Research Brief — Domain: The Founding (2016 Origins)

Domain: The real pains of 2016, the birth of the MAGA brand, and who the 2016 coalition actually was. Parent topic: The Enshittification of MAGA (60-min video essay) Role in the essay: This is the steelman core. Everything downstream depends on establishing that the 2016 grievances were real, documented, and legitimate — and on being honest about where the evidence stops.

Research Summary

The 2016 grievance set is empirically solid and survives hostile scrutiny — the China shock, the mortality reversal among non-college Americans, the 2008 rescue asymmetry, and the collapse of prime-age male labor force participation are all documented in peer-reviewed or primary-source material. What does not survive scrutiny is the causal story most often bolted onto them. Three specific places where the essay must be careful:

1. The scholars who documented the pain reject the prescription. Autor, Dorn and Hanson are explicit that they know of no research justifying tariffs as a remedy, and the weight of evidence attributes the shock primarily to China's own internal reforms, not to American trade policy. The pain is real; "they sold you out with NAFTA and PNTR" is a much weaker claim than the pain itself.
2. "Deaths of despair" is a contested framing built on real numbers. The mortality rise is real. The claim that it is caused by generalized economic despair is disputed by serious researchers (Novosad/Rafkin/Asher; Ruhm; Gelman on age composition), and the trend has reversed sharply since 2023 — a fact the essay cannot ignore in 2026.
3. Racism vs. economic anxiety cannot be resolved as either/or, and the literature's most-cited "racism won" paper has a published, unrebutted-in-part statistical rebuttal. Mutz (2018) has a direct methodological critique from Stephen Morgan in Socius. Sides/Tesler/Vavreck's "racialized economics" is the strongest available synthesis precisely because it refuses the binary.

On the brand: the documentary record for MAGA-as-brand-before-movement is unusually clean — a signed trademark application dated days after Romney's 2012 loss, a registered service mark, and "Keep America Great" trademarked before the first inauguration. The stunt-vs-genuine-bid question is not settled and should not be presented as settled; the best-sourced version is "a brand exercise that became real, with the campaign's own people disagreeing about when."

Corrections to research-foundation.md found in this domain are listed in a dedicated section at the end. The most important: the 59.3% figure is not what the paper's abstract says, and should not be used as written.

1. The China Shock, Precisely

1.1 The original findings

Autor, Dorn & Hanson, "The China Syndrome: Local Labor Market Effects of Import Competition in the United States," American Economic Review 103(6), 2013.

• https://www.nber.org/papers/w18054
• https://www.ddorn.net/papers/Autor-Dorn-Hanson-ChinaSyndrome.pdf
• Original finding: rising Chinese import competition explains roughly a quarter of the aggregate decline in US manufacturing employment over 1990–2007, with trade-exposed local labor markets showing higher unemployment, lower labor force participation, lower wages, and larger transfer payments (TAA, disability, unemployment insurance, food stamps).

Acemoglu, Autor, Dorn, Hanson & Price, "Import Competition and the Great US Employment Sag of the 2000s," Journal of Labor Economics 34(S1), 2016.

• https://www.nber.org/papers/w20395
• This is the source of the widely-quoted headline numbers: import competition from China cost the US roughly 985,000 manufacturing jobs between 1999 and 2011, and — once input-output linkages and local demand spillovers are included — 2.0 to 2.4 million jobs overall.
• Precision note for script: the "1 million / 2.4 million" pairing is a fair summary of this paper. But "2.4 million" is the top of a range whose bottom is 2.0 million. Say "roughly two to two and a half million" or cite the range. Do not present 2.4M as a point estimate.

Autor, Dorn & Hanson, "The China Shock: Learning from Labor Market Adjustment to Large Changes in Trade," Annual Review of Economics 8, 2016. — https://www.nber.org/papers/w21906

• The synthesis paper. Key finding for the essay: adjustment was not the frictionless reallocation the textbook predicts. Displaced workers did not move. Wages in affected commuting zones stayed depressed. The adjustment mechanism was exit from the labor force, not relocation to a better job.

1.2 The 2021 persistence follow-up

Autor, Dorn & Hanson, "On the Persistence of the China Shock," NBER Working Paper 29401 (Oct 2021); published in Brookings Papers on Economic Activity, Fall 2021.

• https://www.nber.org/papers/w29401
• https://www.brookings.edu/articles/on-the-persistence-of-the-china-shock/

Verified from the abstract (exact figures):

• Import competition implied a reduction in the manufacturing employment-population ratio of 1.54 percentage points, which the authors state is 55% of the observed change in that value.
• 86% of this net job loss was absorbed via a corresponding decrease in the overall employment rate — i.e. the jobs did not reappear elsewhere in the local economy.
• "Adverse impacts of import competition on manufacturing employment, overall employment-population ratios, and income per capita in more trade-exposed U.S. commuting zones are present out to 2019."
• Population headcount reductions register only for foreign-born workers and native-born aged 25–39, "implying that exit from work is a primary means of adjustment."
• Brookings' own summary: adverse effects "grew increasingly negative until 2013, and then persisted through at least 2018," even though the import surge itself "plateaued around 2010."

⚠️ CORRECTION TO research-foundation.md. The foundation states the follow-up found the shock "accounted for 59.3% of all U.S. manufacturing job losses between 2001 and 2019." That figure does not appear in the NBER abstract and the framing is wrong in two ways: (a) the abstract's comparable figure is 55%, and (b) it refers to the change in the manufacturing employment-population ratio in trade-exposed commuting zones, not to "all U.S. manufacturing job losses" nationally. Those are different quantities. Do not use "59.3% of all U.S. manufacturing job losses." If a 59.3% figure exists in a table in the paper body, it must be located and its exact referent stated before use. Safe formulations: "55% of the observed decline in the manufacturing employment-to-population ratio," or the more conservative "the effects were still visible in 2019, two decades after the shock began."

1.3 The critical honesty point — the authors reject the remedy

This is the single most important nuance in the domain, and it is well documented.

• Autor, Dorn & Hanson, verbatim: "We are aware of no research that would justify ex-post protectionist trade measures as a means of helping workers hurt by past import competition." (quoted in Cato Institute, "The 'China Shock' Demystified" — https://www.cato.org/publications/china-shock)
• In the 2021 persistence paper, they find the Trump-era tariffs "appear to have succeeded in raising the prices of U.S.-made goods but not in expanding employment" in the protected industries. (Brookings summary, above.)
• Their preferred remedies are place-based policies: expanded unemployment insurance, an expanded EITC targeted at shocked labor markets, and regionally tailored retraining. Gordon Hanson has said this directly in Brookings interviews.
• They also found Trade Adjustment Assistance's effect on per-capita income in trade-exposed areas was "vanishingly small" — which is its own indictment of the bipartisan policy answer, and useful for the essay: the safety net that was supposed to catch these workers did almost nothing.

1.4 The critical honesty point #2 — the cause was mostly China's internal reform

• Amiti, Dai, Feenstra & Romalis and related work attribute roughly two-thirds of the effect of China's WTO entry on US manufacturing to China's own tariff reductions — not to US concessions.
• Handley & Limão found that granting China Permanent Normal Trade Relations accounted for only about one-third of Chinese export growth. PNTR did not "open" the US to China: China had held most-favored-nation status since 1980, and Chinese imports increased more than sixfold in the decade before PNTR.
• Source for both: Cato, "The 'China Shock' Demystified" — https://www.cato.org/publications/china-shock

What this means for the essay's honesty: "Washington sold you out by letting China in" is the weakest link in the 2016 story. The dislocation was real and severe. The villain identification was partly wrong. This is a strong, non-obvious beat: the anger was justified, the diagnosis was borrowed, and the cure was sold by people who knew it wouldn't work.

