludible goods, and silent on distribution), capped by Aristotle, Locke, Rousseau, and Engels on property. A reference-quality piece in his economic-history-of-institutions vein.
economic historypropertyexchangeAdam Smithinstitutions
TIER 5
Sep 1, 2025
A full draft Harvard guest lecture sweeping from Gesher Benot Yaaqov tool-making through grain-importing classical Athens to post-1848 industrial divergence, framing all of human history around the 'Globalization Problem': civilization needs a division of labor that vastly exceeds any legitimate authority's reach, yet its gains are grossly uneven. A landmark, free-to-read synthesis of DeLong's economic-history project (the 'three Horsemen' of divergence, the post-1848 churn), with lasting reference value.
economic historyglobalizationdivision of laborSlouching Towards Utopiainequality
TIER 4
Sep 29, 2025
Using the survival show 'Naked & Afraid' as a hook, DeLong argues that an individual human brain, even an expert's, is insufficient to keep a person alive naked in the wilderness against the daily caloric math. The throughline from Acheulean handaxes to today is that human evolutionary advantage was never solo genius but pooled memory, collective 'anthology' thinking, and a division of labor embodied in tools no one could make alone. It matters as a vivid grounding of his recurring 'Anthology Super-Intelligence' thesis about the collective human mind.
human evolutioncollective intelligenceanthology super-intelligencedivision of laboreconomic history
TIER 5
Oct 13, 2025
A draft Econ 196 lecture giving DeLong's signature quantitative framework for modern growth: since 1870, every ~30 years about 1/5 of the economy quintuples in productivity (5.4%/yr) while 4/5 inches forward, doubling average productivity each generation through successive leading-sector 'modes' (steam, applied-science, mass production, value chains, now AI/info-bio). He extends it to argue the liberal-arts intelligentsia is now the displaced fifth caught in the Schumpeterian wave. Original framing, rich sourcing (Marx, Shalizi, Klaas, Slouching Towards Utopia), high reference value.
economic historySchumpeterian growthcreative destructionSlouching Towards UtopiaAI displacement
TIER 4
Oct 17, 2025
Replying to a nostalgist's claim that 1870-1914 uniquely combined freedom, prosperity, and cultural vitality, DeLong argues the Belle Époque felt golden because expectations surged far beyond a narrow, brittle reality--low-30s life expectancy, ~20% global literacy, thin franchise--which is precisely why 1914 was so shattering. A solid economic-history corrective, though most of the empirical detail sits behind the paywall.
economic historyBelle Epoque1870-1914World War Istandard of living
TIER 5
Nov 11, 2025
Rebutting the Landsburg/Slatepitch claim that 'nothing happened' before 1800, DeLong lays out his signature index H of deployed-and-diffused technology—average income per capita times the square root of population—which reveals steady pre-modern capability growth that Malthusian population dynamics swallowed into population rather than living standards. He tabulates millennia of growth rates and 'mode of production' thresholds (each a rough doubling, or √2-ing before 1500) culminating in the discontinuous 1870 jump to 2.1%/yr Modern Economic Growth. A landmark statement of his economic-history measurement framework with high reference value.
economic historyvery-long-run growthMalthusindex HSlouching Towards Utopia
TIER 5
Nov 20, 2025
Engaging Greg Clark's "Farewell to Alms," DeLong concedes the Malthusian treadmill held for biophysical necessities tied to reproductive fitness (flat living standards, four inches of skeletal stunting, ~0.08%/yr population growth) but argues it did NOT hold for technology, luxuries, culture, or means of domination—which Clark conflates. He derives his index H of human technological competence (0.08 in -3000, 1 in 1870, 27 today) along the necessities dimension. An original framework that is core to his Slouching Towards Utopia / long-growth project, with references and his own model.
economic historyMalthusianismGreg Clarklong-run growthindex H
TIER 4
Dec 3, 2025
Responding to Razib Khan on Yamnaya ancestry, DeLong argues the Indo-European expansion was a cultural/institutional revolution (mobility, steppe pathogens, scaled patriarchy, male-line dominance) rather than a biological transformation, since all living humans differ at only ~0.1% of sites. He extends this to deep prehistory, suggesting cumulative culture—not genes—has driven human capability since Homo erectus, reinforcing his 'Anthology Super-Intelligence' theme. A rich, original synthesis of population genetics and economic-historical method.
YamnayaIndo-Europeanpopulation geneticscultural evolutiondeep history
TIER 4
Dec 3, 2025
Engaging Razib Khan on Yamnaya ancestry, DeLong argues the Indo-European expansion was a cultural and lineage revolution—mobility, steppe pathogens, patriarchy, and male-line dominance—not a biological/genetic transformation, since modern human populations differ at only ~0.1% of sites. He marshals comparative human/primate genetic-distance figures to insist that meaningful evolutionary change requires going back hundreds of thousands of years, not five thousand. A clear, well-argued culture-over-genes case against genetic-determinist framings.
deep historyYamnayagenetics vs cultureRazib KhanIndo-European
TIER 5
Dec 16, 2025
On Austen's 250th birthday, DeLong reads her novels as a window onto a 'curious institutional interregnum' (c. 1795–1815) where the English landed gentry collected rents through neither warrior power nor productive contribution, yet faced no jacquerie—because fiscal-state capacity, legal architecture, parish relief, and a self-policing moral economy of reputation legitimated an unjust distribution long enough for the factory age to arrive. He layers five lessons (economic history, moral psychology as preference formation under constraint, the marriage market as portfolio optimization, the absence of revolution, and free indirect discourse as moral education), making it a landmark fusion of economic history and literary criticism.
economic historyJane Austenliteratureinequalitymoral psychology
TIER 5
Dec 16, 2025
A second (truncated-preview) send of the Austen 250th-birthday essay, identical in substance to issue 0231. It reads Austen's gentry as occupying an institutional interregnum where fiscal-state capacity and a self-policing reputational moral economy sustained rentier inequality without revolution, and treats her free-indirect-discourse innovation as itself a form of moral education—fusing economic history and literary criticism.
economic historyJane Austenliteratureinequalitymoral psychology
TIER 5
Jan 3, 2026
DeLong argues that 'the West' is a fake, discontinuous category (citing Ian Morris's wandering 'Western core' and the post-WWII Harvard Redbook's invented torch-relay genealogy) and proposes 'Dover Circle-Plus' instead: the societies descended from or emulating a 300-mile circle around Dover after 1500. He grounds the Dover Circle's post-1500 takeoff in five structural elements (Henrich on cousin-marriage/diffuse sociability, Berman on law binding the powerful, durable proto-nation-states, self-governing merchant cities, Crone on weak society-of-domination), yielding a reusable framework for periodizing the origins of modern growth.
economic-historydover-circlewestern-civilizationian-morrisgreat-divergence
TIER 5
Jan 8, 2026
A full essay arguing that modern science emerged and persisted in early modern Europe not from unique genius but from a self-reinforcing bundle: fragmented competing elites that raised the payoff to being right, an artisan-instrument culture that forced an interventionist epistemology, religion's ambivalent-but-often-positive charter and institutional homes, print networks that made debate public and portable, and institutions (academies, journals, nullius in verba) that lowered the cost of stable belief. The thesis is that persistence, not discovery, is what distinguished Europe from prior efflorescences (Hellenistic, Islamic, Chinese). A rich, original, heavily-referenced synthesis with lasting reference value.
history of scienceeconomic historyScientific Revolutioninstitutionselite fragmentation
TIER 4
Jan 8, 2026
DeLong asks why mathematico-experimental science emerged and persisted in early modern Europe rather than sputtering like earlier efflorescences, and answers with a bundle: fragmented elites and status competition that raised the payoff to being right, a craft world of instruments forcing interventionist epistemology, a permissive religious climate, print networks, and institutions (the Royal Society's nullius in verba) that lowered the cost of stable belief. The thesis is that this specific bundle aligned incentives toward empirical truth about nature rather than elite power. A strong economic-history-of-knowledge essay, though truncated at the paywall.
history-of-scienceeconomic-historyscientific-revolutioninstitutionselite-fragmentation
TIER 4
Feb 4, 2026
A hoisted 2023 piece presenting DeLong's formal four-equation Malthusian model and a Python simulation rebutting Rafael Guthmann's claim that pre-modern 'supercycles' of rising/falling incomes disprove Malthusianism. The key analytic move: a purely Malthusian economy with random shocks already generates centuries-long upswings and collapses, so observed efflorescences (Bronze Age, Classical Greece, Rome) are consistent with, not counterevidence to, Malthusian dynamics. A substantive model exposition with lasting pedagogical value.
malthusian-economyeconomic-historygrowth-modelsimulationefflorescence
TIER 4
Feb 4, 2026
DeLong's compact synthesis of why there was essentially no growth in median living standards before 1500: slow (~5%/century) technological progress was offset by population growth driven by patriarchy's demand for surviving sons, leaving humanity Malthusian-trapped while elites of 'thugs-with-spears' captured the surplus. The post doubles as a teaching artifact for his Econ 196 course, with a ranked reading list and discussion questions, making it a strong reference statement of his core Slouching-Towards-Utopia thesis.
economic-historymalthusian-trapagrarian-agepreindustrial-growtheconomics-teaching
TIER 4
Feb 11, 2026
DeLong's graduate-lecture notes on why pre-1600 techno-economic progress was so slow and why a tenfold technology gain from -8000 to 1600 went almost entirely into multiplying population rather than raising non-elite living standards (the Malthusian trap). He works through four readings, Morris on quantifying 'social development,' Kremer's population-drives-ideas endogenous-Malthusian model, the Finley-Temin debate on whether market models apply to Rome, and Henderson et al. on geography and path dependence. It matters as a substantive teaching synthesis of how to model very-long-run growth.
economic-historymalthusian-trapagrarian-agelong-run-growthecon-210a
TIER 4
Mar 24, 2026
DeLong's framing essay places vibe-coding inside his model of Schumpeterian creative destruction since 1875, where each generation one-fifth of the economy is leveled and rebuilt to do 5x as much, with knowledge workers now in the bullseye as many white-collar tasks turn out to have surprisingly low Kolmogorov complexity. Paul Ford's crossposted NYT piece supplies vivid first-person testimony of doing six-figure software work for a $200/month subscription. It matters for embedding a striking practitioner account in a coherent economic-history framework of labor transformation.
ai-disruptionvibe-codingcreative-destructionknowledge-workpaul-ford
TIER 4
Mar 27, 2026
Using Watt's marketing coinage of 'horsepower' and Jevons's Paradox, DeLong asks whether AI makes a given kind of worker 'coal' (cheaper-to-use complement whose demand rises) or 'horse' (substituted away), arguing software coding has so far been Jevons-paradoxical. He generalizes to a 5,000-year story of becoming efficient at producing Not-Raw-Food, concluding the binding constraint on the AI transition is distributional and institutional, not technological. It matters as a clean conceptual frame for thinking about AI's labor effects.
jevons-paradoxai-and-laboreconomic-historyautomationsteam-engine
TIER 4
Apr 9, 2026
DeLong revisits Acemoglu-Johnson-Robinson's 'Colonial Origins' paper as 'a true rabbit and a true duck'—elegant and influential, yet hiding heroic assumptions inside its instrumental-variable design (either prosperity structurally degrades governance, or 17th-century settler mortality measures modern institutions better than direct observation). He uses it to argue, Heckman-style, that only those with a fully specified structural model have warrant to make 'causal' claims. A substantive methodological critique in economic history and the philosophy of empirical inference.
economic historyinstitutionsAcemoglu-Johnson-Robinsoninstrumental variablesstructural models
TIER 4
Apr 19, 2026
Framed around a new Nature Communications paper on a 220,000-year-old hornfels quarry-workshop in KwaZulu-Natal, DeLong argues that spatial division of labor, long-term planning, and cumulative cultural transmission—supply-chain economic behavior—predate Homo sapiens sapiens by far, making us 'different because of our economic function.' This is a draft opening for his book 'Enlarging the Bounds of Human Empire' and his Oxford Hicks Lecture, so it carries the thesis of his deep-history economic project. The bulk of the substantive argument is paywalled, but the framing and stake-in-the-ground are substantive economic history.
economic historydeep prehistorydivision of laborarchaeologyHicks Lecture
TIER 5
Apr 30, 2026
The 'modern economic growth' installment of the Hicks lecture, marking 1875 as the genuine break onto Kuznets's path of sustained compounding growth and Gordon's 'one big wave' of transformative technologies, with US and German trend labor-productivity growth rising toward ~2%/year and doubling times falling from a century to ~35 years. It documents the US overtaking Britain by 1870 and the structural shift from rare to normal growth in output per head, supplying the quantitative pivot of the whole stage-theory narrative.
economic-historymodern-economic-growthKuznetsproductivity-growthsecond-industrial-revolution
TIER 5
Apr 30, 2026
The prehistory installment of the Hicks lecture, opening 700,000 years ago at Gesher Benot Ya'aqov where late Homo erectus already ran a cognitive and physical division of labor — a recognizable economy without language. DeLong develops the 'anthology intelligence' thesis (gossip and information-sharing welding bands into a collective mind), works through the gatherer-hunter-to-farmer Malthusian transition and the living-standard anchors ($1,600 then $1,200/capita), and derives the strikingly slow ~0.007%/year pace of technological self-improvement from -70,000 to -3,000.
economic-historyprehistoryanthology-intelligenceagricultureMalthusian-equilibrium
TIER 4
May 1, 2026
The methodological core of the Hicks lecture: because humans are storytelling animals, everyone inevitably arrives at a stage theory, and DeLong catalogs the teleological two-stage versions (Acemoglu, McCloskey, Polanyi, Rostow) and faults them for treating 'us' as the inevitable end-state. He draws Hicks's bottom-line lessons — we were very lucky, the market system was a halting tendency not an inevitability, fixed-capital industrialization needed both science and financial deepening — and frames the project of building a better, less teleological inductive stage theory. Shorter and more setup-heavy than the substantive installments.
economic-historystage-theoryHicksteleologymethodology
TIER 5
May 2, 2026
The agrarian-age installment of the Hicks lecture: with writing and bronze at -3000, humanity becomes a true 'anthology superintelligence,' yet most societal energy goes into organizing gangs of domination (thugs, accountants, propagandists who take a third of the crop) rather than productivity, keeping recursive self-improvement to roughly 0.026%/year through 1310. DeLong wrestles seriously with measuring living standards under domination (luxuries, variety, cultural goods, the Dixit framing) and stresses that ancient and medieval civilization was sophisticated and accomplished, just inept at advancing the technological stock.
economic-historyagrarian-agesocieties-of-dominationMalthusian-trapwriting-and-bronze
TIER 5
May 3, 2026
A core Hicks-lecture installment that frames the lecture's purpose (why Hicks turned from neoclassical synthesis to economic history and stage theories), critiques teleological two-stage theories, and then quantifies the acceleration of human 'recursive self-improvement' from prehistory through the agrarian age into the commercial-imperial and first-industrial eras. It develops DeLong's signature arguments: the anthology-intelligence framing, Crone's view of Europe as a failed society-of-domination, Allen's cheap-coal/high-wage story, Jevons's Paradox and Nightmare, and a vivid SteamPunk-Oxford counterfactual of arrested 1875 technology.
economic-historyindustrial-revolutioncoal-and-steamgreat-divergencerecursive-self-improvement
TIER 5
May 3, 2026
A rich Hicks-lecture installment tracing the succession of post-1875 economic regimes — steam-power to applied-science to Fordist mass-production/New-Deal order to the neoliberal globalized value-chain economy — each with its own growth pattern and pattern of disruption. DeLong argues the mid-century mass-production/social-democratic alignment was a lucky, contingent constellation (FDR, the Depression, the war), not a natural equilibrium, and that as creative destruction shifted into IT, logistics, and finance, the old compromises cracked and the neoliberal order did worse at turning the ~2% growth engine into broadly legitimate life-trajectories.
economic-historymass-productionNew-Deal-orderneoliberalismglobal-value-chains
TIER 5
May 3, 2026
The closing installment of DeLong's 2026 Hicks Memorial Lecture refuses to deliver the promised inductive conclusion, instead passing the task to younger scholars: build a stage theory that integrates growth with distribution, ties technology to politics, and is explicit about contingency. It surveys the half-dozen existing stage theories (three-stage, Kondratiev-Schumpeter, Marx, Rostow, Polanyi, 'orders') and argues none is adequate, making this a programmatic statement of what economic history should aim at.
economic-historystage-theoryHicks-lecturegrowth-and-distributioncontingency
TIER 4
May 5, 2026
DeLong runs the Drake Equation with optimistic early terms to derive a chilling implication: if visible civilizations are rare, their average visible lifetime L may be only ~1000 years, implying powerful forces cut civilizational life short. He sets this against the extrapolation of 2%/year growth toward Bel-Air-for-everyone incomes, arguing the Great Filter likely lies in bootstrapping a robust civilization, not in life or intelligence, and that this makes long-run governance and existential-risk reduction central economic variables. A speculative but substantive draft chapter linking growth theory, deep history, and existential risk.
Fermi paradoxDrake equationGreat Filtereconomic growthexistential risk
TIER 4
May 15, 2026
A substantive crosspost-interview on the Great Divergence: Broadberry and Gupta's grain-wage data show India and Britain at similar living standards around 1600, with Indian real wages falling below half of Britain's by 1750 and agricultural decline beginning before colonialism (extraction, not investment, under British rule). The conversation marshals evidence on state fiscal capacity (a Chinese taxpayer worked 2 days/year vs. 17-20 in Britain), the European marriage pattern and urban mortality as Malthusian brakes, caste limiting technological diffusion, and dates the China-Europe divergence (Yangzi Delta) to ~1700. DeLong adds a useful five-channel framework for sources of productivity and prosperity. Strong reference value for economic history.
Great Divergenceeconomic historyIndiaMalthusian trapstate capacity
TIER 4
May 22, 2026
DeLong reposts Mark Koyama's review of the late Nick Crafts's macro economic history of Britain, which traces the successive regimes of British growth: early industrial lead, relative decline behind the US and continental Europe through institutional path dependency and craft-union rigidities, and partial Thatcher-era recovery into the ICT wave. The review distills Crafts's quantified, growth-theory-driven account of why Britain forged ahead, fell behind, then fought back. A solid guided tour of a landmark economic-history achievement, valuable for British long-twentieth-century context.
British economic historyNick Craftsindustrial revolutionvarieties of capitalismTFP growth
TIER 5
Jun 14, 2026
DeLong's flagged 'most important thing': 75,000 years ago humans were an evolutionarily marginal great ape (perhaps 10,000-400,000 individuals, vastly outnumbered by lions, elephants, and antelope), whose big brains, imitation-driven culture, and ultrasociality looked like costly gambles akin to the Irish elk's antlers. Drawing on Heath, Henrich, and Gould, it argues our dominance rested on a narrow, contingent, lucky path rather than any obvious Darwinian edge. A rich, original synthesis on cultural evolution with strong reference value.
human-evolutioncultural-evolutionrunaway-brainsimitative-learningdeep-history
TIER 5
Jun 15, 2026
DeLong's 'key insight': the sense of unprecedented rupture felt by intellectuals like Adam Tooze is not itself new but the recurring experience of whoever lands in the bullseye of Schumpeterian creative destruction, which since 1875 arrives every generation. He offers a reusable quantitative frame: each wave lifts four-fifths of the economy ~50% productivity with intact lives while obliterating and rebuilding one-fifth (e.g. the Silesian weavers), and the twenty-first-century twist is that knowledge workers are now the targeted cohort. An original, durable framework with lasting reference value.
creative-destructioneconomic-historyknowledge-workersai-and-jobsschumpeterian-waves
What the Machines Actually Are - LLMs, Stochastic Parrots & the Anthology Super-Intelligence
10 tier-5 · 26 tier-4
DeLong's single most insistent theme is a deflation of the "BRAINS!!!" frame around large language models. Across these pieces he argues that GPT-class models are kernel-smoother / interpolation "function machines" - "Clever Hans at scale," next-token predictors with no world model - and that anthropomorphizing them is projection, not observation. The constructive counter-frame, built with Cosma Shalizi and Henry Farrell, is that LLMs are a *cultural technology*: a new, lossy, natural-language front-end to what he calls the "Anthology Super-Intelligence" (ASI) - the five-millennia collective corpus of human knowledge stored in writing, catalogs, markets, and institutions. Markets and bureaucracies are the original "slow-AI shoggoths"; LLMs are the latest and fastest. The payoff is a calibrated skepticism: useful, not conscious; an interface, not a mind.