1.5 The contrary case (steelman the skeptics)

The China shock literature is not unanimous, and a fact-checked essay should say so.

• Caliendo, Dvorkin & Parro (Econometrica 2019) estimate only ~550,000 manufacturing jobs lost to the China shock 2000–2007 — roughly 16% of the total decline in that window. That is a fraction of the ADH-lineage estimates.
• Feenstra & Sasahara (2018) estimate 1.8–2.0 million.
• de Chaisemartin & Lei (2023) identified a methodological flaw in the original 2013 ADH specification. (Cato summary; the underlying critique concerns the shift-share/Bartik instrument.)
• Cato's framing, useful as opposition: the ~1 million directly-attributed manufacturing jobs were "less than 20 percent of the total manufacturing job losses" in 1999–2011; manufacturing's share of the workforce was falling steadily before and after the shock, and the US gained more than 1.5 million manufacturing jobs between 2010 and 2023 while Chinese imports continued.
• Automation is the competing explanation. The standard counterweight is that US manufacturing output kept rising while employment fell — productivity, not trade, did most of the killing. (Note: this claim is itself partly an artifact of computer/semiconductor sector deflators; Susan Houseman's work at Upjohn is the essential citation for why the "output kept rising" story is overstated — https://www.upjohn.org/ — worth one more search at fact-check.)

Editorial guidance: the essay does not need to adjudicate this. It needs to say: the range of credible estimates runs from about half a million to about two and a half million jobs, concentrated in specific places that never recovered. Even the low end is a catastrophe if it lands on your town.

2. Deaths of Despair

2.1 The original finding

Case & Deaton, "Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century," PNAS 112(49), Dec 2015.

• https://www.pnas.org/doi/10.1073/pnas.1518393112 (paywalled to WebFetch; abstract widely reproduced)
• Finding: between 1999 and 2013, all-cause mortality rose for middle-aged (45–54) white non-Hispanic Americans, reversing decades of decline and diverging sharply from every other rich country and from other US racial groups. The reversal was concentrated among those with a high school degree or less. The proximate causes: drug and alcohol poisoning, suicide, and chronic liver disease/cirrhosis.
• This is the paper that put "deaths of despair" into circulation. It was published in December 2015 — six months into Trump's campaign. That timing is itself a usable beat: the statistical confirmation of what those communities already knew arrived while he was on the trail.

2.2 The 2020 book and the 158,000 figure

Anne Case & Angus Deaton, Deaths of Despair and the Future of Capitalism (Princeton University Press, 2020).

• https://press.princeton.edu/books/hardcover/9780691190785/deaths-of-despair-and-the-future-of-capitalism
• 158,000 Americans died deaths of despair in 2017 — suicide, drug overdose, alcoholic liver disease combined.
• The framing: "the equivalent of three fully loaded Boeing 737 MAXes falling out of the sky every day, with no survivors." This is verified as their comparison and is widely quoted. Rhetorical note: the 737 MAX reference is dated — it lands hardest for viewers who remember the 2019 grounding. Consider stating the raw number first and the analogy second, or updating the analogy.
• The book's causal argument is not simply "poor people are sad." Their central mechanism is the American health care financing system — employer-sponsored insurance costs functioning as a head tax on low-wage labor, pricing less-educated workers out of good jobs and transferring their wages upward — plus declining union/labor power, rising corporate power, and the collapse of stable working-class social institutions (marriage, church, community).
• This is a gift to the essay's honesty requirement: the two economists most associated with the "forgotten working class" story blame concentrated corporate power and the health care system, not immigration or trade deals. Deaton in particular has become a critic of his own profession's consensus. The pain diagnosis and the MAGA prescription point in opposite directions.

2.3 What the trend has done since — MANDATORY UPDATE FOR A 2026 ESSAY

The essay is being made in 2026. The deaths-of-despair curve has bent, hard, and using only pre-2023 figures would be a fact-check failure.

• US drug overdose deaths fell from roughly 110,000 in 2023 to roughly 80,400 in 2024 — about a 27% decline, the largest one-year drop ever recorded and a reversal of a two-decade trend. (CDC provisional data; widely reported. Confirm final CDC/NCHS figures at fact-check: https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm)
• Composite "deaths of despair" (overdose + alcohol + suicide) fell roughly 16% nationally in 2024 vs 2023 — reported as the first significant decrease since 1999. Regional reporting confirms the pattern (e.g. https://www.stlpr.org/health-science-environment/2026-06-04/despair-deaths-suicide-missouri-illinois ; https://www.kcur.org/health/2026-06-02/despair-deaths-suicide-missouri-kansas)
• Suicide and alcohol deaths fell more modestly (~4% and ~3%).
• Cause of the decline is itself contested: the "fentanyl hypothesis" (declining potency/purity of the illicit supply) vs. the "infrastructure hypothesis" (naloxone distribution, harm reduction, treatment expansion). See preprints: https://www.medrxiv.org/content/10.1101/2025.10.24.25338732.full.pdf and https://www.medrxiv.org/content/10.64898/2025.12.04.25341579.full.pdf
• Why this matters to the argument, not against it: if the deaths are falling because of public health infrastructure — naloxone access, Medicaid-funded treatment — then the movement that grew out of those deaths has spent 2025–2026 cutting exactly the programs that reversed them. That is a devastating, verifiable enshittification beat. It requires care: the causal attribution is not settled.

2.4 The critiques — steelman them

The "deaths of despair" frame is one of the most contested constructs in social science. The essay should use the numbers and be careful with the word "despair."

• Novosad, Rafkin & Asher (Review of Economics and Statistics / NBER): mortality deterioration is concentrated in roughly the bottom 10% of the education/income distribution, not across the working-class majority. For white Americans, only the worst-off deteriorated; the rest were roughly flat. For Black Americans, the worst-off stayed flat while everyone else improved. — https://paulnovosad.com/pdf/novosad-rafkin-asher-mortality.pdf
• Andrew Gelman: part of the widely-cited 1999–2013 white midlife mortality rise is an artifact of age composition within the 45–54 bucket (the average age inside the bin rose as boomers moved through). Age-adjusting shrinks the effect substantially. — https://statmodeling.stat.columbia.edu (search "Case Deaton age adjustment")
• Selection into the non-college group (Caroline Hoxby and others): as college attainment rose, the remaining non-college population became a more negatively-selected group. Some of the "gap widening" is composition, not deterioration.
• Christopher Ruhm (UVA): argues the mortality rise is driven far more by drug supply conditions — prescription opioid marketing, then heroin, then fentanyl — than by economic despair. His work finds that adverse county economic conditions explain only a small share of the increase in drug mortality. Case & Deaton wrote a direct reply: https://www.princeton.edu/~accase/downloads/Case_and_Deaton_Comment_on_CJRuhm_Jan_2018.pdf
• Matthew Yglesias / Slow Boring provides the accessible popular version of the critique — https://www.slowboring.com/p/the-deaths-of-despair-narrative-is — arguing these are "discrete public health challenges" (opioid supply, smoking/seatbelt policy divergence between states) rather than symptoms of generalized despair.
• Case & Deaton's own shifting frame is a fair criticism: the 2015 paper was about white non-Hispanics; the book reframed around education; later work includes all races. Critics say this mobility of framing weakens the causal claim.

Defensible script formulation: "Between 1999 and 2017, deaths from suicide, overdose, and alcohol among Americans without a college degree rose to a level with no precedent in a rich country outside of wartime or an epidemic. Researchers argue about the word 'despair' — some say it's really an opioid supply story, some say it's concentrated in a smaller group than the headline suggests. Fine. Argue about the mechanism. Nobody disputes the bodies."