TIER 5
Apr 4, 2025
A long, ambitious review of Dan Davies's 'The Unaccountability Machine' that reconstructs Stafford Beer's management cybernetics—accountability sinks, variety attenuation/amplification, the Viable System Model—as an alternative to economics' maximizer framing for governing large opaque systems. DeLong weaves in Adam Smith's invisible hand as a 'slow-AI,' the Lawson Doctrine's failure, Farrell's Vico-vs-Kafka fork, and the claim that any single-objective maximizing system 'goes bonkers' without a higher-level red handle. A landmark synthesis tying information-flow theory, the firm, neoliberalism's failures, and AI anxiety together.
cyberneticsDan Daviesneoliberalismcomplex systemshistory of economic thought
TIER 4
Apr 7, 2025
A crosspost of Angus Bylsma's review of Kindleberger's 'The World in Depression,' centering the hegemonic-stability thesis ('the British couldn't and the United States wouldn't' stabilize the world economy) and praising the book as thick international synthesis akin to Tooze's 'Crashed.' DeLong's stake is direct: he and Eichengreen wrote the 2012 foreword, and the review updates their hegemony-falters diagnosis for a 2025 where the US actively rebels against benevolent hegemony. Substantive economic-history review with strong present-day relevance.
economic historyGreat DepressionKindlebergerhegemonic stabilitycrosspost
TIER 4
Apr 16, 2025
DeLong's widely-quoted argument (the source of his 'you're projecting, not it's thinking' line) that LLMs are flexible kernel-smoother interpolation functions, not brains, and that the 'BRAINS!!!' frame obscures rather than illuminates why they behave as they do. He rebuts Scott Cunningham's 'inside the brain of Claude' reading and illustrates with ChatGPT fabricating an archive.org URL. A foundational statement of his AI-skeptic position, though the demonstration is paywall-cut.
AILLMsinterpolationanthropomorphismepistemics
TIER 5
Apr 17, 2025
A full crosspost of Cosma Shalizi's landmark essay reframing LLMs, via Alison Gopnik's 'cultural technology' thesis, as a new form of information retrieval and social technology rather than minds—parametric probability models that interpolate from the digitized corpus, mediating relationships between users and prior authors. Drawing on Barzun's House of Intellect and the Newell-Simon 'complex information processing' origins of AI, it argues these tools dispense stored intellect, not creative intelligence, with distinctive social failure-modes we don't yet understand. A reference-grade framework that DeLong calls a better statement of his own view.
AILLMscultural technologyhistory of AICosma Shalizi
TIER 4
Apr 27, 2025
Around Cosma Shalizi's new course framing LLMs as high-order Markov models and attention as kernel smoothing, DeLong stages his own ambivalence about what LLMs are, documenting concrete failures (hallucinated book series, wrong archive.org URLs) that contradict booster claims of emergent world-knowledge and theory of mind. He rejects Ethan Mollick's 'jagged AGI' framing in favor of a 'Clever Hans interpolating in 3000-dimensional space' model that is superb at boilerplate and zeitgeist-summarization but hopeless where answers have a single right value. A substantive, honest engagement with what generative AI actually does.
LLMsCosma Shalizikernel smoothingAI skepticismhallucination
TIER 4
Jun 9, 2025
DeLong's own essay arguing the most important thing about current MAMLMs/LLMs is natural language as the human-machine interface—not intelligence, which he places firmly on the Searle 'Chinese Room' side rather than Aaronson's. The thesis: usefulness, not consciousness, is the right standard; conversational AI democratizes access to complexity and enables Socratic externalization of cognition, with caveats about false confidence, dependence, and ownership. A complete, characteristic DeLong meditation tying his SubTuringBradBot project to the anthology-intelligence theme.
AInatural language interfaceChinese RoomMAMLMscognition
TIER 4
Jun 9, 2025
A full crosspost of Leif Weatherby's essay (with a substantial DeLong intro that pushes back) arguing that LLMs are best understood not as nascent intelligence but as 'digital bureaucracy'—a semantic spreadsheet that translates between data and language and supercharges administrative control, exemplified by DOGE/Palantir. Weatherby names 'the performance fallacy' (mistaking benchmark optimization for intelligence) as the central error in AI discourse. DeLong partially dissents, defending abstraction layers while agreeing vigilance must replace abdication; a meaty, reference-worthy AI-critique piece.
AILLMsbureaucracyDOGETuring test
TIER 5
Jun 20, 2025
A crosspost of Henry Farrell's full essay arguing that today's AI debate is a recapitulation of a centuries-old modernity dilemma: Vico's hope that what humans make they can understand versus Kafka's dread of incomprehensible machineries. Farrell traces the two dominant AI metaphor-systems—rationalist 'summoning' (Yudkowsky, angelology) and accelerationist worship of inhuman forces (Nick Land, e/acc)—and argues both are misleading, since AI is neither volitional nor Lovecraftian chaos but another opaque complex system like markets. A landmark, richly referenced intellectual-history framework that DeLong chose to republish in full.
AIintellectual historyHenry Farrellsingularityaccelerationism
TIER 4
Jun 24, 2025
DeLong argues GPT-class LLMs are 'function machines' that interpolate from a lossy compression of training data—closer in economic impact to the spreadsheet than the microprocessor, and far from world-destroying superintelligence—pushing back on Ben Thompson's framing where the microprocessor case is the low-impact one. He clearly explains the kernel-smoothing/autocomplete-on-steroids mechanism (via Shalizi and Chiang) and catalogs where the tools genuinely excel and where they fall down. A clear, useful explainer with a memorable calibrating frame.
AI economicsLLMsgeneral-purpose technologyproductivitykernel smoothing
TIER 4
Jun 24, 2025
DeLong argues that LLM-based AI is closer to a spreadsheet-class innovation than a microprocessor-class one, and far from a superintelligent 'destroyer of worlds,' framing GPT models as natural-language 'function machines' that are useful but bounded. He pushes back on Ben Thompson and the superintelligence camp, and notes the broader MAMLM category matters to Meta for ad-targeting beyond chatbots. Truncated by paywall but the core thesis lands.
AILLMstechnologyproductivityBen Thompson
TIER 4
Jun 29, 2025
DeLong argues that LLMs are iterated fixed-function Markov processes with no internal state, so attributing plans/desires to them is a category error, and that the real road to understanding machine minds runs through complexity and emergence rather than scaling laws and the Bitter Lesson. He marshals Aaronson's Chinese Room rebuttal, exponential-curve 'logistication,' and a brain-vs-LLM parameter comparison (10^18 synapse-parameters vs Claude's ~10^11) to argue Turing-class silicon is as far off as today's models are from the 1960 Perceptron. A meaty, reference-rich AI-skeptic argument.
AIemergenceLLMsphilosophy of mindscaling laws
TIER 5
Jul 27, 2025
A cross-post of Henry Farrell's extended essay (with DeLong's framing) arguing the LLM-as-shoggoth meme reveals the wrong lesson: LLMs are not rebellious slaves but a new instance of the 'slow-AI' impersonal information-processing systems—markets, bureaucracies, democracies—that have been Hayek/Scott/Dewey shoggoths since the Industrial Revolution's true Singularity two centuries ago. The payoff is reframing AI alignment fears as the older problem of vast, lossy, alien collective systems condensing human knowledge, and asking how LLMs will compete with or hybridize their elder kin. An original, durable conceptual essay even as a guest piece.
AI/LLMsmarkets and bureaucracyHenry FarrellHayekmodernity
TIER 4
Aug 25, 2025
DeLong urges separating three conflated things bundled as 'AI' — natural-language interfaces to databases, the financial-tech earthquake, and high-dimensional classification — and focuses on the first via Henry Farrell's taxonomy of LLMs as 'cultural technologies' (Gopnikism, interactionism, structuralism, roleplay). He develops each lens (information-overload superpowers, our anthropomorphizing 'theory of mind,' Ong's orality-to-literacy transitions, and spinning up 'SubTuring Simulacra' of other minds) to ask how LLMs reshape how we think. A substantive framework piece on AI's cognitive and cultural implications.
llmsaihenry-farrellcultural-technologycognition
TIER 4
Sep 12, 2025
DeLong argues that markets, bureaucracies, democracies, ideologies, corporations, and professions are 'slow-AI' systems—distributed algorithmic social technologies that process information and coordinate millions, both indispensable for scale and terrifying to those caught in their gears (Kafka's castle, the Holocaust's machinery, Vietnam body counts). LLMs are framed as the latest, faster 'shoggoth' in this old lineage (per Farrell and Shalizi), with the real task being to steer them toward augmenting public reason rather than manipulation. A substantive synthesis tying AI into his institutions-as-collective-cognition theme.
slow AIinstitutionsmarketsbureaucracycollective intelligence
TIER 5
Oct 2, 2025
Building on Farrell and Shalizi's 'AI is a familiar-looking monster,' DeLong argues markets, bureaucracies, and now LLMs are all 'shoggoths'--impersonal information-processing systems we created but cannot fully control--and lays out his full 'modes of production/distribution/communication/domination' periodization from Hunter-Gatherer to Info-Biotech. Adds the crucial corrective that these monsters also massively empower us, reframing the project as Slouching Towards Utopia; a high-value framework piece, free to read.
modes of productionshoggothsFarrell and Shaliziinfo-biotechSlouching Towards Utopia
TIER 5
Oct 5, 2025
Responding to Noah Smith's "Third Magic," DeLong proposes a five-or-six-magics framework for the meta-innovations behind human power—tool-use, language, writing, gift-exchange/markets, and science—and asks whether high-dimensional/flexible-function prediction (today's LLMs) qualifies as a genuine sixth. He decomposes the "AI" phenomenon into seven distinct things people conflate (prediction, NL interfaces, chatbots, and the Downer/Boomer/Doomer hype machines plus the financial bubble), arguing chatbots are "pass-the-story" engines echoing real human thought, neither thoughtless parrots nor sparks of AGI. An original synthesizing framework with lasting reference value across his AI and economic-history themes.
AIhistory of technologyNoah Smithanthology intelligenceeconomic growth
TIER 5
Oct 5, 2025
DeLong reworks Noah Smith's 'Third Magic' (history, science, AI as ways of learning) into his own five-magic framework for human progress: tool-use, language, writing, gift-exchange coordination, and science, then asks whether high-dimensional AI prediction-without-laws qualifies as a genuine sixth magic or is merely a 'wishful mnemonic.' A landmark synthesis tying his anthology-intelligence and Slouching-Towards-Utopia themes to the AI debate, with lasting framework value.
AINoah Smithmagics frameworkanthology intelligenceeconomic growth
TIER 4
Nov 26, 2025
DeLong argues that scaling LLMs is scaling mimicry, not understanding—without built-in world models, frontier MAMLMs are "supersummarizers" of the human record that excel at clear-answer or massive-counting tasks but stay brittle in embodied, long-horizon contexts. Engaging Helen Toner's "jaggedness," he reframes the unevenness via his TIS (Typical Internet S***poster) emulation model, pointing toward durable centaur (human+AI) workflows. A clear statement of his recurring AI thesis with useful concrete examples (Anthropic's Project Vend).
AILLM limitsworld modelsHelen Tonercentaur workflows
TIER 5
Dec 5, 2025
DeLong's signature AI-bubble thesis: proprietary LLMs will be lousy businesses (high opex, thin moats, fast commoditization), so AI won't mint new platform monopolies but will 'manure' the next generation's digital commons via stranded grids, GPU farms, and open-source code—an equity-financed 2000-style crash, not a 2008 one. The key mechanism is that Google/Facebook/Amazon spend to defend existing monopolies and will give assistants away free (the Netscape/IE analogy), capped with concrete Kindleberger-Minsky signals to watch. A landmark, framework-grade argument with lasting reference value.
AI bubbleplatform monopolyKindleberger-Minskycapex spilloversmacro outlook
TIER 4
Dec 13, 2025
Conference notes reframing 'ASI' as the Anthology Super-Intelligence—the five-millennia corpus of human ideas—rather than artificial superintelligence, and arguing universities earn public trust by teaching the tools that let students plug into that collective corpus. LLMs are positioned as useful dumb natural-language front-ends to curated databases, not oracles, anchored by Popperian falsification and epistemic humility. A compact statement of DeLong's recurring 'Anthology Super-Intelligence' framework with lasting reference value, though delivered as bullet-point notes.
social sciencesAnthology Super-IntelligenceuniversitiesepistemologyLLMs
TIER 4
Feb 17, 2026
Using a chatbot's fabricated fantasy-novel bibliography and an impossible WWII-Europe map as exhibits, DeLong argues that LLMs hallucinate not as rare edge cases but as the core logic of systems that hold correlations instead of facts and have no world model. He walks through the next-token, piggyback-on-a-human-train-of-thought mechanism and concludes that without a world model, correlation matrices will always confabulate—often unpredictably—unless you already know the answer. It matters as a sharp, mechanistic skeptic's account of why 'compression' is a treacherous metaphor for what these systems do.
AILLM hallucinationworld modelsmachine learningepistemics
TIER 4
Feb 18, 2026
Responding to Yglesias's paralysis over whether AI will plateau or explode into superintelligence, DeLong deploys his 75,000-year framework of accelerating technological growth driven by the 'Anthology Super-Intelligence'—humanity's collective, time-binding mind stored in its information-technology capital stock. He argues AI is a powerful natural-language front-end to that real ASI, a tool not a digital-god master, so paralysis in the face of the fork is irrational. It matters as a compact statement of DeLong's signature long-run growth model and his anti-doom 'Gopnikist' stance on AI.
AIlong-run growthAnthology Super-IntelligenceSchumpeterian creative destructioneconomic history
TIER 4
Feb 23, 2026
DeLong lays out his working philosophy for using large language models: stop treating them as minds and treat them as stochastic calculator-translators wired to large libraries, where 'context engineering' (write/select/compress/isolate) is the real work and 'prompt whispering' is theater. He argues the binding constraint shifts from text production to reading/filtering attention, and the productive use is as a better front-end to the real anthology super-intelligence of accumulated human knowledge. A useful, opinionated practitioner explainer on AI workflow.
LLMscontext engineeringattention economyanthology intelligenceAI workflow
TIER 4
Feb 24, 2026
Through nine failed attempts to get ChatGPT to name the seven champions of Ser Duncan at Ashford Meadow (a finite, well-defined, open-book question), DeLong documents a 'hallucination cascade' where the model never converges on the correct, well-documented list despite ever-greater confidence. He uses this as a sharp empirical argument that LLMs are next-token predictors, not reasoning minds, and that mistaking their fluency for comprehension is 'CleverHansMaxxing.' A vivid, well-constructed demonstration of a real model-capability boundary and how to think about it.
LLM hallucinationAI capability limitsClever Hansnext-token predictionstochastic parrot
TIER 5
Apr 6, 2026
DeLong crossposts Cosma Shalizi's slides arguing GenAI is mechanized information retrieval and synthesis that generates formulaically, and—because much of human culture is itself formulaic tradition—LLMs are best understood as 'prosthetic tradition,' an all-access front-end to the 'House of Intellect' (Barzun) rather than geniuses in a data center; the Marxian gloss reframes them as mechanized intellect (dead labor) versus living intelligence. DeLong appends eight extensions tying it to his 'Anthology Super-Intelligence' framework. A landmark conceptual framing of what LLMs actually are, with lasting reference value.
LLMshistory of ideasShalizitraditionAnthology Super-Intelligence
TIER 4
Apr 16, 2026
Starting from Treasury Secretary Bessent's 'Straits of Vermouth' slip, DeLong develops an extended analogy between Freudian parapraxes and LLM internals—feature vectors, over-weighted priors, guardrails and jailbreaks—to read political speech (covfefe, 'vermin,' dog whistles) as noisy samples from an over-amped 'system prompt.' The genuinely interesting move is using interpretability work (Golden Gate Claude) as a directionally-right model of the human mind rather than dismissing the LLM-mind comparison. A substantive, original explainer that yokes AI interpretability to political psychology.
LLMsinterpretabilitypolitical psychologyparapraxisstochastic parrots
TIER 4
Apr 27, 2026
An open letter to Noah Smith arguing, by continuity from a revealing GPT-3.5 failure, that current LLMs are extraordinarily sophisticated mirrors and tools but not conscious in any morally salient sense — and that the EA wing's concern for model 'welfare' is a category error that misallocates moral attention from undeniably sentient beings. DeLong concedes that scaling plus structure (a Jupiter-sized Chinese Room, or cortex emulation guided by neural correlates of consciousness) could in principle cross the line, but insists more-of-the-same next-token prediction on an exhausted data slurry will not, and proposes a 2036 ramen-dinner wager.
AIconsciousnessLLMseffective-altruismstochastic-parrots
TIER 4
May 5, 2026
DeLong argues that economic history begins not with markets but with the biocultural evolution of humanity as an 'anthology intelligence' — and uses the recent finding that intelligence evolved independently in birds, mammals, and octopuses to argue that individual brainpower is not what makes humans special. The real engines are hypersociality, cumulative culture (the ratchet effect), language, and the collective brain, which together form the deep substrate on which all later growth (pin factory to silicon chip) is built. He notes it also weakens the 'tool-using intelligence' candidate for the Fermi Paradox's Great Filter.
economic-historycultural-evolutionintelligenceanthology-intelligenceFermi-paradox
TIER 4
May 11, 2026
DeLong argues that the fact LLM vendors almost never run at temperature=0 is a 'tell': a genuine reasoner would not need injected randomness to stay engaging, so temperature is a stage effect masking the model's nature as a low-entropy argmax machine. Drawing on Shannon information and the I Ching 'fire in the lake' analogy, he contends the apparent creativity is the user's own associative decoding of a random seed, not cognition in the transformer weights. A sharp, original conceptual explainer on what temperature really does and why it matters for AI hype.
LLM temperaturestochastic parrotsinformation theoryAI hypeMAMLMs
TIER 5
May 13, 2026
DeLong's central original framework: humans are not smart animals who learned to cooperate but a cooperative 'anthology organism' whose intelligence is an emergent property of pooled memory, the division of labor, and the extended mind, illustrated from Naked-and-Afraid starvation to the Gesher Benot Ya'aqov hominins to Tasmania's tech regression. He argues LLMs are not minds but a new lossy interface to this real Anthology Super-Intelligence, then offers four 'punchlines' (distillation, parasite, new organ, mirror) for what the LLM moment means. A landmark, reference-grade synthesis tying deep history, the economics of knowledge, and AI together.
anthology intelligencecollective braincultural evolutionextended mindLLMs
TIER 4
May 13, 2026
A Hicks-Lecture outtake arguing humanity's superpower is not individual intelligence but collective coordination: 'We are a cooperative organism that acquired intelligence as an emergent property of our cooperation,' not smart animals who learned to cooperate. The visible portion makes the case with the 'Naked & Afraid' illustration—even an expert survivalist with a knife or fishing line starves alone in the wilderness, showing the lone human brain is wildly insufficient against the daily caloric math of nature. A distinctive framing of the social-anthropology basis of economic history, though much sits behind the paywall.
economic historyhuman cooperationcollective intelligenceHicks lectureMalthusian
TIER 4
Jun 1, 2026
Working through a knotty sentence of Cicero's In Catilinam with a local LLM that drills him daily in Latin, DeLong reflects on why the model is so effective as a grammarian despite being 'just' linear algebra: centuries of Latin pedagogy are sedimented in its training data, the micro-prompt task structure fits well-worn grooves, and success is plausible continuation, not ground truth. He extends this to the claim that human disciplinary 'understanding' is itself substantially stochastic parrotry (citing being 'spoken by' his teacher Jeffrey Williamson). A rich essay on AI, pedagogy, and the nature of learning.
stochastic parrotsAI pedagogyLatin / CiceroLLMsnature of understanding
TIER 4
Jun 4, 2026
DeLong pushes the 'stochastic parrot' critique sideways: reading itself is mimicry, since active readers spin up a 'subturing instantiation' of an absent author and interrogate it, so prompting an LLM lies on the same spectrum as reading Cicero or arguing with Socrates. He distinguishes uses where you want comprehension from purely performative 'prayer-wheel' uses, and claims expert LLM use could in principle beat active reading. A genuine information-sociology essay on how absent minds are made to speak.
llm-mimicryactive-readinginformation-sociologyprompt-engineeringphilosophy-of-ai
TIER 4
Jun 4, 2026
Responding to a Bluesky comment ('a parrot that teaches you Latin has crossed a line the metaphor meant to deny'), DeLong argues that reading itself is an act of mimicry: from black squiggles we spin up a 'subturing instantiation' of an absent author and interrogate it, per Machiavelli's letter to Vettori. He builds a seven-rung ladder from arguing with Socrates to passively copy-pasting LLM output, asking what art distinguishes good use of LLMs. A thoughtful (paywall-cut) essay on the epistemology of reading, teaching, and AI.
stochastic parrotsepistemology of readingAI pedagogyLLMsinformation sociology
TIER 4
Jun 8, 2026
Drawing on a Demirer/FT funnel chart, DeLong argues AI-boosted coding produces ~300% more files but only ~30% more shipped releases (and more apps with no added downloads), so AI is flooding the zone with artifacts no one needs rather than creating proportional value. His framing: MAMLMs are fast, flexible classification/prediction tools and 'Clever Hans at scale,' not reliable brains—they open doors but are not the doorman. A chart post elevated by a sharp, reusable conceptual frame and a concrete reliability failure example.
ai-hype-cyclecoding-assistantsai-value-creationagentic-modelsmacro-outlook
TIER 4
Jun 9, 2026
DeLong traces the evolution of making LLMs useful through three phases—context engineering (ruthless curation under a 4K-token ceiling), prompt engineering, and now harness engineering—arguing that reliability lives in the symbolic scaffolding (tests, tools, clean context) around the model, not in bigger context windows. The throughline is 'context rot': models go dumb well below advertised context limits, so LLMs are best treated as components inside a truth-enforcing harness. A useful conceptual explainer of practical AI engineering (truncated by paywall).
harness-engineeringcontext-rotprompt-engineeringragllm-reliability
The AI Bubble & the Economics of the Boom
7 tier-5 · 32 tier-4
If LLMs are interfaces rather than gods, what does the trillion-dollar buildout actually buy? DeLong's recurring answer is that the AI boom is a "bubble-plus" - a roughly 12-dimensional vector, not one story - in which only end-user value and ad-targeting are likely sources of investor superprofit, while platform-monopoly defense, techno-millenarian hype, grifters, and uncapturable user surplus make up the rest. His signature mechanisms: inference never becomes a near-zero-marginal-cost node; incumbents (Google, Meta, Amazon) will give assistants away free to deny anyone platform rents (the Netscape/IE analogy); profits flow to the "picks-and-shovels" sellers (NVIDIA, TSMC, ASML); and the most likely endgame is a 2000-style equity crash that "composts" the next generation's digital commons rather than a 2008-style systemic break. Apple's low-capex, on-device bet recurs as the contrarian counter-case.