2.5 The opioid–Trump vote correlation (use carefully)

Goodwin, Kuo, Brown, Juurlink & Raji, "Association of Chronic Opioid Use With Presidential Voting Patterns in US Counties in 2016," JAMA Network Open, June 22, 2018.

• https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2685627
• Sample: ~3,100 US counties; Medicare Part D data on 3.76 million enrollees (20% national sample).
• Unadjusted county-level correlation between chronic opioid prescribing rate and Republican vote share: 0.32 (p<.001); the adjusted correlation reported at 0.42.
• Republican vote share alone explained 18% of the variance in county chronic-opioid-use rates.
• After adjusting for income, education, unemployment, disability rates and other demographics, presidential vote explained only 7% — i.e. roughly two-thirds of the raw association is accounted for by economic and social conditions.
• Authors' conclusion (paraphrase): Republican vote in 2016 "is a marker for physical conditions, economic circumstances, and cultural forces associated with opioid use."
• Caveats to state on screen: Medicare beneficiaries only; prescription opioids only (roughly half of opioid deaths); county-level correlation, not individual-level — this is an ecological analysis and cannot say that opioid users voted for Trump.

This is the single most visually powerful data relationship in the domain (see Visual Research), and also the one most likely to be overclaimed. The honest version: the map of who was in pain and the map of who swung to Trump are close to the same map — and most of that overlap is explained by the underlying economic conditions both share.

3. The 2008 Asymmetry and the Broader Grievance Set

3.1 Banks rescued, homeowners not — the actual numbers

This is the most emotionally load-bearing grievance and the numbers hold up.

TARP (banks):

• Authorized at $700 billion (Emergency Economic Stabilization Act, Oct 3, 2008), later reduced to $475 billion by Dodd-Frank.
• Treasury's own accounting: the bank programs were repaid with a profit to taxpayers. Capital Purchase Program: ~$205B disbursed, ~$227B recovered. — https://home.treasury.gov/data/troubled-assets-relief-program
• Roughly $245B went to banks; AIG ~$182B in total commitments across Treasury and the Fed; the auto rescue ~$80B.
• Zero senior executives of major financial institutions went to prison for conduct related to the crisis. (The standard citation is Jed Rakoff, "The Financial Crisis: Why Have No High-Level Executives Been Prosecuted?", New York Review of Books, Jan 9, 2014 — https://www.nybooks.com/articles/2014/01/09/financial-crisis-why-no-executive-prosecutions/ — a federal judge making the argument, which is much stronger than a pundit making it.)

HAMP (homeowners):

• Announced Feb 18, 2009. Obama promised it would help 3 to 4 million homeowners avoid foreclosure.
• Allocated $75 billion. Through September 2015, Treasury had spent about $10.2 billion on HAMP (plus ~$2B on related programs) — roughly one-seventh of the commitment.
• ~1.3 million homeowners received a permanent modification (Treasury, March 2014 MHA report); the broader Making Home Affordable umbrella claims ~1.8 million families "helped directly."
• SIGTARP (the TARP inspector general) documented that about 70% of homeowners who applied were denied a permanent modification, and that cancelled trial and permanent modifications exceeded the number of homeowners who got permanent help.
• Sources: https://home.treasury.gov/data/troubled-assets-relief-program/housing/mha/hamp ; https://home.treasury.gov/system/files/initiatives/financial-stability/reports/Documents/March%202014%20MHA%20Report%20Final.pdf ; https://www.housingwire.com/articles/tarp-watchdog-questions-if-hamp-results-are-worth-cost/
• The killer detail (needs one more verification pass, but is well-reported): Treasury Secretary Tim Geithner's reported framing that HAMP would "foam the runway" for the banks — i.e. spread foreclosures out over time so banks could absorb them — as recounted by Neil Barofsky, the TARP special inspector general, in Bailout (2012). Barofsky is an on-record named source with a first-person account. This is the "the homeowner program existed to protect the banks" claim and it should be attributed to Barofsky's account, which Geithner has disputed. — https://theintercept.com/2015/12/28/obama-program-hurt-homeowners-and-helped-big-banks-now-its-dead/

Foreclosures: roughly 6–10 million American homes were lost to foreclosure or short sale between 2007 and 2014 depending on the counting method (CoreLogic's completed-foreclosure series is the standard source: ~7.8 million completed foreclosures Sept 2008–2014, plus ~2M+ short sales). Verify the exact figure and series at fact-check — this number is often cited loosely.

The asymmetry in one line: the banks got $700 billion authorized and paid it back with interest. The homeowners got $75 billion authorized, $10 billion spent, and a 70% rejection rate. Nobody went to jail.

3.2 Wage stagnation

• Real median household income did not surpass its 1999 level until 2016 (Census/CPS ASEC). — https://fred.stlouisfed.org/series/MEHOINUSA672N
• Real average hourly earnings for production and non-supervisory workers were roughly flat from the early 1970s to the mid-2010s. — https://fred.stlouisfed.org/series/AHETPI (nominal; deflate by CPI) and Pew's summary: https://www.pewresearch.org/short-reads/2018/08/07/for-most-us-workers-real-wages-have-barely-budged-for-decades/
• Productivity–pay divergence: EPI's series showing net productivity growth vastly outpacing typical worker compensation since 1979. — https://www.epi.org/productivity-pay-gap/ (Note: this series has methodological critics — Robert Lawrence and others argue much of the gap disappears with consistent deflators and inclusion of benefits. Flag if used.)

3.3 Prime-age male labor force participation

• FRED series LNS11300060 — Labor Force Participation Rate, Men, 25–54. Peaked near ~98% in the mid-1950s, fell steadily to roughly 88% in 2015, and has partially recovered since. (Fetch blocked at research time; pull the exact series and current value at fact-check from https://fred.stlouisfed.org/series/LNS11300060)
• The canonical policy treatment: Council of Economic Advisers, "The Long-Term Decline in Prime-Age Male Labor Force Participation," June 2016 — https://obamawhitehouse.archives.gov/sites/default/files/page/files/20160620_cea_primeage_male_lfp.pdf — published five months before the 2016 election, by the outgoing Democratic administration. That is a fact worth using: the White House itself documented the problem and it did not change the election.
• Nicholas Eberstadt, Men Without Work (AEI, 2016) — the conservative treatment of the same data; useful as a cross-ideological corroboration.

3.4 Rural hospital closures

UNC Sheps Center, Rural Hospital Closures tracker — https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/

• 197 rural hospital closures and conversions since January 2005 — 109 complete closures, 88 converted (no longer inpatient but still providing some services).
• 154 since 2010 — 86 complete, 68 converted.
• Peak year in the data: 2019 (~20 facilities).
• Sheps' definitions matter: a "converted closure" still provides some care. Do not say "197 rural hospitals shut their doors" — say "197 rural hospitals closed or stopped providing inpatient care."
• Concentrated in Texas, Tennessee, Georgia, Missouri, Alabama, Oklahoma, Mississippi — largely non-Medicaid-expansion states, which is itself an argument.

3.5 Life expectancy divergence by geography

• Dwyer-Lindgren et al., "US County-Level Trends in Mortality Rates for Major Causes of Death," JAMA / IHME — county-level life expectancy gaps of more than 20 years between the highest and lowest US counties. https://www.healthdata.org/research-analysis/library/us-county-level-trends-mortality-rates-major-causes-death-1980-2014
• Chetty et al., "The Association Between Income and Life Expectancy in the United States, 2001–2014," JAMA 2016 — the gap in life expectancy between the top 1% and bottom 1% of the income distribution was 14.6 years for men and 10.1 years for women. https://jamanetwork.com/journals/jama/fullarticle/2513561 — and critically, the gap widened over 2001–2014, and geography mattered enormously for the poor but almost not at all for the rich.
• This is one of the strongest single facts in the domain: how long you live in America depends on your zip code, and the gap got worse during the years everybody was told the recovery was working.

4. The Bipartisan Abandonment Argument

This is the part of the 2016 case that the show's own editorial guidelines demand be taken seriously — "we hold the left accountable when it screws up."

4.1 The trade votes (the receipts)

• NAFTA implementing legislation, Nov 1993. Passed the House 234–200 on Nov 17, 1993; Senate 61–38 on Nov 20. House Democrats split against it, roughly 102 yes to 156 no — meaning it passed on Republican votes with a Democratic president driving it. VERIFY exact tallies at https://clerk.house.gov/Votes and https://www.senate.gov/legislative/votes.htm
• Permanent Normal Trade Relations with China (H.R. 4444), 2000. House 237–197 on May 24, 2000; Senate 83–15 on Sept 19, 2000. Signed by Clinton Oct 10, 2000. Again: a Democratic president, a bipartisan majority, and organized labor on the losing side. VERIFY exact tallies.
• Why this matters structurally: the two policy decisions most associated with the China shock in the popular imagination were bipartisan elite consensus. There is no partisan villain available. That is exactly the condition under which an outsider candidate becomes possible.

4.2 Thomas Frank

Thomas Frank, Listen, Liberal: Or, What Ever Happened to the Party of the People? (Metropolitan, 2016).

• Published March 2016 — before the general election, before anyone thought Trump would win. Frank's argument: the Democratic Party reoriented from a labor party to a party of the professional/credentialed class, treating inequality as a problem of insufficient education rather than insufficient power.
• Also relevant: Frank, What's the Matter with Kansas? (2004) — the earlier, and now often criticized, version of the argument.
• Steelman the counter: Frank's thesis is contested by political scientists who point out that Democrats continued to win low-income voters overall, and that the shift is concentrated among white non-college voters specifically. Larry Bartels' Unequal Democracy and Democracy Erodes from the Top are the strongest scholarly counterweights.

4.3 Piketty's "Brahmin Left"

Thomas Piketty, "Brahmin Left vs Merchant Right: Rising Inequality and the Changing Structure of Political Conflict (Evidence from France, Britain and the US, 1948–2017)," WID.world Working Paper 2018/7.

• https://wid.world/document/brahmin-left-vs-merchant-right-rising-inequality-the-changing-structure-of-political-conflict-evidence-from-france-britain-and-the-us-1948-2017-wid-world-working-paper-2018-7/
• Core finding: across all three countries, left parties transformed over ~70 years from the party of the less-educated into the party of the highly-educated, while right parties remained the party of the high-income/high-wealth. Piketty calls this a "multiple-elite party system" — the Brahmin Left and the Merchant Right — with no party representing the low-education, low-income voter.
• This is the single best framework in the domain for the essay's purposes, because it is (a) transnational, so it isn't a US partisan talking point, (b) quantitative, and (c) it explains why an outsider could capture that vacuum in 2016 without having any coherent program. The vacuum was structural.

4.4 The Tea Party as precursor

• Skocpol & Williamson, The Tea Party and the Remaking of Republican Conservatism (Oxford, 2012) — the essential scholarly account. Key finding for the essay: the grassroots Tea Party's actual concerns were substantially about deservingness — earned benefits for people like them, resentment of benefits for people they saw as not having earned them — and the movement was simultaneously an authentic grassroots phenomenon and an astroturfed elite-funded one (FreedomWorks, Americans for Prosperity, Fox News promotion). https://global.oup.com/academic/product/the-tea-party-and-the-remaking-of-republican-conservatism-9780199832637
• The enshittification rhyme is exact and should be used: the Tea Party is the first cycle of the same pattern — genuine grassroots anger, captured by professional operators, monetized, and delivered to donor-class ends (the 2017 tax cut) that had nothing to do with the original grievance. MAGA is the second, larger run of the same machine. If the essay establishes this, the thesis stops being a one-off story about Trump and becomes a story about a repeatable mechanism.
• Rick Santelli's CNBC rant, Feb 19, 2009 — the founding moment, and note what he was angry about: a proposal to help underwater homeowners. The Tea Party's origin myth is literally the 2008 asymmetry, aimed downward at other homeowners rather than upward at the banks. That redirection is the whole story in miniature.

4.5 Obama-to-Trump pivot counties — the empirical fingerprint

Ballotpedia, "Pivot Counties" — https://ballotpedia.org/Pivot_Counties

• 206 counties voted Obama in 2008, Obama in 2012, then Trump in 2016.
• Located in 34 states. Iowa had the most, with 31.
• Trump won them collectively by more than 580,000 votes, average margin 11.45%. Obama had won them by 12.23% in 2008 and 8.22% in 2012 — a swing of roughly 20 points.
• Demographics (vs. the average US county): more white, fewer foreign-born, more people in poverty, fewer with a bachelor's degree. https://ballotpedia.org/Demographics_of_the_206_Pivot_Counties_that_voted_Obama-Obama-Trump
• Turnout: 75% of the 206 did not keep pace with the national turnout increase from 2012 to 2016; 109 of them saw turnout decline. This is important and under-used: the pivot was not primarily a surge of new Trump voters. In most of these counties it was Democrats staying home plus a smaller Republican gain.
• Full list: https://ballotpedia.org/List_of_Pivot_Counties_-_the_206_counties_that_voted_Obama-Obama-Trump ; 2020 outcomes: https://ballotpedia.org/Election_results,_2020:_Pivot_Counties_in_the_2020_presidential_election (fetch failed at research time — get the retained/boomerang split at fact-check)
• The load-bearing inference: a county that voted for a Black man named Barack Hussein Obama twice and then voted for Trump cannot be explained by simple, stable racism. Something changed — either what those voters wanted, or what was on offer, or how they were being talked to. That is the essay's opening onto the honest complexity.

Best analyses of who those voters were:

• Voter Study Group (Democracy Fund), "The Five Types of Trump Voters" by Emily Ekins (June 2017) — https://www.voterstudygroup.org/publication/the-five-types-of-trump-voters — segments Trump's 2016 coalition into American Preservationists, Staunch Conservatives, Anti-Elites, Free Marketeers, and the Disengaged. Crucially, Ekins finds only one of the five segments is the nativist/racial-resentment archetype, and it is also the most economically distressed and most supportive of government social programs. This is the best single source for "the 2016 coalition was not one thing."
• Yair Ghitza / Catalist and the Voter Study Group panel data on Obama-to-Trump switchers: roughly 9% of Obama 2012 voters voted Trump in 2016; these switchers were disproportionately non-college whites with moderate-to-liberal economic views and conservative views on immigration and race. — https://www.voterstudygroup.org/
• Note the honest finding buried in that data: Obama-Trump switchers were, on average, economically populist and culturally conservative — a combination neither party offered. That is Piketty's vacuum, made of actual people.

5. Racism vs. Economic Anxiety — The Honest Scholarly State

Editorial instruction from the assignment, restated because it matters: do not resolve this as either/or, and do not let the essay's convenience decide it. What follows is the actual state of the literature, with the strongest work on each side.