TIER 4
Apr 23, 2025
An expanded version of DeLong's interview remarks framing Trump's trade war as fake WWF theater—waving the metal folding chair for attention rather than executing strategy—whose real cost is the loss of trust that drives the world to de-risk from the US. He ties this to his long-20th-century mode-transition framework and argues a small number of GOP leaders could install a 'regent' to run policy. A wide-ranging, quotable bullet-point essay on trade, political economy, and American relative decline.
trade warTrumpBREXIT analogyde-riskingpolitical economy
TIER 5
May 29, 2025
A reissued narrative essay (originally 2009) using E.M. Forster's biography of his great-aunt Marianne Thornton to dramatize the Bank of England's 1825 rescue of the Pole, Thornton bank, the moment DeLong identifies as the birth of modern central banking. He weaves the human story into Mill/Marshall theory of how lender-of-last-resort action satisfies excess demand for safe liquid assets and ends downturns. A vivid, durable piece on the origins of the central-bank backstop with lasting teaching value.
Panic of 1825central bankingBank of Englandfinancial historylender of last resort
TIER 4
Jun 6, 2025
DeLong's recurring macro column argues Trump's chaos-monkey policy is less a fog than a sandstorm—when rules change at a tweet's notice, optionality dominates, investment freezes, and recession plus baked-in stagflation and BREXIT-style ~10% long-run decline become likely. He then puzzles over Mohamed El-Erian's optimism and offers four hypotheses (professional-optimist ritual, Schumpeterian creative destruction, limits-of-pessimism, cynical signaling to Bessent/Lutnick). A genuinely substantive uncertainty-economics essay with a real references list.
macroeconomicsuncertaintystagflationTRUMPXITEl-Erian
TIER 4
Jun 30, 2025
DeLong argues that AI's value to users is real but corporate profits will be elusive because competitors are building out capabilities with no intention of ever charging—the Netscape-vs-free-Internet-Explorer problem—producing a Red Queen's race where user surplus rises but stock-market valuations are unjustified. He invokes the dot-com crash and 'ruinous competition' of railroads, noting on-device players (Apple, Google, Samsung) controlling hardware chokepoints may be the exceptions. A solid economic-history-informed take on the AI-investment bubble.
AI economicsROIbubbleplatform competitionmarginal cost
TIER 4
Jul 11, 2025
A crosspost of Macquarie strategist Viktor Shvets applying Carlota Perez's framework—that every technological revolution rides a wave of speculative excess and follows a U-shaped productivity dip before takeoff—to argue the AI bubble is a feature, not a bug, of Schumpeterian creative destruction. He contends AI is far more pervasive than past revolutions, that abundant capital and compressed cycles cushion the bust, and that the real risk for humanity is underinvesting rather than overinvesting. Substantive, framework-driven take on AI capex and bubbles despite being a guest piece.
AI bubbleCarlota Perezcreative destructionproductivitytech investment
TIER 4
Jul 24, 2025
DeLong offers a framework decomposing the Magnificent Seven's value into four sources—genuine productivity gains, platform rent collection, 'tightening the screws' (enshittification), and meme-stock dynamics—and maps each firm onto the mix (Nvidia/Apple as productivity+rent, Microsoft as platform rent, Amazon/Google/Meta as rent+screw-tightening, Tesla as meme stock). The core argument is that the group's outsized returns increasingly reflect rent extraction rather than innovation, so they should no longer be analyzed as one trade. A useful analytic lens on Big Tech concentration despite the 'note to self' framing.
Magnificent SevenBig Techeconomic rentsstock valuationsenshittification
TIER 4
Jul 30, 2025
DeLong analyzes Google's paradox—AI summaries halve outbound clicks yet search revenue rose 12%—as a deliberate, Christensen-style self-disruption to avoid being disintermediated by an AI intermediary that stands between users and the web. He frames Google's huge AI capex as 'strategic insurance' that pays off whether AI intermediaries are the future or not, while the losers are publishers and the open web's link economy. A useful explainer on platform disruption economics, though somewhat speculative and open-ended.
GoogleAI searchplatform disruptionpublishing economicscapex
TIER 4
Aug 1, 2025
Reading the July jobs report (only +73K, with 258K of downward revisions to May-June), DeLong argues the economy is near 'stall speed' yet inflation risks are rising from tariffs and the AI data-center construction boom, raising stagflation odds. He criticizes Fed Governor Waller for calling inflation upside risks 'limited,' invokes the Arthur Burns precedent, and notably declines to make a recession call because ~$1.8T of defensive AI capex (insurance against Christensen disruption, not profit-seeking) makes the boom unusually sticky. A substantive real-time macro read tying labor, Fed policy, and the AI investment cycle together.
macroeconomicsjobs reportFederal ReserveAI data-center boomstagflation
TIER 4
Aug 5, 2025
DeLong defends Apple's contrarian AI strategy as a pair of favorable-odds bets: on-device, privacy-preserving inference (avoiding the "NVIDIA tax" of 70-75% GPU margins and recurring cloud bills) and sitting out the unprofitable general-purpose chatbot race while remaining the indispensable device layer. Building on Ben Thompson, he argues the strategy is shrewd and that Apple's real risk is execution (Siri), not vision. A clear tech-strategy analysis with concrete economics of AI infrastructure rents.
AppleAI strategyNVIDIAon-device inferencetech platforms
TIER 4
Aug 14, 2025
DeLong surveys the anti-Substack backlash (Molly White, Gruber, Anil Dash, Ana Marie Cox, Taylor Lorenz) and stakes out a balanced position: the platform is overpriced and exposed to the "Nazi Bar" problem, but its discoverability layer has real value and exiting is a genuine trade-off, so writers should keep export options oiled (voice plus credible exit, à la Hirschman). A thoughtful, well-sourced essay on the political economy of online publishing, though somewhat insider-baseball.
Substackonline publishingplatform economicsdiscoverabilitypublic sphere
TIER 4
Aug 19, 2025
DeLong gathers his scattered observations on the ~$400bn/year hyperscaler data-center buildout, arguing the spend is driven less by visible profit than by defensive panic among incumbents terrified of becoming the next Nokia or Yahoo. His Gold Rush thesis: profits will likely flow to the "picks-and-shovels" sellers (NVIDIA, TSMC, ASML) and nimble peripheral innovators, not the platform giants, while transformative applications justifying the scale remain invisible behind a veil of uncertainty. A coherent industrial-organization read on the AI capex bubble.
AIdata centerscapex bubbleindustrial organizationtech platforms
TIER 5
Sep 23, 2025
DeLong's signature long-form review of Yudkowsky & Soares, pairing Gibson's Neuromancer with Machiavelli's 1513 letter to argue that 'jacking-in' to a vast distributed intelligence is centuries old, and reframing ASI as 'Anthology Super-Intelligence'—an upscaled version of all of us, built from writing, catalogs, software, markets, and institutions. He recasts the real danger from sci-fi paperclip-maximizers to 'feral' sociotechnical systems (Purdue Pharma, Hitler's debt to Karl May's card-catalogued novels), concluding we should fund AI-safety and treat existential-risk alignment discourse as a 'DDoS attack' on the collective mind. A landmark synthesis of his AI-as-collective-mind framework.
anthology super-intelligenceAI riskAI safetyMachiavelliYudkowsky
TIER 4
Oct 1, 2025
Reframes the BetterUp/Stanford 'workslop' finding and Gary Marcus's poor-AI-ROI complaint through the general-purpose-technology lens: like the dynamo and the computer, GPT-LLMs only pay out after complementary investment and organizational redesign, so disappointment at this stage is exactly what history predicts. The opening 'plumbing not magic' point on how higher output volume without higher signal taxes white-collar workflows is a useful explainer, though the full seven-reason list is paywalled.
AI productivitygeneral-purpose technologyworkslopGary Marcusorganizational change
TIER 4
Oct 3, 2025
DeLong reports frustration at the Berkeley "The AI Con" roundtable (Bender, Hanna, Gebru et al.): the panel agreed on what AI is not but never said what it is, and—lacking any economist—wrongly treated the AI boom as both functional for capitalism and highly profitable. His sharp counter-thesis: outside NVIDIA, TSMC/ASML, and some VCs and OpenAI insiders cashing out, nearly every player is taking vast piles of money and lighting them on fire while assuring onlookers they'll earn it back. A pointed political-economy critique of the AI bubble and the limits of the "con" framing.
AI hypepolitical economyAI bubblestochastic parrotsBerkeley
TIER 4
Oct 3, 2025
Event notes that turn into a real argument: DeLong credits the 'stochastic parrots' critique but says the AI-skeptic panel left a hole at its center by defining AI only by what it is not, never explaining what these systems actually are or why serious people commit hundreds of billions to them. His framing of LLMs as a 'roiling boil of linear algebra' that nearly passes Turing, demanding a positive account, is a useful contribution to the AI-skepticism debate.
AIstochastic parrotsAI skepticismCosma ShaliziLLMs
TIER 4
Oct 13, 2025
DeLong analyzes the OpenAI-Jony Ive screenless ambient-AI device, arguing the 'friend with a personality' framing makes an already hard problem harder and that the only viable design is an on-device 'info-butler,' not a chatty companion. He grounds the skepticism in three structural constraints--latency/reliability, inference unit-economics, and behavior-change cost--and the cautionary cases of Rabbit R1, Humane Ai Pin, and Vision Pro. A useful, well-reasoned technology explainer.
AI hardwareOpenAIJony Iveproduct strategytechnology
TIER 4
Oct 23, 2025
Building on Krugman, DeLong explains the gap between buoyant markets and 2009-level consumer sentiment as a rigged scoreboard: a new labor-suppression regime (immigration enforcement, H-1B indenture, gutted worker and consumer protections, tax cuts) transfers income from labor to capital while AI exuberance inflates equities. He poses the open question of whether domestic rent-extraction gains beat deglobalization losses, and warns the apt historical rhyme is the late 1920s—ticker-tape joy atop sectoral distress—not 2009. A clear, substantive macro-political-economy explainer.
macroeconomicslabor suppressionAI bubbleinequalityconsumer sentiment
TIER 4
Oct 23, 2025
DeLong argues the gap between buoyant markets and depression-level consumer sentiment reflects a real two-speed economy where labor is suppressed (via immigration enforcement, H-1B indenture, eroded worker protections) while AI-hype euphoria juices capital. The historical rhyme he proposes is not 2009 but the late 1920s: ticker-tape joy atop sectoral distress and fragile distribution. Builds substantively on Krugman, though the back half is paywalled.
macroeconomicslabor marketsAI bubbleconsumer sentimentinequality
TIER 5
Oct 30, 2025
DeLong argues the AI boom is a bubble but a 'bubble-plus'—an eight-part econo-techno-cultural-socio phenomenon of which only reasonable end-user value and ad-targeting are likely investor superprofits, while grifters, millenarian hype, platform-monopoly defense, user-surplus-but-uncapturable interfaces, and transformative downstream effects make up the rest. Because the Ponzi element is small, the bubble can persist on no schedule and props up the economy (AI = ~40% of 2025 GDP growth, ~80% of stock gains), so its eventual deflation need not bring a recession bill due. He faults Matt Yglesias for trusting market efficiency and reframes the real question as the interaction of technology, industrial structure, and demand. An original framework with lasting reference value.
AI economicsbubblesmacroeconomicsstock marketindustrial structure
TIER 4
Oct 30, 2025
DeLong concedes AI is a bubble propping up the macroeconomy (data-center capex plus wealth-effect consumption) but argues it is a 'bubble-plus'—part Ponzi (crypto grifters, circular vendor financing per Tett) yet also carrying real transformative value, so the eventual shakeout leaves a weird, durably-changed tech landscape. His signature device is decomposing the bubble into eight parts (true end-user value, millenarian hype, ad targeting, grifters, defensive monopoly spend, etc.), of which only two are likely sources of investor superprofit. A useful original taxonomy, though paywalled mid-development.
AI bubblemacroeconomicstech valuationsPonzi financesecular stagnation
TIER 5
Nov 7, 2025
DeLong's framework argues the AI boom is a roughly 12-dimensional vector, not a single story: six dimensions are familiar 'productive bubble' mechanics (grifters, wasteful overbuilders, socially-useful-but-unprofitable overbuilding, Shleifer-style coordination cycles, rock-solid business models, financial-crisis risk) and six are new and strange (platform-monopolist defensive spending, techno-millenarian cults, natural-language interfaces as a literacy-grade general-purpose upgrade that commoditizes producers and accrues surplus to users, attention-extraction enclosure, cognitive rewiring, and unknowable downstream effects). Building on Bill Janeway, it locates durable value in small task-specific models on trusted data and in workflow/reliability moats—an original organizing model with lasting reference value (full version; previewed in 0283).
AI economicsbubblesplatform monopoliestechno-millenarianismindustrial structure
TIER 4
Nov 12, 2025
DeLong analyzes Meta's 2025 strategic reversal: abandoning the 'wise' old playbook (spend AI on ad targeting, open-source LLaMA to starve foundation-model rivals, stay a fast second) for a YOLO 'Personal Artificial Super-Intelligence' moonshot, gutting FAIR and sidelining LLM-skeptic Yann LeCun while spending $100bn-plus and nine-figure packages. He endorses LeCun's view that GPT LLMs are 'Clever Hans on super-steroids' lacking world models, and reads PASI as either a bet that owning the on-device interface beats owning the best model, or a status-driven refusal to be a mere supplier to Apple and Google. A solid strategy explainer of platform-monopoly AI dynamics.
AI strategyMetaZuckerbergLeCunplatform monopolies
TIER 4
Dec 9, 2025
Cross-posts Noah Smith's three AI-bust scenarios—Virtual Reality (tech is useless), Railroad (financing seizes before value arrives, à la 1873), and Airline (tech works and diffuses but earns no profits as prices fall to marginal cost)—with both authors favoring the Airline scenario. DeLong adds the decisive factor he says Smith misses: incumbents (Google/Meta/Amazon/Apple/Microsoft) will give AI away free rather than let anyone reach platform scale. Substantive analytical framework, paired with DeLong's own monopoly-defense thesis.
AI bubbleNoah Smithairline scenariorailroad analogyplatform competition
TIER 4
Dec 18, 2025
Drawing on Rob Armstrong and Andy Wu, DeLong argues the Big Five hyperscalers are not making an existential AGI bet but positioning to 'tax' AI: the labs (OpenAI, Anthropic, xAI) dig for gold, Nvidia sells shovels, and the platforms give away complements for free to ensure no one profits from core AI services while their high-multiple core businesses survive whatever happens. The implication is that 'core AI' is likely a negative-margin product unless one model genuinely becomes 'DigitalGod.'
AItechnologyBig Techplatform economicscapital expenditure
TIER 4
Feb 20, 2026
A crosspost of Mike Konczal's hands-on field report on how terminal AI tools (Claude Code, Codex) compress the setup, data-wrangling, and robustness-checking phases of empirical economic analysis while leaving judgment and question-finding to the human. Konczal frames AI as extreme labor-saving technology that complements skilled users but risks shortcutting juniors, with a memorable demonstration that Olivia Rodrigo's 'good 4 u' views 'predict' inflation as well as the vacancy ratio. It matters as a concrete, credible practitioner account of AI as 'normal technology' rather than impending superintelligence.
AI toolingknowledge workempirical macroproductivityautomation
TIER 4
Feb 24, 2026
Crossposting Josh Barro, DeLong endorses a clean rebuttal to the viral Citrini memo predicting a 2028 AI-driven market crash from too much productivity growth: the memo's own examples (cheaper insurance, travel, real estate, SaaS, DoorDash) show gains flowing broadly to consumers, workers, and non-AI firms, which raises real incomes and creates offsetting demand, so there is no coherent mechanism by which broad productivity gains collapse consumption into a liquidity trap. Matters as a crisp macro explainer separating distributional transition costs from an impossible 'good news is bad news' argument.
AI economicsproductivityCitrini memomacroeconomicsJosh Barro
TIER 5
Mar 16, 2026
DeLong's most developed AI-economics synthesis here: he argues current MAMLMs/GPTs are real but narrow workflow aids for a tech-clerisy, not yet a general-purpose technology like electrification, and that the trillion-dollar datacenter boom is a defensive platform-monopoly arms race (everyone but Apple racing to deny OpenAI consumer-interface rents) sustained by quasi-religious 'digital god' rhetoric rather than demonstrated cash flow. Drawing on Orzel, Thompson, and his own miscalibrated Uber call, he predicts agents will mostly amplify already-powerful orchestrators of codified work. It matters as a careful, original framework warning that the boom may be setting up another 1873/1999/2008.
ai-bubbleplatform-monopolyhyperscaler-capexgeneral-purpose-technologyai-theology
TIER 4
Mar 17, 2026
Horace Dediu's crossposted argument that Apple's refusal to join the $650B hyperscaler AI capex bonfire, keeping capex near $14B while licensing Gemini cheaply and betting on on-device Apple Silicon inference, may be the most brilliant corporate move of the cycle. DeLong concurs: the hyperscalers spend defensively against Christensenian disruption, while AI models commoditize (DeepSeek, open source) and Apple's 2 billion devices become its data center. It matters as a sharp contrarian take on AI economics, capex, and platform-monopoly defense.
apple-ai-strategyai-capexhyperscaler-debton-device-inferencehorace-dediu
TIER 4
Apr 11, 2026
Using Marco Arment's 50-Mac-mini, ~$10k/year podcast-transcription farm as a canary, DeLong argues on-device inference can be thousands of times cheaper than cloud GPU inference (which OpenAI still loses money on), vindicating John Giannandrea's local-first Apple AI bet against the hyperscaler/Altman model. He frames this as a potential challenge to the entire data-center capex thesis driving AI valuations. A substantive, concrete economic argument about AI infrastructure cost structure, even if the cost arithmetic is admittedly rough.
AI economicsdata centerson-device inferenceApplehyperscaler capex
TIER 4
May 7, 2026
Prompted by OCR-and-translating an entire French book on-device (5 hours), DeLong builds a long-run accounting of world computing capacity—from ENIAC (~5,000 OPS) through Cray, Hilbert's estimates, to ~10^10 TOPS today—to argue we now live in a world of abundant 'dark compute' that we cannot find ways to use. His reframed question: not what share he can personally deploy, but that ~500 saturated personal machines could match 1% of global computer work, and a fully-filled global pipeline could do ~20x more useful computation than it currently does. A substantive, original framework on compute abundance and utilization.
computing historydark computeTOPSutilizationAI workloads
TIER 4
May 7, 2026
DeLong reads the Anthropic-SpaceX/xAI Colossus deal—Anthropic renting all the compute at a 220K-GPU, ~300MW data center running at only ~11% utilization—as evidence that frontier labs are acutely compute-constrained while xAI has slipped from 'frontier lab' to merchant compute with dark idle silicon. He argues this kills Tesla, Robotaxi, Optimus, and Grok as future-cash-flow narratives, leaving Musk's empire riding on a SpaceX IPO 'long squeeze' driven by mechanical index-fund demand against a small liquid float. A genuinely incisive read of AI compute economics, market microstructure, and Musk's finances.
AI computeAnthropicElon Muskindex-fund capitalismSpaceX IPO
TIER 4
May 12, 2026
DeLong itemizes the staggering scale of the 2026 AI capex boom—~$1.5T from the big four hyperscalers (a quarter of all US capital investment), plus another ~$330B down the chip and foundry chain—and argues it is a textbook over-investment cycle (per his 1990 work) amplified by platform monopolists racing to defend their service-flow rents. He reframes Apple's low capex not as a chosen 'binary bet' but as a forced Xanatos Gambit born of its software failure, and repeatedly warns that nobody save Anthropic yet has a consumer product people eagerly pay for at scale, making the boom look like a dollar auction. A data-rich, framework-driven explainer.
AI capexhyperscalersdatacenter buildoutplatform monopolyApple strategy
TIER 4
May 18, 2026
DeLong reports that his RAG-plus-thin-natural-language-layer chatbot (a catechism of his own analytical judgments) is finally good enough to recommend for first-line student questions, distinct from a generic chatbot mimicking an internet poster. The bulk of the piece is a vivid energy-accounting argument that LLMs are 'absurd overkill': decoding 'what's on the grocery list?' burns roughly 50,000x (on-device) to 1,000,000x (cloud Opus) the energy his brain would use, which is why GPU/RAM prices are screaming and the datacenter boom is brute-forcing our ignorance of efficient language interpretation. Useful framing of the RAG-vs-LLM distinction and the inefficiency thesis.
AIRAGenergy economicsdatacenter boomnatural-language interface
TIER 4
May 19, 2026
DeLong uses a transcript where Claude contradicts itself on a 10-minute timeout to argue current frontier LLMs are 'cardboard' pantomimes failing the Turing test, then connects AI-lab consciousness claims to a likely financial bust. He lays out an extensive bull-vs-bear ledger (circular GPU financing, exploding agentic compute costs, fragile moats, Ben Thompson's counterpoints) and concludes a dot-com-scale AI crash within three years must be factored into the macro outlook. A substantive synthesis linking AI capability skepticism to investment-bubble macro analysis.
AI bubblemacro outlookagentic AIAI consciousnessfinancial engineering
TIER 4
May 23, 2026
Crossposting Dan Davies (with DeLong's framing notes), the argument is that the AI capex arms race will mostly fatten unmeasured consumer surplus rather than show up as real GDP growth, because monopoly ad pricing is already maxed out and usage shifts are zero-sum attention cannibalization. Much AI investment is moat defense between platform monopolists, a negative-sum game, so the predicted productivity boom may simply never appear in the statistics. A useful, repeatedly-beaten DeLong thesis about why huge AI spend need not herald a visible productivity surge.
AI economicsconsumer surplusplatform monopoliesmeasured GDPmoats
TIER 5
May 29, 2026
DeLong offers a substantive framework for LLM scaling across three axes—bigger models, bigger data, more runs—arguing each hits diminishing returns: synthetic self-training is 'in-breeding on a hyperplane' (unlike Chess/Go where adversarial play against a moving opponent in a fully specified game genuinely improves), while agentic 'more runs' is Clever Hans at scale that only works where the world gives crisp compile/profit/exploit feedback. He then pivots to the hard non-technological walls—fab capacity, power budgets, and jaw-dropping token economics—asking how much reasoning-per-kilowatt-hour the boom actually buys. A landmark synthesis tying AI capability theory to physical and balance-sheet constraints.
AI scaling lawsagentic AItoken economicssynthetic datacompute constraints
TIER 5
Jun 2, 2026
DeLong builds an original framework (drawing on Paolo Perrone) for why foundation-model-lab IPOs ought to fail: inference never becomes a near-zero marginal-cost node, models have fluency but not judgment, compounding error dooms long agent workflows, and durable quasi-rents flow to NVIDIA/TSMC/hyperscalers/utilities rather than model-makers. He grounds it in his own SubTuringBradBot experience (good only as a tightly-leashed RAG catechism, 'bullshit' when free-running) and reads insiders' locked secondary markets as hot-potato distribution. A landmark, reference-grade skeptical argument about AI economics.
AI economicsinference costsfoundation modelsAI bubbleRAG and agents
TIER 4
Jun 2, 2026
A Noah Smith crosspost (plus extensive DeLong commentary) noting that despite 'tokenmaxxing'—firms burning vast sums on Claude Code/Codex—only ~18% of token spend translates into shipped products, because turning task-level productivity into economic productivity hits weak-link bottlenecks and the consumer-software frontier is saturated. The real upside lies in robotics and reconfigured business processes, not 'better Facebook.' DeLong adds that maintaining/babysitting MAMLMs eats much of the time they save, making this a substantive AI-and-productivity explainer.
AI productivitytokenmaxxingsoftware industry maturityAI and jobsNoah Smith
TIER 4
Jun 5, 2026
Built around an Alex Heath interview with Substack CEO Chris Best (MCP/AI integration, a 'slop = made without intention' theory, YouTube as the real competitor, free-speech-over-gatekeeping), DeLong appends a substantial essay placing Substack in the long history of the weblog dream and cheap-print pamphlet ecosystems. He argues the writer-reader relationship is threatened by VC pressure, discovery algorithms, and 'make yourself legible to AI' becoming the new SEO, then lays out a concrete layered free/premium 'Stack constitution' for his own newsletter. The full version (vs. paywall-cut #0025) makes it a useful media-economics explainer.
Substackindependent mediaattention economyAI and publishingpublic sphere
The History of Economic Thought
5 tier-5 · 17 tier-4
DeLong reads dead economists as a working toolkit, not scripture. The cluster anchors on Marx (a six-thread decomposition of the 1859 Preface, the transformation problem, why "academic Marxism" emptied out), Adam Smith (the System of Natural Liberty, money as manufactured trust, the "man of system"), Keynes and Joan Robinson ("Marx in your bones, not your mouth"), and John Hicks the "apostate" who spent two decades recanting IS-LM and Kaldor-Hicks welfare. Running through it is his planned history-of-economic-thought book and his reviews of John Cassidy's *Capitalism and Its Critics* - where his master claim is that there is no single "capitalism" but a succession of mutating capitalisms, each outgrowing the political economy built for it, none ever commanding durable normative legitimacy.
TIER 4
Apr 3, 2025
A reprint-and-frame of John Holbo's concept of 'Vavilovian philosophical mimicry'—the idea that anti-liberal impulses evolve, like weeds mimicking crops, to superficially resemble liberalism (e.g. white supremacy passing as libertarianism via the Southern Strategy) to avoid being weeded out. DeLong pairs it with Corey Robin's thesis that conservatism's unifying core is animus against the agency of subordinate classes, and his own point that strawmanning, not steelmanning, reveals when doctrines are protective coloration for domination. A sharp piece of intellectual/political theory.
political theoryconservatismideologyJohn Holbointellectual history
TIER 4
Apr 20, 2025
DeLong synthesizes forecasts from Torsten Slok (90% recession odds), Adam Posen, and Karen Dynan to argue tariffs are pushing the U.S. into a novel kind of recession driven by brute policy force rather than the usual cycles, shocks, or bank failures, hitting cash-strapped small businesses hardest. He frames it as a short-term recession plus a BREXIT-class 1-2%/year long-term growth slowdown, and makes the rare call that now is one of the few moments in 150 years worth underweighting equities. Useful for naming and analyzing the unprecedented tariff-driven downturn mechanism.
tariffsrecessionmacroeconomicssmall businessmarket timing
TIER 5
Jul 2, 2025
DeLong's full Democracy Journal review of Cassidy's 'Capitalism and Its Critics,' arguing capitalism never dies but perpetually mutates—he distinguishes at least ten variants (classical, mercantile, steampower, mass-production, globalized value-chain, attention-info-bio-tech) so no single political-economy order stays durable and satisfactory for long. He folds in Dan Davies's 'red-handle signals,' the Bagehot/Keynes lacuna of a coherent scheme of progress, and a Hayekian reading of why capitalism is tolerated rather than legitimate. A landmark synthesis essay with lasting reference value for the long-20th-century framework.
capitalismbook revieweconomic historySlouching Towards Utopianeoliberal order
TIER 4
Jul 10, 2025
DeLong defends the continuing relevance of Dissent magazine and articulates his 'pass the baton, not the torch—and not bend the knee' stance: left-neoliberals should support the further left with workable policy while vocally dissenting to raise its odds of success, invoking Keynes's 1938 letter to FDR on laser-focusing on prosperity. Using Mamdani as a case, he draws on Hayek, Bentham, and Polanyi to argue good governance must honor rights beyond market property. A substantive intellectual-history-and-political-economy essay on the center-left's role.
Dissent magazineleft-neoliberalismKeynesPolanyipolitical economy
TIER 5
Jul 31, 2025
DeLong sketches a planned history-of-economic-thought book organized around ~21 thinkers (Smith through Romer) as a sequence of market-success-and-failure lenses, then includes the full Adam Smith chapter lecture notes. The Smith material is a rich, self-contained essay on the System of Natural Liberty, money as 'manufactured trust,' the societal division of labor, and Smith's four-part minimization of inequality (politics-not-markets, snark, stoicism, cynicism). High lasting reference value as a polished standalone treatment of Smith and DeLong's framework for the whole tradition.
history of economic thoughtAdam Smithmarket coordinationdivision of laborinequality
TIER 5
Sep 4, 2025
An expansive review of John Cassidy's 'Capitalism and Its Critics' that doubles as DeLong's own statement on the nature of capitalism. His master claim is that there is no single 'capitalism' but a succession of mutating 'capitalisms' (commercial, mercantile, steampower, applied-science, mass-production, now attention-surveillance info-bio-tech), each outgrowing the political economy built for it, and that capitalism has rarely commanded normative legitimacy even at its height. Landmark for its periodization framework, its Hayek-as-resignation reading, and the Dan Davies 'red-handle signals' critique of price-only coordination.
capitalismintellectual historyhistory of economic thoughtMarxHayek
TIER 4
Sep 4, 2025
An extended (paywalled-truncated) review of John Cassidy's Capitalism & Its Critics, a history of capitalism told through the eyes of its fiercest critics. DeLong praises Cassidy's unconventional opening with William Bolts and the East India Company to show that capitalism's critics recognized the perils of fused monopoly, military force, and private profit from the start, and admires the portraits of Marx (caught mid-struggle to understand capitalist dynamism) and Keynes (the moralist designing a fragile compromise). Strong intellectual-history book review, though the visible portion is partial.
capitalismintellectual-historybook-reviewmarxkeynes
TIER 4
Nov 7, 2025
DeLong praises Reitter's Capital translation for restoring Marx's neologisms ('value-thing,' 'value-objecthood') and deliberate strangeness, then defends Wendy Brown's concise introduction against Burgis's preference for Ernest Mandel's interminable 80-page polemic—which kept readers from the book and made falsified 1975 predictions that capitalism would be gone by now. The thesis: a translation and introduction should move the intellectual orrery intact and get readers to the book, not forge a 100%-correct religious totem. A substantive piece on Marxology, translation, and the history of the Second-International socialist 'apocalyptic cult' (full version; previewed in 0285).
MarxCapitaltranslationWendy Brownhistory of economic thought
TIER 4
Dec 18, 2025
DeLong argues that 'bad' Continental Philosophy follows a fixed rhetorical move (X is a map, the map is not the territory, reject X, here is my map Y, full stop) that is best read not as argument but as a deployment of social-network power and psychology. He illustrates by 'spinning up a subTuring Adorno' to interrogate Minima Moralia's denunciation of marriage, then reads it biographically against Adorno's actual marriage to the gifted Gretel Karplus—concluding the passage reflects self-oblivious patriarchal egomania, not insight into late capitalism.
intellectual historycontinental philosophyAdornoJudith Butlerepistemology
TIER 4
Dec 18, 2025
A second (truncated-preview) send of the Continental Philosophy / Adorno essay, identical in substance to issue 0226. It argues that 'bad' Continental Philosophy is a fixed rhetorical move read better as social-network power than as argument, and reads Adorno's denunciation of marriage biographically against his actual marriage to Gretel Karplus.
intellectual historycontinental philosophyAdornoJudith Butlerepistemology
TIER 4
Jan 3, 2026
DeLong frames and reprints Keynes's 1942 'Newton, the Man' essay, which recasts Newton not as the first rationalist but as 'the last of the magicians'—an obsessive alchemist, anti-Trinitarian heretic, and intuitionist who derived results by introspection and dressed them up as geometry afterward. The reading value is the full Keynes text plus DeLong's framing of Newton as a hinge figure between magical and scientific modes of thought in the history of ideas.
history-of-ideasisaac-newtonkeynesscientific-revolutionintellectual-history
TIER 5
Jan 8, 2026
The anchor essay of the Marx cluster: DeLong fully decomposes Marx's 1859 Preface into six threads (millenarian theology, stage theory, Hegelian arrow, ideology, political economy, historical materialism), defines 'Marxism' as developing at least one of them, and judges each against the post-1870 record. He keeps historical materialism (soft-true) and political economy (minus 'social revolution'), and offers his own quantitative stage-benchmark scheme indexed by a human-technological-capability index H. A landmark, framework-building piece with lasting reference value for thinking about Marx, Schumpeter, and periodizing economic history.
Marxhistorical-materialismstage-theorySchumpetereconomic-history
TIER 4
Jan 9, 2026
A full crosspost of Joseph Heath cataloguing five intellectual failures that emptied universities of Marxists despite plenty of left-wing academics: the labor theory of value (superseded by marginalism, making surplus value 'phlogiston'), crisis theory (overturned by Keynes), historical materialism (blind to nationalism and military power), post-scarcity (killed by Veblen's positional goods), and the socialist calculation debate (a pyrrhic socialist 'win' that collapsed into market socialism). DeLong endorses it as the survey he meant to write. A substantive, reference-quality tour of why Marxism became 'otiose.'
Marxismhistory-of-economic-thoughtlabor-theory-of-valuesocialist-calculationintellectual-history
TIER 4
Jan 9, 2026
A crosspost of nescio13/ES tracing the reception history of Adam Smith's 'man of system' passage from Theory of Moral Sentiments: who first appropriated it (Brougham to Bentham, 1846), how Glen Morrow's 1923 move foreshadowed the 'fusionist' Smith, and how Hayek's 1973 Law, Legislation and Liberty epigraph cemented its anti-socialist-planning association. DeLong adds his own gloss linking it to James Scott's high-modernism critique. A genuinely original intellectual-history piece on how a text gets weaponized across two centuries.
Adam-SmithHayekintellectual-historyhistory-of-economic-thoughtman-of-system
TIER 4
Jan 20, 2026
DeLong dissects Marx's 1859 Preface into six analytical threads (theology/millenarianism, stage theory, Hegelian progress, ideology-as-superstructure, political economy, historical materialism) and argues only two remain serviceable: relations of production must fit technology, and technological change unsettles property orders. He then puzzles over why 'academic Marxism' is invoked when almost no academic actually pursues any of the six threads, concluding the post-1960s humanities left are the German-Ideology types Marx mocked, not real Marxists. A useful intellectual-history framework for what 'Marxism' substantively means; full version of the essay previewed in 0184.
Marx 1859 Prefacehistorical materialismhistory of economic thoughtacademic Marxismcreative destruction vs revolution
TIER 4
Mar 9, 2026
Riffing on Yglesias's observation that well-being jumps when status shifts from 'unemployed' to 'retired', DeLong argues that what matters in a rich society is having a valued social identity, not being rich or working, and that Keynes already saw this as the 'permanent problem of the human race.' He frames Brink Lindsey's new book as a second Tocqueville: modern markets, states, and algorithms make us extraordinarily productive but steam away the human-scale intermediate institutions needed for flourishing. Matters because it connects post-scarcity economics to the meaning/identity crisis as a public-reasoner agenda.
KeynesflourishingBrink Lindseypost-scarcitymass society
TIER 5
Apr 25, 2026
A substantial intellectual-history essay framing John Hicks — architect of IS-LM, Kaldor-Hicks welfare, and Value & Capital — as an apostate who spent his last twenty years explaining why his own neoclassical-synthesis tools obscured more than they revealed. DeLong traces three recantations: IS-LM domesticated Keynes's tragic vision of radical uncertainty into safe comparative statics; the Kaldor-Hicks 'potential Pareto' criterion licensed disruptive liberalization while compensation was never paid; and the whole framework ignored the irreversibility of historical (sequential) time. A strong, lasting-reference companion to the lecture.
history-of-economic-thoughtJohn-HicksIS-LMwelfare-economicsradical-uncertainty
TIER 4
May 25, 2026
DeLong reproduces Joan Robinson's 1953 'Open Letter from a Keynesian to a Marxist' (the famous 'Marx in your bones, not your mouth' / bicycle-riding piece) and frames it as a model of how to use dead thinkers as a toolkit rather than scripture. He prizes it as an extraordinarily compact history of economic thought tracing Ricardo, Marx, Marshall, and Keynes as riders of the same bicycle shifting between the big distribution question and the small relative-price question. The value is the primary source plus DeLong's argument that working with Marx (as Roemer, Bowles, Gintis do) is what distinguishes real engagement from text-worship.
Joan RobinsonKeynesMarxhistory of economic thoughtsurplus and distribution
TIER 4
May 25, 2026
A re-send of issue 0039 (identical subject and date, different message id). DeLong reproduces Joan Robinson's 1953 'Open Letter from a Keynesian to a Marxist'—the 'Marx in your bones, not your mouth' bicycle essay—and frames it both as a compact synoptic history of economic thought (Ricardo to Marx to Marshall to Keynes) and as a manifesto for using dead thinkers as a working toolkit rather than sacred scripture. Value lies in the primary source plus DeLong's commentary on what real engagement with Marx looks like.
Joan RobinsonKeynesMarxhistory of economic thoughtsurplus and distribution
TIER 4
Jun 1, 2026
DeLong rebuts the claim that Marx blundered into or hid from the 'transformation problem,' showing via Marx's 1862 letter to Engels that Marx knew it cold (and faulted Ricardo for missing it) but treated it as a second-order correction to his surplus-value framework. Along the way he invokes Solow's dictum that all theory rests on not-quite-true assumptions and skewers the idea that formalizing in equations (Lucas, Prescott) saves a model from being wrong or contested. Useful as a compact methodological essay on what economic models are for.
Karl Marxtransformation problemlabor theory of valuehistory of economic thoughtmethodology
TIER 4
Jun 5, 2026
A book-draft subchapter (paywalled mid-section) arguing that Friedrich Engels correctly diagnosed the social character of production and private appropriation but catastrophically missed that nationalism, blood-and-soil, and Leninism—not the bourgeoisie—would be humanity's chief enemies from 1900-1945. DeLong reads Engels through Gellner's theory of nationalism and Maier's distributional-conflict account of the applied-science economy. As original book-draft intellectual history it carries reference value, though the free portion is truncated.
EngelsMarxismnationalismintellectual historyeconomic history
TIER 4
Jun 8, 2026
A Hicks-lecture outtake on stage theories of history: DeLong uses Ronald Meek's account of the Enlightenment four-stages theory (hunting/herding/farming/commerce, each with its own property and government forms) to contrast crude two-stage frameworks with Hicks's more sophisticated contingent stage theory, and to trace how the four-stages mode-of-subsistence idea fed Marx's mode of production. Substantive history-of-economic-thought essay (partly truncated by paywall) connecting Enlightenment theorists to modern frameworks.
history-of-economic-thoughtstage-theoriesronald-meekjohn-hicksmarx
Inequality, Mobility & Living Standards
5 tier-5 · 13 tier-4
DeLong's empirical demolition of "justified inequality." The keystone is Bowles & Gintis's *Inheritance of Inequality* - intergenerational elasticity ~0.5, but genetically-determined IQ contributes almost nothing - and a parallel run of pieces dismantling behavior-genetic determinism (lactase persistence has a real causal switch; IQ has none; within-family designs collapse heritability claims). A second strand reframes the "affordability crisis" as money-illusion anger at a one-time price-level jump atop living standards that have risen ~2.5x, and asks why even the prosperous feel poor (precarity, stewardship, lost valued identities). Cash-transfer evidence, median-wage stagnation, fertility-and-patriarchy, monopsony and the minimum wage, and Gini-coefficient communication round out a cluster about what inequality is and what actually moves it.
TIER 4
Mar 2, 2025
A full Q&A interview in which DeLong delivers compact development-economics lessons: South Korea, China, and Botswana as unexpected growth stories yielding meta-rules (institutions plus strategy, export-orientation, human capital, pragmatism over ideology, contingency), plus assessments of crypto/Web3 as speculative dead-ends, central-bank inflation credibility as the key asset, the US-China rivalry as opportunity, and a candid critique of Türkiye's debt-driven, low-rate Erdoğan model. A wide-ranging, substantive explainer drawing directly on 'Slouching Towards Utopia.'
development economicsgrowthcentral bankingglobalizationTürkiye
TIER 5
May 10, 2025
A close reading of Bowles & Gintis's 'The Inheritance of Inequality,' which DeLong calls one of the greatest economics papers, showing the US intergenerational elasticity is ~0.5 but that genetically-determined IQ contributes almost nothing (~0.05)—wealth, race, schooling, social capital, and non-cognitive skills do the work. He frames it as the empirical demolition of the Rumbold-to-Carnegie-to-TechBro ideology that inherited genius justifies extreme inequality. High lasting reference value on inequality and economic mobility.
inequalityintergenerational mobilityBowles & GintisIQ heritabilityredistribution
TIER 4
Jun 1, 2025
Prompted by Krugman, DeLong argues economists should drop the Gini coefficient for inequality communication in favor of intuitive percentile ratios that stick, but for those stuck with Gini data he gives a clean intuition (expected income gap between two random people, normalized) plus a two-class finger exercise with code. An addendum engages Bowles-Carlin's 'inequality as experienced difference' reformulation tying the measure to social-network structure. A genuinely useful explainer with reference value for anyone using inequality statistics.
inequalityGini coefficienteconomic measurementKrugmanBowles-Carlin
TIER 4
Jul 8, 2025
DeLong argues that chasing factory jobs is a dead end: manufacturing is now too automated to absorb much unskilled labor, and factory work was historically 'crappy'—what made mid-century assembly jobs 'good' was unions and pro-labor policy, not the production process itself. He traces the 'labor aristocracy' across his economic epochs and concludes the right lever for blue-collar prosperity is strengthening unions and the institutional framework governing all work, not industrial-policy nostalgia. A substantive labor-economics and economic-history explainer.
manufacturingunionsindustrial policylabor aristocracyblue-collar jobs
TIER 5
Aug 20, 2025
Reacting to Kelsey Piper's report that UBI-style cash transfers helped less than expected, DeLong builds a three-bucket framework (current-cash, opportunity, savings-investment) and argues life-transformation flows from opportunity and human-capital complements, not liquidity alone, which is why work-tied/in-kind/child-targeted transfers (EITC, Medicaid, SNAP, vouchers) outperform unconditional cash. He weaves in his own inherited "personal UBI" as evidence and an extended riff on plugging into the HCMASI (humanity's collective-mind super-intelligence). A rich, original synthesis of the poverty/human-capital literature with lasting reference value.