5.1 The "identity/racial attitudes" case

Sides, Tesler & Vavreck, Identity Crisis: The 2016 Presidential Campaign and the Battle for the Meaning of America (Princeton, 2018). https://press.princeton.edu/books/hardcover/9780691174198/identity-crisis

• Winner of the 2019 Richard E. Neustadt Award. Widely described as the definitive social-science account of 2016.
• Central finding: attitudes about race, ethnicity and religion — not personal financial hardship — distinguished Trump voters from other Republicans and from Obama voters who defected. "No other factor predicted changes in white partisanship during Obama's presidency as powerfully and consistently as racial attitudes."
• Their key concept — and the one the essay should actually use — is "racialized economics": "the belief that undeserving groups are getting ahead while your group is left behind." This is not "it was racism, not economics." It is "the economics were experienced and interpreted through a racial frame."
• Their account of why Trump specifically: Republican primary fragmentation, enormous free media coverage, and identity-based polarization around immigration.

Michael Tesler, Post-Racial or Most-Racial? (2016) — the prior work establishing that Obama's presidency itself increased the salience of racial attitudes in ordinary partisan choice ("racial spillover"). This is the mechanism that makes 2016 legible: not that America got more racist, but that racial attitudes got more load-bearing in partisan sorting.

Diana Mutz, "Status threat, not economic hardship, explains the 2016 presidential vote," PNAS 115(19), April 2018. https://www.pnas.org/doi/10.1073/pnas.1718155115

• Used a nationally representative panel (2012 and 2016 waves) to argue that changes in vote choice tracked perceived threats to group status (dominant-group status threat around race, gender, and America's global standing) rather than personal economic circumstances.

5.2 The published rebuttal to Mutz — DO NOT SKIP THIS

Stephen L. Morgan, "Status Threat, Material Interests, and the 2016 Presidential Vote," Socius 4, 2018. https://journals.sagepub.com/doi/full/10.1177/2378023118788217

• Morgan reanalyzed Mutz's own released data files and argued that (a) the relative importance of economic interests versus status threat cannot be effectively estimated with her cross-sectional models, and (b) her panel data are consistent with economic interests being at least as important as status threat. He published his Stata code publicly to enable replication.
• Mutz's reply: "Response to Morgan: On the Role of Status Threat and Material Interests in the 2016 Election," Socius 2018. https://journals.sagepub.com/doi/10.1177/2378023118808619 — she argues Morgan's results depend on variable recategorization without independent empirical justification, and asks why Trump supporters would oppose a stronger safety net if material interests drove them.
• Morgan's rejoinder: "Correct Interpretations of Fixed-effects Models, Specification Decisions, and Self-reports of Intended Votes: A Response to Mutz," Socius 2018. https://journals.sagepub.com/doi/full/10.1177/2378023118811502
• Andrew Gelman's commentary on the exchange, useful as a neutral third party: https://statmodeling.stat.columbia.edu/2018/05/14/status-threat-explain-2016-presidential-vote/ and https://statmodeling.stat.columbia.edu/2018/07/01/status-threat-explain-2016-presidential-vote-diana-mutz-replies-criticism/

⚠️ CORRECTION/AMPLIFICATION TO research-foundation.md. The foundation cites Mutz's status-threat research approvingly as fitting the synthesis. That is fine as far as it goes — but the foundation does not mention that Mutz's paper drew a formal published rebuttal in a peer-reviewed journal, with code, using her own data. If the essay cites Mutz's headline ("status threat, not economic hardship") without the caveat, it is citing a contested finding as settled. Either cite it with the Morgan exchange named, or don't cite the headline at all and use Sides/Tesler/Vavreck's "racialized economics" instead, which is more robust and better serves the argument anyway.

5.3 The economics-mattered case

• The Nation's review of Identity Crisis ("What Political Scientists Get Wrong About 2016") — https://www.thenation.com/article/archive/identity-crisis-and-the-roots-of-2016-book-review/ — argues the authors under-weight economics and that survey questions about personal financial situation are the wrong instrument for measuring a community-level economic collapse. This critique is substantive, not just ideological: a laid-off worker's own household finances may look stable while their town dies around them.
• Autor, Dorn, Hanson & Majlesi, "Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure," American Economic Review 110(10), 2020. https://www.nber.org/papers/w22637 — the strongest economics-side evidence, by the China shock authors themselves: trade-exposed commuting zones shifted toward Republicans, and had the 2002–2014 trade exposure been 50% lower, the counterfactual suggests Michigan, Wisconsin, and Pennsylvania would have elected the Democrat in 2016. This paper is the citation for "the economic shock had a measurable electoral effect," and it comes from the same team, in a top-five journal.
• Ekins/Voter Study Group segmentation (above) showing most Trump voters were not the nativist archetype.
• The Nation / Dissent left-economic critique tradition — https://www.dissentmagazine.org/article/beyond-the-backlash/

5.4 The defensible synthesis

Both literatures are real and they are measuring different things. The honest statement, which the essay can make without hedging into mush:

Racial and status attitudes were the strongest predictors of which individuals voted for Trump. Economic shocks were the strongest predictors of which places swung to Trump. Both are true because they describe different levels of analysis — and the place-level shock is part of what made the individual-level attitudes politically activatable.

That formulation is defensible against both Sides/Tesler/Vavreck and Autor et al., because it is what both of them actually found. It also avoids the trap: saying "it was just racism" makes the essay's steelman impossible, and saying "it was just economics" is contradicted by the best individual-level data.

One more honest note the essay should include: Trump lost among voters making under $50,000 in 2016 and won among voters making over $50,000 (exit polls). The "MAGA is the white working class" story is a place story and a within-white-voters story, not a raw income story. Getting this wrong is the single easiest way for a hostile fact-checker to discredit the whole essay. Verify 2016 exit poll income crosstabs at fact-check.

6. The Creator's Contested Claim — Did Left Cultural Messaging Drive Persuadables to Trump?

This section exists to protect the essay from its own creator's prior. The claim is researched here seriously and neutrally. The finding is: there is a real phenomenon underneath it, the timeline is a serious problem for the 2016 version of the claim, and the causal step cannot be established with existing evidence.

6.1 The strongest case FOR

David Shor and "popularism."

• The argument: parties should identify which of their positions poll well, campaign on those, and stay quiet about the unpopular ones; Democrats have systematically failed to do this because the party's staff and messaging apparatus is drawn from a college-educated, ideologically atypical slice of the electorate.
• Shor's related and more important structural claim: educational polarization. As the Democratic coalition became more college-educated, the people writing the messages became less and less representative of the median voter, and the party's positions drifted in directions its own persuadable voters disliked.
• Best accessible summaries: Ezra Klein's NYT interview with Shor (2021); Eric Levitz's New York magazine interview ("David Shor's Unified Theory of American Politics," July 2020); the Inside Higher Ed overview https://www.insidehighered.com/opinion/columns/higher-ed-gamma/2025/01/31/ongoing-struggle-between-popularism-and-progressivism
• Important scope note: Shor's own diagnosis is primarily about 2020, not 2016, and his best-known finding was about the Latino shift in 2020 and the effect of "defund the police" on 2020 down-ballot races. It is not a 2016 argument.

The "Great Awokening" data.

• Zach Goldberg (then Georgia State) documented, across multiple independent national survey series (ANES, GSS, Pew), a rapid and unusually large leftward shift among white liberals on race, immigration and identity questions beginning around 2012–2014 — in several series white liberals moved to positions to the left of the average Black or Hispanic respondent. Corroborated by media word-frequency analysis (LexisNexis counts of "racism," "white privilege," "systemic," etc., rising by orders of magnitude 2011–2019).
• Musa al-Gharbi — https://musaalgharbi.com/2023/02/08/great-awokening-ending/ — dates the onset to 2011 ("Beginning in 2011, there was a rapid shift in the ways people associated with the knowledge economy talk about, and engage on, 'social justice' issues") and the peak to 2020–2021. He is explicit that these shifts predate Trump rather than being a reaction to him. He locates the shifters specifically among highly-educated white liberals, knowledge-economy professionals, women, and early-career workers — and notes "everyone else in America continued shifting right." His book We Have Never Been Woke (Princeton, 2024) develops the structural explanation: elite overproduction — Awokenings are power struggles among "symbolic capitalists," established elites versus frustrated elite aspirants, and he counts multiple prior episodes going back to the late 1910s.
• al-Gharbi is the essay's best citation here, not Goldberg — al-Gharbi is a Columbia/Stony Brook sociologist publishing with Princeton UP, his framework is structural rather than culture-war, and he is not aligned with the American right. Goldberg's data are real but his subsequent public affiliations make him a citation a hostile viewer will use to dismiss the whole segment.