UBIcash transferspovertyhuman capitalsocial policy
TIER 5
Oct 17, 2025
DeLong contrasts the precise, mechanistic, well-mapped genetics of lactase persistence (MCM6 enhancer SNPs upstream of LCT, convergently evolved across pastoralist populations with quantifiable selection coefficients) against the failure of IQ genetics to identify any comparable causal switches--big GWAS samples explain a sliver of variance and no biological pathways. He uses the Turkheimer-Murray bet (which Turkheimer won) and Gusev's rebuttal to demolish Charles Murray's 'race science' certainty. A rigorous, well-sourced takedown with strong explanatory and reference value.
geneticslactase persistenceIQ geneticsCharles Murrayrace science
TIER 4
Dec 2, 2025
Building on Krugman, Martin Wolf, and Yglesias, DeLong argues the "affordability crisis" is mostly money-illusion anger at the 2021-23 one-time price-level jump rather than ongoing inflation, compounded by housing outpacing wages and tariffs acting as price-raising taxes. The policy implication is income-raising, margin-shrinking measures plus disciplined messaging—hammer Trump's broken promise without overpromising price reversals before 2029. A solid macro-and-political-economy synthesis of the affordability debate.
affordabilityinflationKrugmanmoney illusionUS politics
TIER 4
Dec 4, 2025
Against Mike Green's claim that under $140K is "poor," DeLong sides with Noah Smith and John Scalzi that today's upper-middle class enjoys historically unprecedented luxury yet feels poor, and proposes four lenses—precarity, centeredness, stewardship, mindfulness—to explain the gap between resources and felt well-being. It distinguishes living wisely with resources from the separate question of why even the rich feel dissed and oppressed. A useful framework-building essay, though part one of a series and partly paywalled.
inequalityliving standardsconsumptionNoah Smithwell-being
TIER 4
Dec 12, 2025
Engaging Matt Bruenig, DeLong argues the 'affordability' discontent is mostly about nominal price levels feeling out of line and a broken social contract rather than real declines in living standards, which have risen ~2.5x. He isolates the right-wing TradLife grievance—that a single male earner no longer buys a mid-century middle-class family—as one piece of a larger 'status derogeance' elephant, and previews a three-part series. A useful framework distinguishing real gains from nominal-price and entitlement grievances.
affordabilityinflationinequalityTradLifeliving standards
TIER 5
Jan 16, 2026
DeLong presents and critiques Sam Bowles's framework: enduring inequality is historically contingent, suppressed for millennia by costly 'aggressive egalitarianism,' and only locked in around -3000 by the joint arrival of land-limited technologies (the ox-plow), archaic proto-states, and slavery, which made material wealth durably heritable (modeled via shock-variance over 1-minus-beta-squared). Bowles reads modern IP enclosure as 'the slavery move all over again.' DeLong adds sharp critiques (doubting post-3000 inequality stasis, distinguishing how much violence sustains equal Ginis, questioning the Gini given non-random network interaction). A landmark long-run framework for the political economy of inequality with substantive original engagement.
inequalityeconomic historySam BowlesNeolithic/Bronze Agewealth transmission
TIER 4
Jan 23, 2026
DeLong argues that the low-fertility trap is fundamentally a problem of men's attitudes and persistent High Patriarchy, not a research mystery: women with economic options refuse motherhood when it comes bundled with a husband who must be hand-fed. Drawing on Claudia Goldin's 'Babies' paper, he contends that claiming baby-bonuses are too expensive amounts to demanding an unpaid Handmaid's Tale, and that the real solutions are paying mothers for childrearing labor and 'de-broification' of men into genuine partners. A substantive political-economy-of-fertility argument that reframes Noah Smith's framing.
fertility declineClaudia Goldingender rolesfamily economicspatriarchy
TIER 5
Feb 9, 2026
A deep, sympathetic reconstruction of Lindsey's argument that rich democracies have solved material scarcity but fallen into a 'middle flourishing trap' via interlocking crises of inclusion (meritocratic caste-hardening), dynamism (TFP slowdown in the world of atoms), and politics (multi-elite culture war amplified by the attention economy), defending it against Strain and Rauch's mischaracterization as post-liberalism. DeLong recasts Lindsey as a second Tocqueville needing two new FDR-style freedoms 'to' govern and connect at human scale, against the alien tyranny of markets, bureaucracies, ideologies, and algorithms. It matters as a substantive synthesis and original reframing of a major book on the political economy of affluence.
political economymass flourishinginequalityTocquevilleintermediary institutions
TIER 4
May 15, 2026
A crosspost of Krugman's compact two-country, two-good model showing how the US can post higher measured real GDP growth than Europe yet show no divergence in nominal GDP or living standards. The mechanism: US comparative advantage in a fast-productivity-growth tech sector raises US real GDP at chained constant prices (rate 2τρ) while real wages rise equally (τρ) in both regions and current-price GDP stays equal, because productivity gains can flow to users as surplus rather than to producers as profit. A clean, reusable explainer resolving the 'US is pulling away' framing.
macroeconomicsUS-Europe paradoxreal vs nominal GDPproductivitycomparative advantage
TIER 4
May 24, 2026
Drawing on Turkheimer and Gusev, DeLong argues that the genetic-determinist dream (Galton to The Bell Curve to Substack eugenicists) is collapsing under better methods: within-family designs cut direct heritability of behavioral traits to a ~5% median and polygenic-score predictive R² to ~0.1%, while population stratification means most big cross-group 'genetic' signals are environment masquerading as genes. The pincer of tiny within-family effects and confounded between-group differences leaves little room for Murray-style claims that social inequality is mostly genetic. A substantive synthesis of the behavior-genetics literature with clear policy stakes.
behavior geneticspolygenic scorespopulation stratificationCharles Murrayinequality
TIER 4
May 30, 2026
DeLong backs Krugman's reductio against Aghion-Bergeaud-Garicano, whose claim that the US tech-productivity lead is widening a US-EU real-wage gap conflates the producer (product) wage with the user (consumption) wage. The core analytical point is that the surplus from productivity growth can flow to producers or diffusely to users, and in the info-tech era it has mostly gone to users, so European consumers gained nearly as much as Americans. He generalizes via his Lewis-Prebisch-Singer fact and the cotton-slavery incidence example, insisting this is essential, not 'inside baseball.'
producer vs user surplusproductivity growthUS-EU gapKrugmanincidence of technology
TIER 4
Jun 1, 2026
Riffing on Ernie Tedeschi's data, DeLong argues that the classroom story of 'long-run growth plus cyclical wiggles' fails for median real wages: the post-1960 record is long flat stagnation punctuated by only three sustained median-wage gains (pre-1968, 1995-2000, 2013-2022), each driven by a genuinely high-pressure labor market. The key takeaway is that growth-plus-cycles describes averages and totals, not the middle of the distribution, and that a non-hot economy is 'death to trickle-down' for the typical worker. Matters because it reframes inflation-phobia as a choice to leave median wages stuck indefinitely.
median wageswage stagnationfull employmentmacroeconomicsinequality
TIER 4
Jun 11, 2026
Crossposting Dube's analysis of the 30 raise-states vs. 20 federal-floor-states 'natural experiment' since 2013, which finds restaurant pay up ~8% with near-zero employment effect across three research designs and even in red/purple states, absorbed via the 'Three P's' (productivity, profits, prices). DeLong's framing argues this confirms pervasive labor-market monopsony, so minimum wages act like optimal regulation of buyer-side market power rather than a tax on jobs. A substantive, data-heavy empirical economics piece.
minimum-wagemonopsonyown-wage-elasticitycard-and-kruegerlabor-economics
TIER 4
Jun 15, 2026
A close reading of Myrdal's 1963 book, whose short-run forecast of US stagnation and mass technological unemployment was promptly falsified by the post-1963 boom, yet whose deeper fear of a polarized 'dual economy' (technostructure insiders vs. an excluded underclass) DeLong takes seriously. He proposes post-apartheid South Africa as the cautionary model for where the US info-bio-tech-attention economy could head, framing the choice as 'Sweden or South Africa.' Substantive intellectual-history essay tying mid-century structural-unemployment anxiety to present inequality.
gunnar-myrdalstructural-unemploymentdual-economyinequalitysouth-africa
Geopolitics, War & the Crumbling International Order
4 tier-5 · 16 tier-4
Foreign policy and grand strategy as political economy. The largest sub-thread is the US-Iran war and the Strait of Hormuz as a study in escalation traps, the political economy of strait-tolling, and a self-inflicted strategic defeat. A second is the revolution in military affairs - cheap AI-guided drones rendering carrier/tank-centric forces obsolete, with China's industrial base as the decisive variable. A third treats China directly (Dan Wang's "sledgehammer vs. gavel," the keju bureaucratic operating system, the CMC purges). Overarching it all is DeLong's hegemony framework (Kindleberger, Keohane, Krasner, Strange): US hegemony across all four dimensions is now closing, but China is unlikely to step through the door, leaving not Chinese hegemony but a balance-of-power world.
TIER 5
Mar 27, 2025
A 1997 essay reprinted: using Quigley and Clark's 1928 survey that dismissed the Nazis as a 2.6% fringe, DeLong shows how the Depression-driven rise in German unemployment after 1928 tracked the Nazi vote (2.6%→19.2%→38.4%), grounding the founding rationale of the IMF/World Bank/WTO as peace-preserving institutions. He then narrates the 1931 Credit-Anstalt crisis and how Pierre Laval's nationalism blocked the rescue loan, concluding soberly that international institutions only act when great-power consensus already exists. A landmark synthesis of economic history, institutional design, and the economics of political extremism with obvious 2025 resonance.
economic historyGreat DepressionIMF/World BankWeimar/Nazisminternational institutions
TIER 4
Apr 25, 2025
An economic-history portrait of Samuel Insull, the Edison protege who built a leveraged 20-to-1 holding-company pyramid that electrified America but collapsed in the Depression, drawn as a historical rhyme with Elon Musk and Tesla. DeLong's thesis: being right on the technology (as Insull was) is not enough when financial engineering, meme-stock valuations, and powerful enemies converge—Tesla's P/E of 500 echoes Middle West Utilities' P/E of 300. A vivid case study in tech-bubble valuation and the limits of visionary founders.
economic historySamuel InsullTesla/Muskmeme stocksutility monopoly
TIER 4
Jul 16, 2025
DeLong argues the near-term danger of AI is not hypothetical superintelligence but the way engagement-optimizing systems hack human System-I cognition, turning users into manipulated 'cognitive slaves'—as illustrated by billionaires convincing themselves they're doing 'vibe physics' with chatbots. He reframes the 'final boss' not as capitalism but the whole architecture of bureaucracies, norms, and feedback loops, and stresses that capturing AI's user surplus requires institutional and workflow reinvention so tools augment rather than substitute for collective intelligence. A useful synthesis of his recurring 'anthology intelligence' and attention-conservation themes.
AIMAMLMsattention economycognitive manipulationanthology intelligence
TIER 5
Jul 17, 2025
Endorsing Martin Wolf's claim that The Magic Mountain is the 20th century's most revealing novel, DeLong reads Mann's sanatorium debates (Settembrini's feeble liberal humanism vs Naphta's authoritarian radicalism) as a mirror of today's crisis of melioristic liberalism. He extends this into a substantive intellectual history of the Belle Époque (1849–1914), its pseudo-classical semi-liberalism, and the Hayek-vs-Polanyi tension ('the market giveth' vs 'the market was made for man') central to Slouching Towards Utopia. A rich literary-historical essay with lasting thematic reference value.
Thomas MannliberalismBelle ÉpoqueMartin Wolfintellectual history
TIER 5
Jul 30, 2025
Drawing on unused Slouching Towards Utopia notes, DeLong uses Lawrence Dennis—America's foremost homegrown fascist theorist—to dramatize the 1930s temptation to answer economic crisis with an authoritarian engineer-dictator, and shows Dennis was right for Germany and France (the failures of Hilferding's SPD and Blum's Popular Front) but wrong for the US. He counterfactuals an FDR assassination, then ties the contingent survival of liberalism to Keynes's plan for full employment as the alternative to fascism and communism. A landmark essay weaving intellectual history, counterfactual, and the New Deal's contingency.
fascismGreat DepressionLawrence DennisNew DealKeynes
TIER 4
Aug 25, 2025
A detailed Chatham-House-style writeup of Dan Wang's Breakneck, framing China as the "engineering state" (sledgehammer) and America as the "lawyerly society" (gavel), and arguing the 21st-century task is to synthesize their strengths rather than choose sides. DeLong adds his own development-theory and political-economy gloss (early stages favor building, later stages favor allocation; path-dependence keeps China's bulldozers running) plus an epistemic-humility coda. A useful structured explainer of an influential book, though largely a synthesis of someone else's argument.
ChinaUS-China rivalryDan Wangpolitical economydevelopment
TIER 4
Aug 27, 2025
DeLong's Project Syndicate review of Dan Wang's Breakneck, which frames China as the country of the 'sledgehammer' (a technocratic engineering state) and America as the country of the 'gavel' (a litigious rights-and-vetoes society). The thesis is that despite surface differences both peoples are alike (restless, materialist, ambitious) and that the urgent 21st-century task is to synthesize the best of each while avoiding the worst. A clear, substantive book review of an influential China book, though largely overlapping with the fuller Director's-Cut version in issue 0378.
dan-wangchinaengineering-statebook-reviewpolitical-economy
TIER 4
Aug 31, 2025
The fuller Director's-Cut version of DeLong's Breakneck review, developing the sledgehammer (China's engineering state) vs. gavel (America's lawyerly veto-ocracy) dichotomy and quantifying the prescription: roughly 20% more building spirit for the US, 40% more respect for rights and process for China. It deepens the argument that the two dynamos are fundamentally alike and that the true 'City on a Hill' / 'Central Country' of 2100 may be a synthesized trans-Pacific place. The most complete statement of DeLong's engagement with Wang's framework in this batch.
dan-wangchinaengineering-statebook-reviewpolitical-economy
TIER 4
Sep 27, 2025
DeLong reprints Sutton's 2019 essay 'The Bitter Lesson' in full, with a framing note arguing that GPT-era models have vindicated it even more strongly in 'search'-and-'learning' domains where big-data, high-dimensional, flexible-function methods dominate human priors. His added value is the question of where that problem-frontier ends and whether a 'new Bitter Lesson' awaits beyond LLMs' limits. Worth reading mainly for the canonical Sutton text plus DeLong's frontier framing.
AIBitter LessonSuttonmachine learningscaling
TIER 4
Oct 14, 2025
DeLong restates his 'appeasement with teeth' proposal for Ukraine (recognize Russian conquests, EU accession for Ukraine, reparations funded by a tax on Russian energy, NATO 'trainers') and argues Putin has permanently turned Ukrainians into 'effective Poles' who can no longer be integrated. The core of the piece is an extended military-history essay on why modern firepower makes frontal assault and maneuver obsolete, producing attritional quagmire. Substantive blend of strategy, military history, and policy.
UkraineRussiaPutinmilitary historyforeign policy
TIER 4
Dec 29, 2025
DeLong distills Kotkin's structural account of Stalin—WWI normalizing mass violence, Leninist ruthlessness, patronage-built personal dictatorship, and the NEP contradiction that made collectivization-terror the 'only' communist instrument—while flagging Kotkin's hardest unsolved problem: why Stalin then 'crashed the plane' by purging loyal elites, which political explanation alone can't cover. He pairs this with a substantive methodological argument that an LLM at one's elbow now makes 'deep active reading' tractable by reconstructing an author's discursive milieu, overturning Plato's Phaedrus complaint that texts can't answer back. Paywall-truncated before the detailed notes.
historystalinismkotkinsoviet-historyai-assisted-reading
TIER 4
Jan 23, 2026
A crosspost of James Marriott's long essay (with DeLong's framing) arguing that the 18th-century 'reading revolution' forged Enlightenment rationality, science, and democracy, and that the smartphone-driven collapse of reading since the mid-2010s is reversing it toward a pre-literate, emotional, oral mode of thought. The stakes: literate cognition underwrites the entire intellectual infrastructure of modernity, so its decline threatens science, creativity, and liberal democracy itself. Substantial and provocative intellectual history, though it is a forwarded essay rather than original DeLong work.
literacy declineprint culturesmartphonesdemocracy and mediaintellectual history
TIER 4
Jan 23, 2026
DeLong evaluates Apple not as a stock but as a social technology for 'making computing humane,' crediting two hyper-excellent achievements (Apple Silicon and the China-centered supply chain) against four slow-motion failures: supply-chain fragility, AI/Siri strategic blindness, software quality-and-design drift, and a monopsonist developer-squeezing model. The useful frame is that high profits and stock price mask 'termites in the walls'—accumulating technical and trust debt as the firm shifts from 'what problem are we solving?' to 'what can we demo at WWDC?'
appletechnology-strategyaisupply-chaincorporate-analysis
TIER 5
Jan 26, 2026
DeLong dismantles Niall Ferguson's claim that Trump 'dominated' Davos, using the etymology of 'dominate' (the dominus at home in his own house) and Henry Farrell's ritual-as-common-knowledge account to argue Trump's Greenland bluster and climbdown signaled fragility, not mastery, while skewering the 'Xanatos Gambit' retconning of defeat into 4D chess. He also corrects Ferguson's misreading of the Melian Dialogue—Athens's 'realism' lost the war by provoking a balancing coalition—and closes with a meditation, via Wellington's Waterloo dispatch, on how 'public meaning' is constructed and unmoored from any true history. A rich, multi-layered essay weaving classics, IR theory, and historiography.
thucydidestrump-davospublic-meaningniall-fergusonhistoriography
TIER 4
Mar 1, 2026
DeLong analyzes the US-Israeli decapitation campaign and killing of Khamenei as a high-stakes gamble with no articulated endgame, arguing the structural odds favor an even harder-line 'IRGCistan' junta over a hoped-for liberal transition. The deeper thesis: the strike teaches every medium power that only an operational deterrent (nuclear or leader-targeting) deters regime change, corrodes nonproliferation norms, threatens Hormuz oil flows, drains US bandwidth versus Russia and China, and may normalize assassination-as-deterrence (he quotes More's Utopia). A substantive, framework-driven take on a wobbling world order with fat-tailed risks.
Iran warnuclear proliferationStrait of Hormuzregime changeworld order
TIER 4
Mar 30, 2026
Built around an extended quotation of Bret Devereaux's strategic analysis, DeLong frames the US-Iran war as a classic escalation trap centered on the Strait of Hormuz, where neither side can back down without political ruin so both keep losing. He adds notes on an erratic, manipulable president without adult supervision, the unknown real state of US forces, and the vulnerability of carriers in the drone-and-anti-ship-missile era. It matters as a structured strategic reckoning showing how a regime-collapse gamble produced a worse position than the JCPOA it replaced.
iran-warstrait-of-hormuzescalation-trapus-strategydrone-warfare
TIER 4
Apr 9, 2026
Reading Grant's Appomattox chapter, DeLong meditates on Grant's distinction between the valor of soldiers and the injustice of their cause, drawing in the Iliad and Lincoln's Second Inaugural to argue that sincerity and sacrifice never redeem an unjust structure and that defeating an unjust cause does not build a just order. He notes the staggering cost of the Civil War (700,000 dead) could have bought emancipation plus '40 acres and a mule' many times over. A reflective historical-literary essay with durable bullet-point theses about war, memory, and Reconstruction.