6.2 The strongest case AGAINST

• The timeline problem, and it is fatal to the naive version. "Defund the police" as a mass slogan dates to summer 2020. "ACAB" entered mainstream American discourse in the same window. The 1619 Project was August 2019. The DEI-training expansion was largely 2020–2021. None of this existed in 2016. Any claim that these specific slogans drove 2016 persuadables to Trump is anachronistic and will be destroyed on contact by an informed critic. What did exist by 2015–2016: Black Lives Matter (founded 2013, mass-salient after Ferguson, Aug 2014), campus speech controversies (Missouri, Yale, Nov 2015), the "political correctness" frame — which Trump attacked explicitly and repeatedly in 2015.
• David Forrest (Oberlin), "What's Wrong with the 'Great Awokening'?" — https://publicseminar.org/essays/whats-wrong-with-the-great-awokening/ — disputes three core Awokening claims: that white liberals drifted left of racial minorities (he argues strong majorities of liberals of all races hold similar views); that white liberals overstate racism's prevalence (he argues the shift tracked reality); and that the shift was elite-led "saviorism" (he argues it was responsive to minority-led movements, especially BLM). Note that Forrest's own critique is from the left — he thinks the problem is that the Awokening was too identity-focused and insufficiently class-linked.
• Sides/Tesler/Vavreck's finding cuts against the causal claim: what predicted defection to Trump was racial and immigration attitudes, not backlash to a specific set of slogans. If left messaging had been the driver, you'd expect the strongest effect among voters with moderate racial attitudes who were alienated by tone. That is not what the data show.
• Popularism's own empirical stumbles: Shor publicly warned of near-inevitable Democratic collapse in 2022; Democrats substantially outperformed. Critics — Timothy Noah's "popularism vs. deliverism" (https://timothynoah.substack.com/p/popularism-v-deliverism), The New Republic (https://newrepublic.com/article/164144/popular-things-wont-save-democratic-party) — argue that message discipline cannot substitute for material delivery, and that popularism is unfalsifiable in practice because "which positions are popular" shifts with elite cueing.

6.3 What the evidence actually supports — the line the essay can defend

1. A real, large, measurable attitude shift among college-educated white liberals began around 2011–2014 and peaked around 2020–2021. (Well documented; al-Gharbi, Goldberg, multiple survey series.) ✅
2. That shift made the Democratic coalition's messaging class less representative of its own persuadable voters. (Strongly supported; educational polarization is one of the best-documented facts in American politics.) ✅
3. Trump ran explicitly against "political correctness" in 2015–2016 and it was a durable applause line. (Verifiable from the record.) ✅
4. Specific slogans like "defund the police" drove 2016 vote switching. ❌ False — they postdate 2016. They are relevant to 2020 and 2024, and to the later acts of this essay, not to the founding.
5. Left cultural messaging was a primary cause of the 2016 realignment. ⚠️ Not established. It is one plausible contributing dynamic. The best individual-level data point elsewhere.

Recommended script framing: "There's a version of this argument I think is true and a version I think is anachronistic. The true version: by the mid-2010s the people writing the Democratic Party's messages had become demographically and ideologically unlike the people they needed to persuade, and that gap kept widening. The anachronistic version: that 'defund the police' cost Democrats 2016. It couldn't have. Nobody had said it yet." This is exactly the "we hold our own side accountable and we don't fudge the evidence" move the brand is built on, and it will earn more credibility than either the creator's original claim or a flat denial of it.

7. The Brand's Birth

7.1 The trademark timeline — verified

Trump's own account (to the Washington Post, Jan 2017): he says he came up with it the day after Romney's loss; he first considered "We Will Make America Great" (rejected — didn't have "the right ring") and "Make America Great" (rejected — implied America had never been great); he settled on "Make America Great Again." He claimed he was unaware of Reagan's 1980 use until 2015 — and, in the same breath, noted that Reagan "didn't trademark it."

Why this is the best single artifact in the whole essay: a man who says he did not know a former president had used the phrase nonetheless immediately noticed that the former president had failed to secure the intellectual property. The slogan was a brand asset before it was a movement. Registered, defended, licensed, and monetized. Everything the essay argues about what MAGA became is present in the filing date.

⚠️ USPTO VERIFICATION REQUIRED. The trademark record should be pulled directly from USPTO TSDR (https://tsdr.uspto.gov/) so the essay can show the actual document on screen — filing date, signature date, registration date, serial and registration numbers, and the owner of record (Donald J. Trump for President, Inc.). This is a public record and makes for a strong visual. Do not rely on secondary sources for the dates that appear on screen.

7.2 June 16, 2015 — the escalator

Verified quotes from the announcement (transcript: https://time.com/3923128/donald-trump-announcement-speech/):

• On Mexico: "When Mexico sends its people, they're not sending their best... They're bringing drugs. They're bringing crime. They're rapists."
• On China: "When was the last time anybody saw us beating, let's say, China in a trade deal? They kill us." And: "China has our jobs and Mexico has our jobs."
• The thesis line: "Sadly, the American dream is dead. But if I get elected president I will bring it back bigger and better and stronger than ever before."
• The slogan: "We need somebody that literally will take this country and make it great again."
• On money: "I'm really rich." / "I don't need anybody's money... I'm using my own money." He presented a summary claiming a total net worth of $8,737,540,000.

The structural point for the essay: in a single speech he names the two real, documented shocks — China and the collapse of the American dream — and attaches them to a third thing that is not a shock but a scapegoat (Mexican migrants). The grievance is real. The causal chain is a splice. That splice is the founding act of the whole enterprise and it is on tape.

Also on tape and useful: the announcement drew immediate reporting that some attendees were paid actors — Extra Mile Casting emails offering $50 to appear at the event were reported by The Hollywood Reporter (June 17, 2015). The Trump Organization initially denied it; the casting company was later reported to have been paid. Verify current state of this reporting before use — it is a perfect miniature of the thesis (the crowd for the announcement of a populist movement was partly hired) but it must be nailed down.

7.3 Stunt or genuine bid — the honest answer is "contested"

For the "stunt/brand exercise" reading:

• Michael Cohen, sworn testimony to the House Oversight Committee, February 27, 2019. Verified quote: "Mr. Trump would often say, this campaign was going to be the 'greatest infomercial in political history.' He never expected to win the primary. He never expected to win the general election. The campaign — for him — was always a marketing opportunity." And: he ran "to make his brand great, not to make our country great," with "no desire or intention to lead this nation — only to market himself and to build his wealth and power." Full text: https://edition.cnn.com/2019/02/27/politics/cohen-testimony-read/ ; official hearing record: https://www.congress.gov/116/chrg/CHRG-116hhrg35230/CHRG-116hhrg35230.pdf ; C-SPAN video: https://archive.org/details/CSPAN_20240513_140300_Michael_Cohen_at_2019_House_Oversight_Committee_Hearing
• ⚠️ MANDATORY ON-SCREEN CAVEAT: Cohen pleaded guilty to lying to Congress (Nov 29, 2018) about the Trump Tower Moscow project, and to campaign finance and tax charges. He was testifying while awaiting a three-year sentence, and Republicans on the committee spent the hearing making exactly this point. The essay must state this in the same breath as the quote, not in a footnote. Doing so is good for the essay: it demonstrates the standard of evidence the show holds itself to, right at the moment it's making its most quotable claim.
• Corroborating pattern (not corroborating testimony): Trump flirted with running repeatedly without running — a 1987 New Hampshire trip (Portsmouth Rotary Club, Sept 1987, funded by a local activist), a 1999–2000 Reform Party exploratory committee, and the 2011 birther-era flirtation. Roger Stone had urged him to run for decades. Reported plans for a "Trump TV" venture if he lost fit the infomercial theory (reported by the Financial Times, Oct 2016, and others).
• Michael Wolff, Fire and Fury (2018) — the election-night account of a shocked, unhappy Trump. Reliability caveat: Wolff's book has documented accuracy problems and disputed quotes. Use as color, corroborated by other accounts, never as a load-bearing source.

Against the "stunt" reading:

• Trump's trade grievance is genuinely decades old and consistent. The 1988 Oprah Winfrey interview — in which he complains about Japan, trade deficits, and America being "laughed at" — is the essential clip. He said essentially the same thing in a 1987 full-page newspaper ad (NYT/WaPo/Boston Globe, Sept 2, 1987) that cost him a reported ~$95,000. That is 28 years of consistency on one issue. It undercuts the pure-cynicism reading.
• Corey Lewandowski & David Bossie, Let Trump Be Trump (2017) — the campaign-insider account arguing it was real. Self-interested, obviously.
• Sam Nunberg and other early aides have given interviews describing internal belief that it was real earlier than outsiders assume. Nunberg's public statements are volatile; treat with care.
• Roger Stone has claimed long-term intent. Stone is a convicted felon (later pardoned) who lies as a professional practice; cite only for what he said, never as evidence of what is true.
• The strongest structural counterargument: whatever he intended in June 2015, by the time he was signing FEC filings, hiring state directors, and spending his own money on a primary calendar, the distinction between "stunt" and "bid" had stopped being meaningful. The honest answer is that a brand exercise and a presidential campaign are the same object in this case, and always were. That is a better line than picking a side.

7.4 The "oh shit" moment — candidates, all requiring verification

There is no single documented moment; the campaign's own people have named different ones. Do not assert one. Present the candidates and let the ambiguity do the work.

1. The announcement's aftermath (June–July 2015). The Mexico remarks cost him business deals (NBC, Univision, Macy's, Serta) and raised his primary polling. That inversion — punishment by institutions producing reward from voters — is reportedly what several aides identified as the signal.
2. The Phoenix rally, July 11, 2015. An overflow crowd at the Phoenix Convention Center, far beyond what the campaign had booked for. Multiple insider accounts point to this as the moment the operation realized the demand was real. Verify the attendance figure and the sourcing.
3. The John McCain comment, July 18, 2015 ("He's not a war hero... I like people who weren't captured"). Every political professional predicted this would end him. It did not. This is the moment the old rules were shown to be inoperative — arguably the more important realization than any crowd size.
4. The first debate, August 6, 2015 — 24 million viewers, a Fox News record at the time, and the Megyn Kelly exchange that he survived. Best framing: the "oh shit" moment wasn't a moment of ambition, it was a moment of discovery — the discovery that the punishment mechanisms of American politics had stopped working. Everything downstream in the essay follows from that discovery.

Two additional verified anchors for this section:

• The June 16, 2015 announcement used a casting company to supply paid attendees — this is stated flatly in the standard reference account of the campaign. (https://en.wikipedia.org/wiki/Donald_Trump_2016_presidential_campaign ; original reporting: The Hollywood Reporter, June 17, 2015.) The Mexico remarks cost him NBC, Macy's, Univision, and NASCAR.
• An Economist/YouGov poll released July 9, 2015 was the first major national poll to show Trump as the Republican front-runner — 23 days after the announcement. That is the date the stunt reading stopped being available to anyone paying attention.

8. Foreign Amplification, In Proper Proportion

The instruction is to frame this as amplification of pre-existing organic division, not its cause. The primary sources support that framing — including the bipartisan Senate report itself.

8.1 Primary source: Senate Select Committee on Intelligence, Volume 2 (bipartisan)

Report of the Select Committee on Intelligence, U.S. Senate, on Russian Active Measures Campaigns and Interference in the 2016 U.S. Election, Volume 2: Russia's Use of Social Media. PDF: https://www.intelligence.senate.gov/sites/default/files/documents/Report_Volume2.pdf

Verified verbatim findings (quoted from the report text):

• Finding 1: "the IRA sought to influence the 2016 U.S. presidential election by harming Hillary Clinton's chances of success and supporting Donald Trump at the direction of the Kremlin." The Committee found IRA social media activity "was overtly and almost invariably supportive of then-candidate Trump."
• Finding 2: the campaign was "part of a broader, sophisticated, and ongoing information warfare campaign designed to sow discord in American politics and society." IRA operational planning "goes back at least to 2014."
• The key sentence for the essay's framing: "the preponderance of the operational focus... was on socially divisive issues—such as race, immigration, and Second Amendment rights—in an attempt to pit Americans against one another and against their government." The Committee describes IRA operatives "consistently us[ing] hot-button, societal divisions in the United States as fodder." The divisions were already there. The IRA was a lever, not a source.
• Finding 3: the IRA also targeted Republican primary candidates — Ted Cruz, Marco Rubio and Jeb Bush were "targeted and denigrated." Useful and under-used: the operation was not simply pro-Republican, it was pro-chaos and pro-Trump specifically.
• Finding 4: "no single group of Americans was targeted by IRA information operatives more than African-Americans." Supporting data: over 66% of IRA Facebook ads contained a race-related term; the "Blacktivist" page generated 11.2 million engagements; five of the top ten IRA Instagram accounts focused on African-American issues; 96% of IRA YouTube content targeted racial issues and police brutality.
• Finding 5 — the anti-overclaiming finding: "paid advertisements were not key to the IRA's activity." Facebook ad spend was about $100,000 over two years, against IRA operational costs of roughly $1.25 million per month. The 3,393 purchased Facebook/Instagram ads are "comparably minor" next to 61,500+ organic Facebook posts, 116,000 Instagram posts, and 10.4 million tweets.
• Finding 6: the IRA "coopted unwitting Americans to engage in offline activities," including obtaining assistance from the Trump Campaign in procuring materials for and promoting rallies.

Documented reach numbers (from the report, sourced to the platforms):

• Facebook: as many as 126 million Americans came into contact with IRA-manufactured content via IRA Facebook pages at some point between 2015 and 2017. 470 pages/accounts identified; 3.3 million users followed IRA-backed pages; 76.5 million engagements (30.4M shares, 37.6M likes, 3.3M comments, 5.2M reactions).
• Facebook ads: 11.4 million people saw at least one of the 3,393 IRA ads. 44% of impressions occurred before Election Day; 56% after. Roughly 25% of the ads were never seen by anyone.
• Instagram: 20 million users reached per Facebook's testimony (researchers led by Renée DiResta argue this is an undercount). 133 IRA Instagram accounts; 12 with 100,000+ followers.
• Twitter: 3,800+ IRA accounts, 8.5 million tweets, 72 million engagements, 1.4 million users engaged with IRA-originated tweets. 57% of IRA Twitter posts were in Russian; 36% in English.
• Context the report itself provides: during the campaign season 128 million US Facebook users generated ~9 billion election-related interactions. In the final month alone, 67 million US users generated 1.1 billion Trump-related interactions. Total 2016 social digital ad spend was $1.4 billion.