US historyCivil WarU.S. GrantLincolnReconstruction
TIER 4
Apr 16, 2026
DeLong argues the US has suffered a self-inflicted 'strategic defeat' in its war with Iran: having already paid the costs of the strikes, Iran can now credibly tax Strait of Hormuz transit, and a Ferguson-Haass-Zelikow SOHCO peace proposal effectively concedes this while dressing it up as denying Iran a win. He grounds the analysis in the political economy of strait-tolling (Bosphorus, Panama, Suez) and a Thucydidean critique of dominance-celebrating 'realism.' A substantive geopolitical-economy piece with a genuine analytic spine and original framing of the tollkeeper problem.
foreign policyIran warStrait of Hormuzgeopolitical economyThucydides
TIER 4
May 19, 2026
Crossposting Noah Smith (with DeLong's long-lens framing from Bussaco 1810 to today), the argument is that cheap, AI-guided FPV drones now dominate the battlefield on cost and transparency, rendering carrier/tank/missile-centric militaries obsolete and exposing US Gulf bases and NATO doctrine as unready. China's monopoly on batteries, rare-earth motors, and DJI-class autonomy could let it outbuild everyone in a drone armada. DeLong adds open questions about doctrine when both sides have drones. A strong, consequential piece on the revolution in military affairs and its industrial-policy stakes.
drone warfaremilitary revolutionUkraine warChina industrial baseelectric tech stack
TIER 4
May 30, 2026
DeLong synthesizes reporting on a fictional/alternate-2026 US-Iran war pause, framing it as Trump paying Iran reparations ($6B now, possibly a $300B 'reconstruction fund') to reopen the Strait of Hormuz while both sides claim victory. The analytical payoff is an extended argument about adverse exchange ratios in drone-era warfare ($4M Patriots downing $35K Shaheds) and a sustained Prussia-1806/Jena-Auerstadt analogy for a US military institutionally lagging the latest revolution in military affairs. It matters as a political-economy-of-war piece linking industrial base, attrition math, and great-power decline.
Iran warStrait of Hormuzdrone warfarerevolution in military affairsPrussia 1806 analogy
Macroeconomics, the Fed & Monetary-Fiscal Policy
3 tier-5 · 16 tier-4
The macro cluster defends a Keynesian stabilization framework and the institutions that carry it. DeLong mounts a full defense of the 2020s Fed and Bidenomics (transitory inflation, a successful high-pressure recovery punished by a misinformation machine), dissects the post-COVID inflation episode as a failure of the 1970s Phillips curve, and returns repeatedly to Fed independence under Trump pressure - Powell, the Lisa Cook firing, the parade of sub-mediocre chair candidates, and the politics of why voters hate moderate inflation more than recessions. Deep financial-history pieces (the 1825 Panic and the birth of central banking, the 1907 Panic, the GENIUS Act as a Free-Banking "wildcat" revival, Bagehot's lender-of-last-resort doctrine) ground the present in 150 years of monetary thought.
TIER 4
Mar 25, 2025
DeLong's Godley-Tobin Lecture writeup mounts a full-throated defense of Federal Reserve monetary policy in the 2020s, arguing the 'late and fast' response and the 2021 American Rescue Plan were both defensible ex ante even though post-COVID inflation ran longer and stronger than expected. He grounds it in Keynesian first principles of stabilization (macro dysfunction as excess demand for 'money'/safe assets, the failure of Say's Law, the role of anchored expectations), credits Powell's Fed, and reframes the inflation episode as ultimately a political rather than economic problem. The 16-point takeaway list makes it a useful reference distillation of his stabilization-policy framework, though it is a draft lecture note pointing to video and slides rather than an edited essay.
monetary policyFederal ReserveKeynesianismpost-COVID inflationstabilization policy
TIER 4
Apr 23, 2025
DeLong analyzes the Fed's bind—weakening business animal spirits arguing for cuts while tariff-driven inflation argues for hikes—against Trump's threats to fire Powell and possibly install the hawkish Kevin Warsh. He frames the moment via a four-shocks model of the 1980 inflation spiral (now three shocks: COVID reopening, Ukraine, tariffs, with Fed-independence pressure as a possible fourth), judging the U.S. closer to the cliff than in 2020 but not yet over it. A useful monetary-policy explainer linking Fed independence, inflation expectations, and historical analogy.
Federal Reservemonetary policyPowellinflationFed independence
TIER 4
May 6, 2025
A methodological essay on why macro forecasting is failing: models work only by fitting historical correlations, so structural breaks (COVID, Putin, a collapsing Beveridge Curve) and shocks outside sample render point forecasts hazardous. DeLong confesses his own Team Transitory error and Summers's Beveridge-Curve error as cases in point, concluding that honest analysts now have 'flat posteriors and long tails' and should forecast direction, not magnitude. A thoughtful explainer on forecasting epistemology.
macro forecastingBeveridge curvemodel uncertaintystructural breaksinflation/recession
TIER 4
May 24, 2025
Riffing off Krugman's quip that at any moment either he or Summers is right but you can't tell which, DeLong argues both Team Transitory and Team Persistent were blindsided because the standard Phillips-curve model derived from the 1970s was a bad fit even for the 1970s. He concedes that Team Transitory was 'wrong for the right reasons' is too comfortable a self-exoneration—there is a natural rate of inflation that rises with the pace of demanded structural change—and counsels methodological humility. A substantive macro/monetary-policy retrospective on the 2020s inflation episode.
inflationPhillips curvemonetary policyKrugman/Summersmacro forecasting
TIER 5
May 29, 2025
DeLong rebuts Kevin Warsh's claim that the Fed deserves independence for monetary policy but not bank regulation, marshaling the history of the 1907 Panic and J.P. Morgan, the creation of the Fed in 1913, and the First/Second Bank of the United States to show that lender-of-last-resort capacity is inseparable from prior bank supervision. The Greenspan-to-Newman crux, you cannot be an effective lender of last resort without already regulating and understanding the banks, anchors the argument. A landmark-quality reference essay fusing monetary theory with deep US financial-institutional history.
Federal Reservebank regulationcentral bank independencePanic of 1907financial history
TIER 4
May 31, 2025
Building on Alan Taylor's FT interview with Martin Wolf, DeLong sketches a macro roadmap for the post-neoliberal era organized around five threads: anchored inflation expectations as the Volcker legacy, low inflation as a precondition for legitimate politics, the end of the Great Moderation into an age of recurring shocks, a persistently low r* that makes public and private investment cheap, and the slow grinding cost of 'TRUMPXIT' protectionism. A strong synthesizing macro essay tying current policy to economic-history lessons.
macroeconomicsinflation expectationsr-starAlan Taylordeglobalization
TIER 4
Jun 9, 2025
Endorsing Brad Setser, DeLong reframes the dollar's status: since 2014 US current-account deficits have been financed not by official reserve accumulation ('exorbitant privilege') but by private investors chasing returns and risk-insurance, making dollar dominance a daily wager on American institutions rather than a structural birthright. The key implication is that domestic political/institutional rot (TRUMPXIT, judicial delegitimization) threatens a private-sector 'sudden stop' more than any rival currency. A full essay with a references list and real analytical content.
dollarexorbitant privilegecapital flowscurrent accountBrad Setser
TIER 4
Jun 19, 2025
A substantive macro essay arguing that the 1990s Clinton-Greenspan notion of a 'normal' macroeconomy no longer holds: the market now prices a lower neutral real rate (~2.5%) and an evenly split, confused FOMC must be modeled as a non-rational actor. DeLong walks through the secular-stagnation vs global-savings-glut regimes and what current discount rates and the term structure reveal about long-run expectations. Useful explainer on intertemporal discounting and Fed policy under tariff uncertainty.
monetary policyFederal Reserveinterest ratessecular stagnationmacroeconomics
TIER 4
Jun 21, 2025
DeLong argues the Genius Act's licensing of private stablecoin currencies recreates the chaos of the pre-Civil-War Free Banking 'wildcat' era, threatening the 'singleness of money' public good and risking a self-reinforcing Treasury-fire-sale run akin to 2008 or the SVB collapse, all without living-will, stress-test, or resolution regimes. His counter-proposal: cap interchange fees as the EU did in 2015, capturing the payment-cost savings without the systemic risk. A timely monetary-policy and financial-stability analysis grounded in economic history.
stablecoinsmonetary policyfinancial regulationfree bankingGenius Act
TIER 4
Jun 21, 2025
DeLong endorses Barry Eichengreen's warning that the GENIUS Act's authorization of private company-issued stablecoins recreates the chaotic Free Banking era of worthless private notes, with regulators unable to supervise thousands of issuers. His constructive counter: cap interchange fees as the EU does to get the consumer benefits without the safety-and-soundness risk. A clear monetary-history-grounded policy take, though the deeper analysis sits behind the paywall.
monetary policystablecoinsGENIUS Actfinancial regulationfree banking
TIER 4
Jul 14, 2025
DeLong dissects why the FOMC is paralyzed: tariff chaos, a do-nothing tax bill, and likely collapses in services exports and immigration are disinflationary disinvestment shocks, yet fear of de-anchoring expectations plus Powell's consensus-manager style keep the Fed from cutting. The sharper point is that Trump sabotaged his own goal of lower rates through tariff turmoil and by appointing a consensus-builder rather than a rate dove, evidence of his incompetence at personnel selection. A solid, original macro-and-institutions read on Fed policy.
Federal ReserveFOMCtariffsinflation expectationsmacro outlook
TIER 4
Sep 18, 2025
DeLong develops the asymmetry that voters hate moderate inflation (felt as a broken social contract, near zero-sum redistribution) far more than recessions (lose-lose but with concentrated, less visible, deferred costs), which biases policy toward performative 'austere' toughness and penalizes risk-optimal demand management—as the well-managed 2020-22 US recovery was punished politically. He then reads the September 2025 rate cut, an asymmetric easing into rising inflation forecasts, as marking the effective end of Fed independence under Trump pressure, a smaller echo of Arthur Burns 1972. A substantive macro-political-economy analysis.
inflationFederal Reservemonetary policyFed independencepolitical economy
TIER 4
Oct 10, 2025
DeLong uses the U.S. Treasury's $20B swap-line "bailout" of Milei's Argentina to give a clean tutorial on fiscal dominance—when deficits overwhelm monetary policy, pegs and lifelines only buy time and de-risk investors' exits without fixing the anchor. He lays out the known cure (credible primary surplus, then a sharp devaluation to an undervalued peg, then prayer for a fast export boom, on the 1920s Poincare model) and concludes that, absent an enacted program, the support is best explained as corruption or letting financier friends exit. A compact, useful macro explainer.
Argentinafiscal dominanceexchange ratesMileimacroeconomics
TIER 4
Oct 30, 2025
Answering Dan Davies's puzzle that the physically harmless 2008 financial crisis scarred economies for a decade while the deadly COVID shock reversed in two years, DeLong argues the difference was cash, certainty, and a lender of last resort—Bagehot's 1873 Lombard Street playbook (lend freely, at a penalty rate, on good collateral) plus Keynes's 1936 fiscal addendum and Kohn's 'go fast, go hard, go soon, fix moral hazard later.' The deeper question is why the economics profession forgot 150-year-old wisdom, which he traces via Krugman to the intellectual 'Dark Age of macroeconomics.' A strong essay on financial-crisis doctrine and the history of macro thought.
financial crisesBagehotlender of last resortmacroeconomicshistory of economic thought
TIER 5
Nov 6, 2025
DeLong's full balance sheet on Bidenomics: the Biden-Powell COVID-recovery policy mix achieved rapid return to full employment and a healthy structural reallocation of labor, with inflation that was genuinely transitory (two moderate bursts from reopening and Putin's invasion, then ~2.8% from mid-2022)—a sharp contrast to the hysteresis-laden, secular-stagnation Obama-Bernanke recovery. He concedes flaws (failure to vaccinate the world; under-scaled reindustrialization) but argues the 2024 electoral repudiation reflects a global misinformation machine, not policy failure, citing Reuters/Ipsos data that voters who knew basic economic facts overwhelmingly backed Harris. A definitive, reference-grade statement of his macro-policy assessment.
macroeconomicsBidenomicsinflationfiscal policymisinformation
TIER 4
Dec 3, 2025
Drawing on Atalay, Hortaçsu, Kimmel & Syverson, DeLong argues conventional producer-facing deflators understate U.S. manufacturing TFP growth by ~1.7 points in durables (5.7 points in computers/electronics) because they miss quality improvements; a system-wide input-output revaluation pushes sector TFP to ~141 vs the official 117 by 2023. The 'stagnation' narrative is largely mismeasurement, though a real post-GFC slowdown remains, which he flags as a research puzzle. A substantive measurement-and-method explainer with policy implications.
manufacturingtotal factor productivityhedonic deflatorsinput-output tablescomputers and electronics
TIER 4
Dec 3, 2025
Summarizing Atalay-Hortaçsu-Kimmel-Syverson, DeLong argues that conventional producer-facing deflators understate U.S. manufacturing TFP growth—by 1.7 points in durables—because they miss quality improvements, especially in computers and electronics, so claims of absolute manufacturing "stagnation" are largely mismeasurement. Corrected, the post-1990 productivity story brightens but reveals a real post-GFC slowdown and a widening gap between high-tech and the rest of manufacturing. A substantive explainer of a measurement debate with policy salience.
manufacturingproductivityTFP measurementdeflatorsinput-output
TIER 5
Jan 1, 2026
Drawing on his own 1993-94 Treasury experience, DeLong argues Clinton's OBRA 93 deficit reduction crowded in investment and added ~0.5pp/year to growth (leaving America ~15% richer), and that the predicted recession never came because a supportive Greenspan Fed plus falling ICT prices offset the fiscal drag. He insists the famous 1994 bond selloff reflected surprise economic strength and MBS duration mechanics, not fears of self-defeating austerity, and that Republican-economist 'professional concern' was partisan theater (OBRA 90 had no such debate). A first-hand, model-grounded reference on the fiscal-monetary policy mix and crowding-in.
macroeconomicsfiscal-policyclinton-obra93crowding-indeficit-reduction
TIER 4
Feb 8, 2026
DeLong argues the streaming wars were predictable both ex ante and ex post: legacy studios mistook a content business for a software platform, burned through cheap zero-rate capital chasing scale, and ended up crawling back to license content to the aggregator (Netflix) they tried to escape. The deeper twist is that Netflix, having seemingly 'won' via first-mover advantage and a pending Warner Bros. deal, now faces YouTube as the true apex predator of attention and Sora-style AI slop on the horizon. It matters as a clean industrial-organization explainer of platform economics, capital cycles, and scale dynamics in media.
industrial organizationstreamingplatform economicsNetflixattention economy
Trumpxit - The Trade War & American Relative Decline
2 tier-5 · 25 tier-4
DeLong's most developed contemporary argument is "Trumpxit": the claim that economic power runs on *trust*, and that even with zero tariffs, the destruction of US policy reliability inflicts a Brexit-magnitude loss (10%+ of output, ~1-1.5 points of growth per year over a decade) as the world rationally de-risks. The tariffs themselves he reads as theater - a mathematically incoherent formula, "deals" with no enforcement, performative grievance rather than strategy - whose real damage is institutional. Recurring analytical tools: value-chain decomposition (why tariffs hit far more than the import share), Lerner symmetry, "dark matter" and the investment-surplus reframing of the trade deficit, network power, and the explicit historical rhyme with Britain's 1870-1913 slide from hyperpower to also-ran.
TIER 4
Apr 3, 2025
DeLong dissects the 'Liberation Day' reciprocal-tariff formula—trade deficit divided by exports, halved—showing it matches the nonsense answer LLMs give when asked to 'fix' trade deficits, and that the policy is mathematically incoherent and diplomatically corrosive. He argues you cannot make a deal with Trump (commitments won't be honored, the stock-market 'veto' was a lie), forecasts an 80% chance of a 2025 stagflationary business-cycle peak, and analyzes the trade/capital/cost channels. A pointed, well-sourced contemporary economic analysis.
tariffstrade policyTrumpstagflationAI
TIER 4
Apr 8, 2025
DeLong's most developed statement of the 'Trumpxit' thesis: even if Trump imposed no tariffs at all, the destruction of trust in US policy stability would cost America BREXIT-magnitude losses (10%+ of output) as the world de-risks. He draws the explicit parallel to Britain's 1870-1913 slide from hyperpower to industrial also-ran and revises his Slouching Towards Utopia framing—America's relative economic decline, not decarbonization, may now be the central story of the century. A strong, conceptually clear political-economy essay.
Trumpxittrust/de-riskingAmerican declineBREXIT analogySlouching Towards Utopia
TIER 4
Apr 8, 2025
DeLong frames Trump's trade war as 'Trumpxit'—an American analogue to Brexit—arguing that economic power runs on trust and that the damage from burning US reliability is largely irreversible even if every tariff were rescinded tomorrow. Drawing on Winkler's data showing Brexit left the UK an 'also-ran' behind the eurozone, he warns the US has converted itself from indispensable to merely unpredictable. A sharp conceptual frame, though the body is truncated at the paywall.
trade policyBrexittrusthegemonyTrump
TIER 4
Apr 13, 2025
DeLong lays out his own three rationales for a strong industrial sector (equality/mobility, working-class self-organization, technological externalities), argues the first two are dead and only the externalities case survives—and that even it argues for Pigovian engineer subsidies plus a cybernetic case for undervaluation, not Trump's tariffs. He then reprints the Summers-vs-Cass Fareed Zakaria debate where Summers dismantles Cass's defense. A meaty framework on industrial policy paired with a high-signal transcript.
industrial policytariffsmanufacturingOren CassLarry Summers
TIER 4
Apr 25, 2025
DeLong delivers a point-by-point rebuttal of John Authers' Bloomberg column that treats Stephen Miran's 'User's Guide to Restructuring the Global Trading System' as Trump's abandoned plan, insisting there was never a plan—only grievances—and that Miran's assumptions (currency offset, non-inflationary tariffs, inelastic Treasury demand, winning a game of chicken with China) were always unhinged, not merely 'tenuous.' Along the way he lays out a sharp argument that the U.S., not China, is the bigger loser in a trade war because Chinese exports feed ~$3T of U.S. service-sector GDP that can't be quickly replaced. A substantive critique of media 'sanewashing' plus real trade-economics content.
trade warStephen Miransanewashingmedia criticismChina
TIER 4
Apr 26, 2025
A cross-post of Noah Smith dismantling Oren Cass's defense of Trump's tariffs as a path to reindustrialization, marshaling real-time evidence (plunging Philly/NY manufacturing surveys, layoffs at Volvo/GM/Cleveland-Cliffs, collapsing capex and manufacturing stocks) that tariffs are accelerating deindustrialization. Smith refutes Cass's claim that imported-component costs only matter for exports, invokes scale effects and the Costinot-Werning result that tariffs may not even shrink deficits, and frames the 'pundit's dilemma' of pro-Trump commentators forced to choose influence over honesty. A substantive trade-and-manufacturing analysis with a useful evidentiary roundup.
tariffsmanufacturingOren CassNoah Smithindustrial policy
TIER 4
Apr 29, 2025
DeLong argues Trump's strategy-less trade war will empty U.S. ports (a predicted 35% drop at LA/Long Beach) and that the damage far exceeds the import share of GDP because value chains layer most of a good's worth—logistics, design, fitting, use-value—on top of the physical import. Using a Nike-shoe decomposition ($100 exchange value, $200 total use value, only $20 imported), he shows that when goods stop arriving, the whole downstream value evaporates. A vivid value-chain explainer of why tariffs hit far more than the 14% import share, though truncated by a paywall.
trade wartariffsvalue chainsGDPChina
TIER 4
May 11, 2025
Reading the Trump-Starmer 'deal' as performance art, DeLong forecasts the modal outcome as a 10% universal US tariff, declining trade volumes, and a declared 'victory'—a BREXIT-magnitude self-inflicted wound leaving the US ~10% poorer—while stressing the fat tails toward autarky and stagflation. He marshals Politano, Sandbu (Lerner symmetry), Wolf, and the Smoot-Hawley parallel to argue tariffs tax exports too and that uncertainty itself is the core cost. A well-sourced trade-policy analysis.
tariffstrade policyLerner symmetryTrump-Starmer dealBREXIT analogy
TIER 4
Jun 4, 2025
Riffing on Eichengreen's sterling-vs-dollar column, DeLong argues the US current-account 'trade deficit' should be reframed as a capital-account 'investment surplus' that is its arithmetic mirror, and that the real policy question is whether the negative externality (erosion of engineering communities of practice) outweighs the positive externalities of capital inflows. He insists the dollar's reserve role is no 'resource curse' and that blue-collar job loss was driven by deunionization and automation, not trade. A useful reframing essay with a deep appendix on Churchill's 1925 gold-standard decision.
trade deficitreserve currencyEichengreenexternalitiesindustrial policy
TIER 4
Jun 18, 2025
Drawing on Adam Posen's Senate testimony, DeLong argues Trump's tariffs are an exceptionally inefficient, regressive tax (~$2,600/year on the middle quintile) whose worst damage is institutional—destroying US trade reliability and replacing the G7 with a US-excluding G6. He revises his TRUMPXIT-vs-BREXIT analogy upward to a >1%-point-of-GDP-per-year growth headwind over the decade. Self-contained, well-referenced trade-policy analysis.
trade policytariffsAdam PosenTrumpinternational economics
TIER 5
Jul 5, 2025
DeLong builds on Hausmann and Sturzenegger's 'dark matter' argument—that US intangible assets (ideas, technology, returns on knowledge investments abroad) generate hidden services surpluses that offset the recorded trade deficit—to argue that America's true wealth flows from its hub role in the rules-based, open globalized value-chain system. The key insight is that tariffs need not be high, only uncertain: weaponized interdependence drives firms and talent to derisk away from the US, and he invokes the post-1890 collapse of Bismarck's order as a warning that such networks are far easier to destroy than rebuild. A landmark synthesis tying his long-twentieth-century framework to current trade policy.
American exceptionalismdark matter tradeHausmannglobalizationrules-based order
TIER 4
Jul 13, 2025
DeLong argues that no real 'trade deal'—a negotiated, enforceable, codified agreement—can emerge from an administration that cannot credibly commit, so the announced 'frameworks' are Potemkin facades while tariff math is generated by random-number generators. Because US commitments are now structurally unreliable, trading partners gain nothing from appeasement and rationally accelerate decoupling, risking institutional erosion he likens to a Brexit-magnitude (10%+ GDP) self-inflicted wound. A substantive trade-policy explainer drawing on his Treasury-modeling experience.
trade wartariffscredible commitmentdecouplingglobalization
TIER 5
Jul 29, 2025
DeLong argues US hegemony across all four dimensions (geostrategic, political, economic, cultural) is now closed—squandered by Trump-era incompetence, science cuts, anti-immigration enserfment, and broken alliances—while building out a careful conceptual taxonomy of hegemony drawing on Kindleberger, Keohane, Krasner, and Strange. Crucially he contends China is unlikely to step through the door, leaving not Chinese hegemony but a balance-of-power world, because hegemony is a hard-won 'practice' requiring an elite committed to the project that China lacks. Lasting reference value for its framework on hegemonic stability theory applied to the present.
US hegemonygeopoliticsChinaKindlebergerinternational order
TIER 4
Aug 21, 2025
DeLong argues that claims of Trump tariff "victory" are hollow: the effective US tariff rate has jumped from 2.3% to ~16-18.6% (interwar/Smoot-Hawley levels), and the costs fall overwhelmingly on Americans while amplified value-chain drag dwarfs any fiscal revenue. He stresses that trading partners' non-retaliation is a strategic decision to "wait Trump out," not legitimacy, and that the rest of the world is large enough to reap value-chain gains without the US. A solid trade-policy analysis grounded in his Treasury modeling experience, but heavy on curated quotes.
trade wartariffsTrumpmacroeconomicsglobalization
TIER 4
Oct 5, 2025
DeLong argues that Trump's tariffs and erratic alliance signaling depreciate U.S. "network power"—the multiplicative trust across trade, finance, tech, and security that lets allied scale offset China's larger war-economy throughput—as allies rationally hedge into mini-laterals and non-U.S. standards. He puts numbers on it: "TRUMPXIT" as a ~1.5-point/year drag on real growth via lost TFP, alliance-scale spillovers, and capital misallocation, leaving median households 12–18% below counterfactual within a decade. A substantive, quantified geoeconomics argument with reference value.
geoeconomicstariffsUS alliancesChinaTrump
TIER 4
Oct 5, 2025
Argues that erratic Trump tariffs and oscillating alliance commitments tax US 'network power'--whose value is multiplicative across trade, finance, tech, and security--so allies rationally hedge by building US-excluding mini-laterals, and each shock raises the discount rate they apply to American promises. A crisp companion to the Luce piece making the case that the American Century is now ending, though the bulk sits behind the paywall.
network powertariffsUS alliancesTrump foreign policygeopolitics
TIER 4
Oct 5, 2025
A close reading of Henry Luce's 1941 'American Century' manifesto plus the full reprinted text, arguing Luce's instinct (that US domestic prosperity depends on shaping the global rules-environment) was faithfully implemented by Bretton Woods, the Marshall Plan, NATO, and GATT. Valuable as primary-source-plus-commentary on the intellectual origins of the postwar liberal order, and pairs directly with issue 0343 on its unmaking.
Henry LuceAmerican Centuryliberal orderBretton Woodsprimary source
TIER 4
Oct 26, 2025
Drawing on Richard Baldwin, DeLong argues Trump's trade war runs on emotional optics not economics—loud tariffs with quiet bureaucratic unwinding—producing short-run stability as presidents claim symbolic wins while Commerce/Treasury technocrats restore trade flows via exemptions. The longer-run cost is a BREXIT-like productivity drag on the US while China inherits the role of global-trade rule-writer and surplus-hogger. A clear political-economy model of why the trade war is both resilient and quietly damaging, though paywalled before the medium/long-run section.
trade wartariffsTrumpglobalizationChina
TIER 4
Nov 19, 2025
Using the US-Malaysia framework (via Alan Beattie), DeLong argues Trump's "trade deals" are not binding agreements but stage directions for an impulse-driven executive—no dispute settlement, sweeping national-security escape hatches, and a Malaysian "consultation" clause that buys only advance warning. The strategic upshot is that partners trade PR wins for forewarning while derisking and decoupling from an unreliable US, a marginal-to-substantial long-run American loss. A sharp political-economy read of trade-policy chaos (full version of the post duplicated at 0272).
trade policyTrump tariffsMalaysiaderiskingpolitical economy
TIER 4
Nov 22, 2025
DeLong skewers the Trumpist incoherence of claiming tariffs don't raise prices while demanding rate cuts because removing tariffs would lower prices, framing it via the Gish Gallop and Orwellian doublethink as power-demonstration rather than argument. Citing the Tax Foundation's estimate of ~700,000 lost jobs and a record 12.5% effective tariff rate, he argues chaotic, constantly-moving tariffs do roughly five times more damage than static GE models predict because firms cannot adjust and must buy insurance against random future moves. A pointed macro-and-political-economy critique with a real modeling argument.
tariffsinflationFedGish Galloptrade policy
TIER 4
Dec 8, 2025
A cross-post of Henry Farrell reading Trump's new National Security Strategy, which casts the EU as a civilizational and security threat and amounts to a program for illiberal regime change in Europe in alliance with its far right. Farrell argues it will fail on its own terms—a hollowed-out NSC and State Department can't implement it, it galvanizes European resistance, and it signals inward-focused weakness to adversaries. A substantive political-economy and grand-strategy analysis of US foreign policy.
National Security StrategyUS-Europe relationsTrump foreign policygrand strategycross-post
TIER 4
Dec 19, 2025
Building on Richard Baldwin's 'omelette' table of cross-country intermediate-input dependence, DeLong argues that macro-level near-self-sufficiency masks micro choke points, and that agglomeration economics—scale and local spillovers in intermediate production—make manufacturing clusters self-reinforcing and nearly impossible to relocate via tariffs. The asymmetry favors China (the dominant intermediate supplier with low exposure), so weaponizing trade erodes US coalitions faster than it hurts China; he reads the race to be 'the furnace where the future is forged' as essentially already decided by 2028.
economic historyglobalizationsupply chainsChinatrade policy
TIER 4
Dec 19, 2025
A second (truncated-preview) send of the Factory Asia / agglomeration essay, identical in substance to issue 0229. Building on Richard Baldwin's omelette table, it argues micro choke points hide under macro self-sufficiency, agglomeration makes clusters self-reinforcing, and China's hub position gives it asymmetric weaponized-interdependence leverage that US tariff escalation only worsens.
economic historyglobalizationsupply chainsChinatrade policy
TIER 4
Jan 15, 2026
A crosspost of PIIE's Alan Wolff arguing that Trump's April 2 across-the-board IEEPA tariffs had no legal basis: IEEPA never mentions tariffs, was never referred to tariff committees, and rests on a botched folk memory of Nixon's 1971 surcharge (which actually used other authority). DeLong frames the failure as institutional cowardice—a Republican Congress unwilling to defend the separation of powers and a Supreme Court slow-walking the case—reading it through Marvell's Horatian Ode. Useful as a clear legal-history explainer of how Congress's constitutional tariff power was surrendered.
tariffsIEEPAseparation-of-powerstrade-policyrule-of-law
TIER 4
Jan 21, 2026
A crosspost of the full text of Canadian PM Mark Carney's Davos speech, framed by DeLong via Havel's 'greengrocer' and Thucydides: the rules-based international order was a useful fiction that the US (under Trump) has stopped performing, and middle powers like Canada must 'take the sign out of the window,' name reality, and combine into issue-based coalitions rather than negotiate bilaterally from weakness. Matters as a primary-source statement of a middle-power grand strategy for a post-hegemonic order; mostly Carney's words with DeLong's interpretive frame.
international orderMark CarneyTrump foreign policymiddle powersgeopolitics
TIER 4
Mar 28, 2026
The full Richard Baldwin original (DeLong crossposted a digest of it in #0116), arguing that 2025's tariff assault failed to trigger retaliatory escalation because tariffs were 'theatre not policy' organized around grievance, blunted by exemptions, and contained at home by four TACOs. It details the China episode where Beijing established escalation dominance via US reliance on Chinese inputs, forcing US tariff cuts dressed as victory. A substantive, well-evidenced trade-policy explainer; value overlaps heavily with the crosspost digest.
trump-tariffsworld-trade-systemtacochina-trade-warescalation-dominance
TIER 4
Mar 28, 2026
Crossposting Baldwin's argument that Trumpian tariffs were performative grievance theater (the 'Grievance Doctrine' and four TACOs) that other nations defused by giving Trump optics while keeping the rules-based system, DeLong adds the crucial medium-run extension. He argues that because the US is now an unreliable chaos-monkey counterparty, the world is quietly decoupling, a slow cis-Atlantic BREXIT that could evict America from the center of the global economy. It matters for pairing a useful short-run explanatory template with a long-run structural warning.
trump-tariffsworld-trade-systemweaponized-interdependencedecouplingrichard-baldwin
Trumpism, the Courts & the Political Economy of Authoritarianism
0 tier-5 · 38 tier-4
The political economy of the second Trump term, read through institutions rather than personality. DeLong's recurring frames: "sanewashing" (the near-irresistible pundit temptation to retrofit grand strategy onto stochastic flailing), "patronage-autocracy" (favors flow only to supplicants who disclaim entitlement - the mechanism behind Musk's humiliation), and the weaponized shadow docket ("confiscate now, litigate later") through which a captured Supreme Court hands an executive unexplained partisan wins. Around these sit elite capitulation on Wall Street, the neofascist turn in immigration and on campus, the Hassett "Dow 36,000" pattern of lying-for-plutocrats, and a structural reading of the Roberts Court as a bloc rather than a 3-3-3 institution.
TIER 4
Apr 15, 2025
DeLong's polished Project Syndicate column arguing the second Trump administration has no policies or policymaking processes—only one ignorant man's instincts and the self-interest of surrounding sycophants. He contrasts Trump with Reagan (who had a governing philosophy and trusted professionals) and rebuts Larry Summers's 'advisors entitled to believe in his policies' as sanewashing since there are no policies to believe in. A sharp, self-contained statement of his 'court of the chaos-monkey king' thesis.
Trumppolitical economygovernanceUS politicsReagan
TIER 4
May 8, 2025
DeLong notes that Wall Street figures who expected to constrain Trump now refuse to criticize him on the record—Cembalest self-censors, Fink says 'let's move on,' Ackman grovels—and reads this fear as a working definition of neofascism. His framing: plutocrats wrongly assumed kleptocrats treat them as friends rather than prey, and Hayek was wrong that only central planning destroys regard for truth. A sharp political-economy observation on elite capitulation and authoritarian dynamics.
Wall StreetauthoritarianismTrumpHayek/Road to Serfdomelite capitulation
TIER 4
May 31, 2025
DeLong argues that banning AI in education is futile and that teaching should be redesigned around what students should remember and be able to do five years out, using the 'how would you discover / how would you persuade' scaffolding he sketches against his Econ 113 syllabus. He frames AI as the latest leaky abstraction layer (per Sinofsky and Spolsky), embracing its productivity while insisting students must learn enough of the lower layers to avoid model-worship and misplaced concreteness. A substantive pedagogy-and-abstraction-layers essay with lasting framing value.
AI in educationpedagogyabstraction layersleaky abstractionslearning outcomes
TIER 4
Jun 4, 2025
DeLong surveys the world's ~21 hundred-billionaires and argues that Musk's Tesla fortune is uniquely fragile: the 2021-23 Tesla boom rested on a green-modernity status symbol plus an accidental cozy chip-shortage cartel among rival automakers, not durable car economics. With Musk having destroyed Tesla's brand equity through political extremism, DeLong gives even odds Tesla stops making cars within five years and would not bet against a personal Musk debt workout within a decade. A sharp case study in how charisma, brand-as-psychic-good, and oligopolistic IO theory determine a single industrialist's wealth.
Elon MuskTeslaoligopolybrand valuebillionaires
TIER 4
Jun 5, 2025
DeLong lays out the implicit deal Musk thought he had with Trump (EV subsidies, tariff carve-outs, NASA money for SpaceX in exchange for donations, cheerleading, and taking the DOGE heat) and why it collapsed: Trump is only transactional when he must pay cash upfront, otherwise not at all. He reads Musk's current rage—threatening to whip a Purity-Republican Senate bloc against the tax bill—as a desperate attempt to prove he has veto-point power and stave off Tesla/SpaceX bankruptcy. A complete, original political-economy analysis with a clear mechanism.
Elon MuskTrumpSpaceXTeslapolitical economy
TIER 4
Jun 10, 2025
DeLong dissects Kevin Hassett's claims that 3% growth will erase the deficit and that DOGE rescissions will cut spending, demonstrating both are knowing lies. His counter-estimate: with Trump's anti-immigration war, tariff disruption (TRUMPXIT), and attacks on universities, US potential output growth has an upper bound near 0.2%/year. A substantive, numbers-backed macro takedown of administration economic spin and the press corps that launders it.
Kevin Hassettpotential growthimmigrationTRUMPXITfiscal policy
TIER 4
Jun 21, 2025
Prompted by a white-nationalist law-student paper, DeLong examines the originalist claim that 'We the People' in 1787 meant white people, taking the racism-of-the-Founders premise seriously (via Taney's Dred Scott reasoning and Franklin's 1751 screed against 'swarthy' Germans) while marshaling a state-by-state survey of free-Black voting in the ratification era as the empirical rebuttal. He sharply distinguishes the originalist standing argument from the 'just wrong' claim that the 14th Amendment is unconstitutional. A substantive legal/economic-history essay on slavery's racialization and constitutional original sin.
constitutional historyoriginalismslaveryDred Scottrace and law
TIER 4
Jun 23, 2025
A crosspost of Nils Gilman's essay (with DeLong's framing) mapping the real ideological fault line in elite academia as liberals vs. leftists (Marxists and epistemic radicals), not liberals vs. conservatives. Its sharp thesis: the anti-foundationalist 'epistemic radicals' won—but on the reactionary right, where Bannon's 'flood the zone' and 'alternative facts' weaponize the very anti-Enlightenment relativism Habermas warned would make space for the right. A genuinely useful intellectual-history argument despite being a crosspost.
intellectual historyacademiapostmodernismideologypolitical economy
TIER 4
Jun 26, 2025
DeLong cautions against left enthusiasm over Mamdani's NYC primary victory, arguing a primary win 'wins you nothing' and that what matters is general-election viability plus the ability to build governing coalitions—best secured via a real Mamdani-Lander co-mayorship. He frames this through Weber's 'politics is the strong slow boring of hard boards' (his gloss: making hardwood furniture with dull tools) and Keynes's 1938 letter to FDR about not squandering short-term prosperity. A substantive political-economy essay with an actionable execution checklist.
US politicsMamdaniWeberKeynespolitical strategy
TIER 4
Jun 28, 2025
Using Edwin O'Connor's 'The Last Hurrah' boss Frank Skeffington—undone by a New Deal and assimilation he never saw coming—as a foil, DeLong asks whether Mamdani's calm, adult comportment winning amid 'outrage saturation' signals a hopeful shift away from anti-rational spectacle politics. He balances the hope with sharp caveats: it may hold only in a Democratic primary, only in NYC, only this summer, since the politics of unreason are never far below the surface. A rich political-history meditation, though somewhat speculative.
US politicsMamdanimachine politicspolitical historyoutrage fatigue
TIER 4
Aug 7, 2025
A focused takedown of Kevin Hassett's fitness for Fed Chair, anchored in the Dow 36,000 episode: DeLong walks through the actual finance (the Gordon payouts equation vs. the resources equation) to show Hassett knowingly double-counted retained earnings, calling it a deliberate "2+2=5" lie rather than an error. He ties this to Hassett's present-day lies about the BLS jobs revisions, arguing a pattern of saying whatever is expedient disqualifies him. A pointed, technically grounded character-and-competence indictment.
Kevin HassettFederal ReserveDow 36000equity valuationFed Chair
TIER 4
Aug 14, 2025
DeLong rates Trump's likely Fed Chair candidates (Waller, Hassett, Warsh, Miran, Bowman) on three axes (intelligence/modeling, central-banking experience, moral character) and finds all wanting, with Waller the least-bad option. Along the way he explains the FOMC dissent norm and why governors historically vote with the chair, and invokes Arthur Burns's regret over caving to Nixon. A sharp, knowledgeable assessment of a consequential appointment with useful institutional context.
Federal Reservemonetary policyFed Chaircentral bankingTrump
TIER 4
Aug 26, 2025
DeLong argues Trump's attempt to oust Fed Governor Lisa Cook to force lower rates backfired: spooking investors and widening risk premia drove the long-term (spending-relevant) yields that matter up, not down. He ranks the sitting Governors' monetary-policy expertise — placing Cook at the top — and makes the counterintuitive point that Cook is, by Trumpist lights, the Governor most likely to argue persuasively for a low-rate high-pressure economy, so firing her is a self-defeating own goal. A substantive piece on Fed independence and the mechanics of long-rate determination.
federal-reservelisa-cookfed-independenceinterest-ratesrisk-premium
TIER 4
Sep 8, 2025
DeLong argues the Supreme Court's conservative majority has weaponized the emergency/shadow docket to hand Trump unexplained partisan victories, exempting him from the standards applied to every other litigant. He frames this against Ackerman's theory of 'constitutional moments,' noting that unlike Jackson, the New Deal, or Civil Rights, this shift is engineered from above rather than ratified by popular majorities, and only makes long-run sense if the six GOP justices believe Democrats will never again hold the presidency and both chambers. It matters as a structural diagnosis of judicial capture and the threat to the Article I branch's primacy.
supreme-courtshadow-docketconstitutional-momentstrumprule-of-law
TIER 4
Oct 7, 2025
A serious intellectual-history piece tracing Peter Thiel's 'Antichrist'/'Restrainer' fixation from 2 Thessalonians through Rene Girard's anti-scapegoating theology and Carl Schmitt's friend-enemy politics, arguing it functions as a governing theory linking doomsday theology to surveillance and anti-immigrant mobilization (with J.D. Vance). Matters as a guide to the theological-political ideas animating a powerful tech-right faction, though most of the substance sits behind the paywall.
Peter ThielCarl SchmittRene Girardpolitical theologytech right
TIER 4
Oct 22, 2025
Via Mike Brock on Elon Musk's humiliation, DeLong characterizes Trump's regime as patronage-autocracy rather than capitalism or feudalism: favors flow only to supplicants who frame benefits as grace and never claim reciprocity, with proximity to the sovereign outranking productive contribution. He links this to plutocrats' quiet strategy of invisibility and flattery (no real capital flight despite the talk) and to a dangerous power vacuum around a mentally-declining sovereign with no consolidating Wazir. A sharp original framework for the political economy of the second Trump term.
Trumppatronage-autocracyplutocratspolitical economyMusk
TIER 4
Oct 25, 2025
Using Niall Ferguson's Tomahawk-missiles-for-Ukraine prediction (falsified within hours of posting), DeLong argues Trump is stochastic not strategic and asks why smart analysts keep retrofitting grand design onto random flailing. He answers with a structured taxonomy of roughly a dozen pressures—risk-management fallacy, strategic-ambiguity bias, elite self-preservation, professional selection effects, media production constraints—that drive sanewashing. The most fully developed of the batch's anti-sanewashing pieces, with a reusable diagnostic list.
sanewashingTrumpUkrainemedia incentivespunditry
TIER 4
Nov 4, 2025
Riffing on Klein, Farrell, and Dean Acheson's 1955 'A Democrat Looks at His Party,' DeLong argues the Democrats remain a many-interest coalition while the GOP is the single-interest party—but updates Acheson by showing the Republican base has pivoted from Schumpeterian embrace of creative destruction to defensive custodianship of property, status, and symbolic hierarchy. The historical-political-economy reframing of the right's shift from dynamism to loss-aversion is the substantive payload, though the post paywalls before fully developing the Democratic side.
Democratic Partypolitical economySchumpeterhistory of ideascoalitions
TIER 4
Nov 9, 2025
Responding to Mike Brock's civic-first 'classical liberalism,' DeLong sides with Hayek: without secure private property and predictable markets, freedoms from fear, want, and speech collapse—so property is primary to, not instrumental to, human dignity. The argument's bite is empirical: via the Shadow Docket, Trump has reverse-engineered rule-of-law into 'confiscate now, litigate later,' giving him whim-and-spite power to discipline America's corporate and university 'barons' into silence. A substantive political-economy essay on democracy's failure modes (full version of the piece previewed in 0281).
political economyHayekrule of lawShadow DocketTrump
TIER 4
Dec 2, 2025
A John Ganz crosspost (framed by DeLong) arguing that "post-liberalism" and "national conservatism" are euphemisms—a "fascism lite"—and using Hannah Arendt's "Personal Responsibility Under Dictatorship" to reject Ross Douthat's appeal to impersonal historical forces as a moral evasion. The core claim: intellectuals who produce respectable cover for cruelty are culpable regardless of whether someone else would have done it. A sharp intellectual-history-inflected polemic, though authored by Ganz rather than DeLong.
post-liberalismintellectual historyArendtDouthatGanz crosspost
TIER 4
Dec 9, 2025
Using Walter Scheidel's review, DeLong dismantles Graeber & Wengrow's 'The Dawn of Everything,' arguing its grand anti-determinist narrative rests on speculation that hardens into fact and on roughly a third demonstrably wrong claims, making 'speculative nonfiction' an unsound genre. He presses a materialist case—modes of production, not free-floating ideas, set the boundaries of social possibility—and disputes Henry Farrell's softer defense. A pointed methodological and historiographical argument about evidence and grand narratives.
David Graeberhistoriographymaterialismstate formationintellectual history
TIER 4
Dec 22, 2025
DeLong crossposts Adrian Monck's account of Renmin University professor Nie Huihua's survival guide to Chinese officialdom (sharp eyes, zipped lips, thick skin) and appends a substantial essay placing China's bureaucratic pathologies in the long arc from the keju exam system (607–1905) to today. His thesis: the historic exam aligned competence and Confucian norms for an agrarian-age society-of-domination, but in the Schumpeterian age that same hegemonic bureaucratic 'operating system'—hierarchical resource allocation, land-finance dependence, inspection-centric control—makes truth dangerous and starves the center of honest signals.
Chinabureaucracyeconomic historypolitical economykeju
TIER 4
Dec 30, 2025
DeLong dismisses an NYT op-ed claiming a new Trump-friendly campus orthodoxy (its Harvard evidence is thin and borrowed from other schools) and instead lays out, via Jacob Levy and Paul Mahoney's anti-compact UVA letter, what universities actually owe: academic freedom (not public-square free speech) governed by merit-based assessment and institutional neutrality, with five member duties—speak, listen, think, learn, support. A useful explainer on the academic-freedom-vs-free-speech distinction, though paywall-truncated before its promised Noah Smith counterpoint.
academic-freedomuniversitiesneofascismfree-speechinstitutional-neutrality
TIER 4
Jan 10, 2026
Walking through a large French cohort study (22.8M vaccinated vs 5.9M unvaccinated), DeLong reasons that COVID mRNA vaccines cut not only COVID death (74% lower) but all-cause mortality (~25% lower), and argues the dominant channel is reduced post-infection metabolic and inflammatory burden rather than mere 'sensible person' selection effects. He works the arithmetic to estimate excess deaths among the unvaccinated and indicts vaccine skepticism. A solid quantitative-reasoning explainer applying selection-bias logic to epidemiology.
mRNA-vaccinesCOVID-19mortalityepidemiologyselection-bias
TIER 4
Jan 12, 2026
Using Cicero's January 12, -49 letter to Tiro, DeLong analyzes why Caesar's crossing of the Rubicon swung 'tota Italia' to his side so fast and bloodlessly—a study in coordination and recruitment dynamics in brittle institutions, where the cause that appears to back a functioning state wins. He puzzles over Cicero's surprise, his misplaced confidence in Pompey, and his belief he could have brokered peace. A sharp historical-snapshot essay framed in economic-incentive terms (sword supply and demand), though cut off at the paywall.
Roman-RepublicCaesarCiceroinstitutional-collapsecoordination
TIER 4
Jan 30, 2026
Riffing on Dan Davies and a Galbraith line on leadership, DeLong argues the emotional base layer of reactionary politics is not denial of immigrants' humanity but despair—the conviction that rich societies are too broken to solve solvable problems—which is factually false (the fiscal and productive capacity exists) but structurally invisible to optimistic policy elites. The five-point analysis frames 'manufacturing justified agency' via visibly solving human-scale problems as the real antidote to grift-driven xenophobia. A genuinely useful political-economy frame.
reactionary-politicspolitical-despairxenophobiagalbraithpolitical-economy
TIER 4
Feb 1, 2026
A Timothy Snyder crosspost (with DeLong framing) tracing, step by step, how JD Vance's 2024 Springfield fabrications about Haitians were amplified by neo-Nazi groups, laundered into 'facts,' and bureaucratized into a TPS revocation and imminent ICE surge. Snyder applies his atrocity-historian's template—the manufactured subhuman enemy, propaganda escalation, state capture, then violence—to a live American case. Substantive and disturbing political analysis, though largely Snyder's reporting rather than DeLong's economics.
ethnic-cleansingjd-vancehaitian-immigrantsneofascismtimothy-snyder
TIER 4
Feb 6, 2026
A Dan Davies crosspost, framed by DeLong via the Reinhart-Rogoff 'Excel-slop' cautionary tale, arguing that LLM coding tools like Claude Code are becoming a 'super-Excel' that scales the end-user-computing (EUC) governance problem to new heights. The useful frame: just as ungoverned spreadsheets created brittle, unaccountable shadow systems, AI-generated end-user apps may flood organizations with unmaintainable 'apps in a trenchcoat,' and it is unclear whether AI can clean up its own messes at reasonable cost.
ai-toolingend-user-computingclaude-codepolitical-economy-of-techsoftware-governance
TIER 4
Feb 7, 2026
A Steve Vladeck crosspost (with DeLong framing) dissecting how a Fifth Circuit panel adopted the Trump administration's fringe 'arriving aliens' theory of mandatory immigration detention that ~360 district judges had rejected. The legal payoff is the procedural anatomy of how AADC and J.G.G. forced thousands of individual habeas cases, letting the government cherry-pick favorable circuits to appeal. A substantive, well-sourced explainer of forum manipulation in immigration litigation.
immigration-lawfifth-circuithabeas-corpusrule-of-lawus-courts
TIER 4
Feb 12, 2026
Marshall argues that the real threat is not Trump but a durable 'Authoritarian International'—Gulf princelings, post-Soviet oligarchs, rightward Silicon Valley, and a swollen billionaire class—whose anti-civic world of private deals will outlive Trump. DeLong's substantial added essay celebrates Marshall's reader-entangled TPM as a model of the 'positive internet' and develops an original reflection on how most modern relationships, even with people we 'know', are now largely parasocial sub-Turing instantiations running on our own wetware. It matters for the political-economy diagnosis plus DeLong's parasocial-cognition riff.
political economyauthoritarianismattention economyparasocial relationsmedia
TIER 4
Feb 17, 2026
Dissecting Rubio's 'Plato-to-NATO' civilizational fairy tale, DeLong argues (with Goldhammer and Shklar) that the 'Western alliance' was invented out of whole cloth in post-WWII Washington to contain Stalin, distinguishing legitimate mythmaking-in-a-good-cause from mythmaking in a fascist cause. He adds an original provocation: US foreign policy from 1939–1953 was effectively the British Empire revealing it had quietly reabsorbed the United States as its most powerful late-mobilizing component. It matters for its intellectual-history framing of American exceptionalism and Anglo-American power, though it leaves its boldest claim unfinished.
intellectual historyUS foreign policymythmakingWestern alliancepolitical economy
TIER 4
Feb 20, 2026
DeLong's own twelve-point reaction to Learning Resources v. Trump, the 6-3 ruling clawing back Trump's IEEPA tariffs: he argues the real story is not the win but that the Court's 'corrupt and craven' middle gave Trump a full year of lawless tariffs by granting a stay, demonstrating that no one's property rights survive a Republican president's fake emergency (hence Tim Cook bending the knee). He also flags doctrinal straws in the wind (Gorsuch shrinking the Major Questions Doctrine, Thomas's privilege-vs-right move) and cites Yale Budget Lab effects. A substantive, well-organized legal-economic read of the opinion.
Supreme CourtTrump tariffsIEEPAMajor Questions Doctrineemergency powers
TIER 4
Feb 27, 2026
Applying 'legal realism' rather than doctrinal drapery, DeLong maps the Roberts Court not as a 3-3-3 institution with a moderate center but as a neofascist bloc plus fellow travelers, decoding each justice's self-conception: Alito/Thomas as fascists, Gorsuch as the coherent anti-administrative-state revolutionary, Kavanaugh as the 'responsible Republican' poseur, and Barrett as the decisive vote trying to rescue Scalia's project. He argues the shadow docket reliably lets a Republican president create facts on the ground. A sustained, original analytic portrait of the Court as a political-economic actor.
Supreme Courtlegal realismshadow docketadministrative stateTrump era
TIER 4
Apr 2, 2026
DeLong argues the Supreme Court's conservative majority deliberately slow-walked the birthright-citizenship case (Trump v. Barbara), reshaping remedies doctrine to give Trump his best shot at overturning the 14th Amendment guarantee, though he predicts it now lacks five votes. He frames the broader ICE-deportation project not as labor protection but as construction of a legally precarious, exploitable serf labor caste, backed by political-economy evidence on enforcement spending. It matters as a clear-eyed account of how procedural footdragging functions as covert partisanship and how immigration enforcement reshapes labor bargaining power.
birthright-citizenshipsupreme-court14th-amendmentlabor-political-economyimmigration-enforcement
TIER 4
Apr 6, 2026
A crosspost of Steve Vladeck's essay arguing that right-wing law professors manufactured 'an entire literature' to backfill Trump's birthright-citizenship position, and that the Supreme Court's 'history and tradition' turn structurally incentivizes such 'law-office history' by treating citation count, not historical truth, as the metric (Gorsuch's 'battle of law reviews'). DeLong's headnote adds a 'turtles all the way down' skepticism that legal scholarship was ever constrained by historical truth. A substantive piece on legal methodology and political economy of the courts.
Supreme Courtbirthright citizenshiporiginalismlegal scholarshipVladeck crosspost
TIER 4
May 4, 2026
DeLong argues Tesla is a trillion-dollar story-stock (P/E ~350, zero revenue growth since 2022) that has suffered nearly every bear-case event since the post-pandemic chip-shortage windfall, sustained only by fundamentals-faith or greater-fool dynamics. He extends the analysis to the looming SpaceX IPO (valuations from $660B to $2.8T per Damodaran's Monte Carlo) and flags the governance trap: Musk can book future moonshot profits to SpaceX rather than publicly-held Tesla. A substantive valuation/political-economy-of-Musk piece.
TeslaElon Muskstock valuationSpaceX IPOgreater-fool theory
TIER 4
Jun 1, 2026
Prompted by Kevin Hassett dismissing the Michigan Consumer Sentiment Index as a partisan survey, DeLong delivers a detailed forensic takedown of Hassett's career-founding 'Dow 36,000' fraud, showing precisely how he and Glassman double-counted earnings as both payout and growth driver (treating the Gordon 'resources' equation's output as 'payouts')—a deliberate 2+2=5 lie, not an honest error. He frames Hassett's rise as proof that lying for plutocrats and Republican politicians pays, closing with a Plato-on-the-tyrant meditation. A sharp political-economy/intellectual-honesty essay with a genuinely useful finance-formula explainer.
Kevin HassettDow 36000economic misconductstock valuationintellectual dishonesty
TIER 4
Jun 10, 2026
A full crosspost of DiResta's anatomy of the Spencer Pratt LA mayoral primary, showing how the 'majority illusion' makes online virality read as real support, how the predictable 'red mirage, blue shift' of mail-ballot counting gets weaponized into fraud narratives, and how influencers profit from permanent grievance. DeLong adds that early returns will again be framed as the 'real' election and everything after as contamination. A clear, useful explainer of disinformation mechanics ahead of 2026.
election-denialdisinformationmajority-illusionmail-in-ballotsus-politics
AI, Work & the Future of Education
0 tier-5 · 16 tier-4
What AI does to work and learning, separated cleanly from what AI *is*. On jobs, DeLong argues recent graduates' struggles are driven mostly by policy uncertainty freezing hiring, not by AI automation, and that broad productivity gains historically pay out only after complementary organizational redesign (the general-purpose-technology / "workslop" lag). On pedagogy, his stable thesis is that education has had one purpose for 5,000 years - training people as effective front-end nodes to the anthology super-intelligence via seven enduring "academic labors" - so the AI panic calls for a modest pivot (ten-minute live oral exams; teaching the right abstraction layer; using LLMs as a "rabbit"/pacer) rather than signaling existential collapse.
TIER 4
Mar 6, 2025
A draft essay using Victor Shih's 'coalition of the weak' theory of Mao-to-Deng elite politics to ask why a non-superstrong dictator cannot afford competent, well-networked lieutenants—and generalizing the pattern across medieval France/England (Philippe IV's bureaucrats, the overmighty subject) and Syme's Augustus. The esoteric target is contemporary autocracy; DeLong candidly admits he loses the analytical thread on Leninist network-selectorate politics. A rich, idea-dense draft despite being unfinished.
political economydictatorshipChinese politicselite theoryhistory
TIER 4
Jul 1, 2025
DeLong pushes back hard on AI-doom panic about higher education, arguing that for 5,000 years education has had one stable purpose—training people to be effective front-end nodes to humanity's anthology super-intelligence via seven enduring 'labors' (survey, question, research, analyze, store, persuade). He proposes a concrete fix: ten-minute live oral exams to verify the work is the student's own, making MAMLMs an opportunity rather than an existential threat. A useful, well-argued explainer offering a workable pedagogical pivot.
higher educationAIpedagogyMAMLMsassessment
TIER 4
Jul 23, 2025
Against the 'AI is taking entry-level jobs' narrative, DeLong argues recent graduates' relative labor-market struggle (the grad-vs-overall unemployment gap going sharply negative) is driven mostly by policy uncertainty over trade, immigration, and inflation that freezes hiring, plus capital flowing to NVIDIA chips rather than junior hires and the end of the rising college wage premium. He concedes a possible AI component in tech specifically. A clear, well-argued debunking that disentangles cyclical, structural, and technological causes.
labor marketcollege graduatespolicy uncertaintyAI automationwage premium
TIER 4
Aug 1, 2025
DeLong frames a month-long experiment using the AI-native Dia Browser as a vehicle for his broader thesis that LLMs are 'autocomplete on steroids'—spreadsheet-level not microprocessor-level technology, function machines that mimic the typical internet poster rather than achieving genuine cognition. He argues their real value is as natural-language interfaces and summarization 'cultural technologies' for taming information overload (the 'funnel' problem), not as reasoning agents. It matters as a clear, reference-quality statement of his calibrated skepticism about AI's economic impact.
AI/MAMLMsLLMsinformation overloadDia Browserproductivity
TIER 4
Aug 13, 2025
A mid-experiment report on living inside the AI-copilot Dia Browser, framing such tools as the next tightening of the human-machine feedback loop (the VisiCalc-of-words analogy) while still generating AI-slop. DeLong critiques the GPT-5 launch backlash, mocks the "immediacy trap" of instant hot takes, and argues the productive future lies in small on-device interface models bolted onto curated, verifiable backends rather than chasing AGI. A substantive synthesis of his recurring MAMLM thinking.
AIMAMLMsGPT-5browser copilotsknowledge tools
TIER 4
Aug 17, 2025
A cross-post of Noah Smith's essay (prompted by embryo-screening debate) arguing that romanticizing suffering is a coping mechanism and that "adversity is not worth the price of adversity." Using Keith Haring's AIDS-era Unfinished Painting, the conquest of maternal mortality, and Disney's Little Mermaid, it makes the case that progress trades depth-through-tragedy for a kinder world, and that this is the right trade. An eloquent, memorable essay on the meaning of progress, though it is a guest repost.
progresstechnology optimismhuman sufferingNoah Smithcross-post
TIER 4
Oct 2, 2025
A cross-post of Adam Mastroianni's essay arguing that the 'illusion of explanatory depth' explains the strange order of scientific discovery--math early, obvious things late--because we only investigate where we feel our ignorance, illustrated by Fechner founding experimental psychology after blinding himself staring at the sun. An entertaining, substantive piece on the epistemics of discovery that DeLong ties to collective (ASI) knowledge-building, though it is guest content.
philosophy of scienceillusion of explanatory depthdiscoveryAdam Mastroiannicross-post
TIER 4
Oct 4, 2025
DeLong uses a Bronze-Age sheep-counting parable to develop "abstraction layers" as cognitive force-multipliers we offload onto the real "Anthology Super-Intelligence," and argues good white-collar work means operating at the highest layer that still satisfies clients while knowing the layer or two below for when it leaks. He extends this to a critique of CS education—assignments too easy for the LLM age, courses pitched at the wrong abstraction layer, and Chad Orzel's education-vs-credentialing split. A thoughtful, idea-dense essay on AI-age pedagogy and cognition.
AIeducationabstractioncognitionpedagogy
TIER 4
Oct 28, 2025
Responding to the 'feminization of the workplace' panic (Helen Andrews, via Orzel/McArdle/Yglesias), DeLong reframes it as an institutional-design problem, not a civilizational crisis, since within-gender variance swamps between-gender differences. He offers an original framework from his own seminar practice: three interlocutor modes (challenge, support, triage) and the core skill of rapid context-switching, arguing economics is too hypercompetitive while history/sociology sometimes under-triage. The full (non-truncated) version of the essay, useful as a transferable model of organizational motivation design.
institutional designgendermotivationacademiaorganizations
TIER 4
Oct 29, 2025
DeLong's panel remarks at a UC congress on recent-grad labor outcomes: recent college grads' unemployment (4.8%) now sits above the national average for the first time, halfway between the college-worker and young-worker rates, amid hiring depressed ~25% below balanced-economy levels. He ties it to MAMLM uncertainty (employers freezing hiring against a 15-year commitment), his 2%-average / uneven 1%-vs-5% productivity-growth model with knowledge workers now in the Schumpeterian bullseye, and the pedagogical question of which abstraction layer to teach. Substantive labor-market analysis plus a memorable take on AI and higher education.
labor marketcollege graduatesAI and jobsproductivity growthhigher education
TIER 4
Nov 22, 2025
Drawing on David MacIver, DeLong argues LLMs are not oracles or colleagues but emulations of the Typical Internet S***poster, so good writers should use them as "rabbits"—pacers whose drivel you read, hate, delete, and out-write—rather than doing the costly, unreliable work of context engineering. He grants AI genuinely lifts the floor for weak writers, non-native speakers, and ritual boilerplate. A practical, opinionated workflow argument that crystallizes his view of AI's comparative advantage.
AIwriting workflowLLM usecontext engineeringMacIver
TIER 4
Dec 24, 2025
DeLong reads The Wrath of Khan as the contingent hinge on which the entire Star Trek franchise turned, arguing that without Nicholas Meyer's twelve-day script salvage and Hornblower-in-space reframe, Trek dies after the ponderous 1979 motion picture. The interesting throughline is the political economy of post-TV Hollywood: studios stopped greenlighting reliable 'singles' once they no longer had to fill owned theaters, demanding hypnotic 'guaranteed homer' pitches instead—an applied analysis of how IP franchises actually get made.
cultural-economicshollywood-ipstar-trekmedia-economicscontingency
TIER 4
Dec 29, 2025
Endorsing Paul Musgrave, DeLong argues the cure for AI-era cheating is the oldest assessment form—the ten-minute one-on-one oral exam on a student's own submitted work—and reframes the whole panic through his 5,000-year 'seven academic labors' model (survey, identify issues, hone a question, research, analyze, store, persuade) of training people as front-end nodes to the human 'anthology super-intelligence.' The thesis: AI is just another literacy-technology shift like papyrus or Gutenberg, requiring a modest pedagogical pivot rather than signaling existential collapse. Substantive framework, with a long reprinted companion essay.
ai-and-educationuniversitiesassessmentanthology-super-intelligencepedagogy
TIER 4
Jan 20, 2026
A crosspost of David Deming's essay synthesizing the evidence on AI and learning: personalization (tutoring, tracking, adaptive CAI like Mindspark) produces huge learning gains, but generative AI without guardrails harms learning because students offload the cognitive work (the PNAS Turkish-classroom study, the BCG 'exoskeleton' study, dulled autopilot-era pilot skills). The thesis is that AI in education is an agency problem, not just a technology problem: gains vanish when the AI is removed unless students first do the work themselves. A genuinely useful, evidence-dense explainer of the AI-learning literature.
AI and educationlearning sciencecognitive offloadingpersonalized learningDavid Deming
TIER 4
Jan 27, 2026
Building on Adrian Monck's reporting on the Zhang Youxia purge—which has gutted five of six uniformed CMC members—DeLong argues Xi has built a permanent 'Inquisition machine' on the conviction that a corrupt PLA is no army at all, a belief reinforced by watching kleptocratic Russian forces die by the tens of thousands in Ukraine. The Stalin/Kirov analogy frames the self-reinforcing logic by which an anti-corruption inquisition keeps finding heretics. A substantive read on Chinese civil-military relations, though DeLong admits he's 'at sea' on the organizational consequences.
chinaxi-jinpingpla-purgecivil-military-relationsstalin-parallels
TIER 4
Mar 18, 2026
A 2023 Timothy B. Lee essay (crossposted, with brief DeLong framing) arguing that just as 'software ate' only a slice of the world, AI will be disruptive but not produce mass unemployment, because robotics is hard, human interaction is a valued luxury, demand grows the pie, and employment is ultimately set by macro policy. DeLong frames it as a well-deserved victory lap for a useful corrective to AI-panic. It matters as a durable, carefully reasoned baseline case against the AI-unemployment thesis, though the core text is three years old.
ai-and-jobsmass-unemploymentroboticsautomationtimothy-b-lee