The proportion argument, made with the report's own numbers: the IRA spent $100,000 on Facebook ads in an election where $1.4 billion was spent on social digital advertising — roughly one seventy-thousandth. A quarter of its ads were never seen at all. More than half its impressions landed after the election was over.

8.2 The persuasion-effect research

Eady, Paskhalis, Zilinsky, Bonneau, Nagler & Tucker, "Exposure to the Russian Internet Research Agency foreign influence campaign on Twitter in the 2016 US election and its relationship to attitudes and voting behavior," Nature Communications 14, 62 (2023). https://www.nature.com/articles/s41467-022-35576-9

Verbatim from the abstract: the authors "demonstrate, first, that exposure to Russian disinformation accounts was heavily concentrated: only 1% of users accounted for 70% of exposures. Second, exposure was concentrated among users who strongly identified as Republicans. Third, exposure to the Russian influence campaign was eclipsed by content from domestic news media and politicians. Finally, we find no evidence of a meaningful relationship between exposure to the Russian foreign influence campaign and changes in attitudes, polarization, or voting behavior."

• Sample: 1,496 US Twitter users, three survey waves April–October 2016, linked to their actual Twitter feeds. 10% of users accounted for 98% of exposures.
• Authors' own caveats, which must be stated: observational data, so no causal claim; Twitter only, so nothing about Facebook, Instagram, or YouTube; and nothing about non-social-media interference such as the GRU hack-and-leak operation, which is a separate and much better-evidenced intervention.

Also useful: Thomas Rid's assessment that the IRA was "the least effective of all Russia's interference campaigns" despite receiving the most coverage — see Rid, Active Measures (2020).

8.3 The framing the essay should use

Three claims, all defensible, in this order:

1. It happened, it was real, it was Kremlin-directed, and the Senate confirmed it on a bipartisan basis. (SSCI Vol. 2, above; Mueller Vol. I; the Feb 16, 2018 indictment of 13 individuals and 3 entities.)
2. Its measurable persuasive effect on vote choice is, as far as the best available research can tell, approximately nothing. (Eady et al. 2023.)
3. Both of those are true because the IRA was not manufacturing division — it was buying cheap leverage on division Americans had already built. The report's own words: they "used hot-button, societal divisions in the United States as fodder."

And the line that connects this domain to the rest of the essay: the IRA understood something in 2014 that the essay is about — that an American political audience could be farmed. They were early, they were foreign, and they were bad at it. The people who got good at it were domestic, and they were selling merchandise.

⚠️ Do not let the essay say or imply that Russia caused Trump's win, that the racism of 2016 was substantially manufactured abroad, or that the IRA "swung" the election. The creator's notes raise the question "how much of that racism existed organically vs. how much was amplified by foreign operatives." The research answer is: overwhelmingly organic; the foreign contribution was amplification of existing content, concentrated on people who already agreed, with no detectable persuasion effect. Saying so strengthens the essay — it forecloses the most common right-wing dismissal ("you people think Russia did it") and it keeps the responsibility where the thesis needs it: domestic.

9. Human Impact & Storytelling — Specific, Namable, Filmable

The steelman fails if it stays at the level of coefficients. These are the concrete anchors available for this domain.

9.1 Places with documented, filmable collapse

• Lordstown, Ohio (GM Assembly Plant). Trumbull County was an Obama-Obama-Trump pivot county. Trump told a Youngstown rally in July 2017: "Don't move. Don't sell your house... We're going to get those [jobs] values up." GM announced the plant would be "unallocated" in November 2018; the last Chevy Cruze came off the line March 6, 2019; ~1,700 jobs directly, thousands more in suppliers. This is the single best case study in the entire domain — real pain, an explicit promise, a documented broken promise, in a county that had voted for Obama twice. It is the whole essay in one town, and it happened during the first term.
• Galesburg, Illinois (Maytag). The 2004 closure that Obama himself used as a set-piece; Knox County. The refrigerator plant moved to Reynosa, Mexico. A long-running local-journalism record exists.
• Hickory / Catawba County, North Carolina (furniture) and the NC textile belt. The classic ADH-identified trade-exposed commuting zones.
• Martinsville and Henry County, Virginia (furniture). Documented in Beth Macy's Factory Man (2014) — a well-reported, filmable narrative with named people, and Macy is also the author of Dopesick, which links the two grievance streams in one body of work.
• Huntington, West Virginia and the Appalachian opioid corridor — see Macy's Dopesick (2018) and the Hulu series; and the DEA ARCOS data showing pill volumes per capita.
• Kermit, West Virginia — population ~400; pharmacies received millions of hydrocodone pills over a few years per the ARCOS data. Verify the exact town/figure from the Washington Post database before use; the "town of 400 got 9 million pills" formulation circulates in several variants and should be pinned to the Post's own numbers.

9.2 Named human sources available on the record

• Neil Barofsky — SIGTARP, author of Bailout, on-record about HAMP's design failures and the "foam the runway" account.
• Elizabeth Warren — chaired the Congressional Oversight Panel for TARP; her contemporaneous COP reports are primary sources and she is a Democrat criticizing a Democratic administration's program, which is exactly the cross-cutting credibility the essay wants.
• Anne Case and Angus Deaton — extensive on-camera interviews; Deaton is a Nobel laureate and an unusually quotable critic of his own profession.
• David Autor — MIT, on-camera repeatedly, and on record saying tariffs won't fix what he documented. Getting him saying that in his own words, on tape, is the single highest-value clip in the domain.
• Beth Macy — reporter with named subjects in Martinsville and Roanoke.
• Sherrod Brown — the Ohio Democrat who voted against NAFTA and PNTR and spent thirty years saying this would happen. He is the living refutation of "nobody warned you," and he lost his seat in 2024, which is its own bitter beat.

9.3 Vivid details worth building beats around

• The trademark was signed six days after Romney lost. Not the movement, not the platform — the trademark.
• Trump noticed Reagan "didn't trademark it" while claiming not to have known Reagan used it.
• A quarter of the Russian ads were never seen by anyone, and more than half the impressions landed after Election Day.
• 109 of the 206 pivot counties had lower turnout in 2016 than in 2012. The realignment that broke American politics was, in most of those places, people giving up.
• HAMP: $75 billion promised, $10.2 billion spent, 70% of applicants denied.
• The Tea Party started as a televised rant against helping homeowners — five months after the banks were rescued.
• The paper that proved the China shock never healed was published in 2021. The paper that proved the mortality reversal was published in December 2015 — while Trump was already on the trail. The data arrived after the anger did. It didn't arrive after the pain.
• The crowd at the announcement of the great populist movement was partly hired at $50 a head.

10. Visual & B-Roll Research (Domain-Specific)

10.1 Archival footage to source

Legal note carried over from the foundation document and worth repeating here: House/Senate floor footage is public domain (exceptions: State of the Union, joint sessions, Speaker elections). C-SPAN-produced content — committee hearings, Washington Journal, campaign event coverage — is copyrighted and C-SPAN's terms require a license for monetized use. The escalator announcement and the Cohen testimony are both in the copyrighted category. Get legal review; do not assume fair use covers monetized long-form.

10.2 Data visualizations to build (all from named public datasets)

1. Manufacturing employment by county, 1990–2019, animated. Data: BLS Quarterly Census of Employment and Wages (QCEW), county-level NAICS 31-33 — https://www.bls.gov/cew/ ; or Census County Business Patterns — https://www.census.gov/programs-surveys/cbp.html Treatment: a choropleth that drains color out of the eastern half of the country between 2000 and 2011. Jon Bois-style "scroll to feel time" — let the viewer sit in the years.

2. Trade exposure vs. 2012→2016 vote swing, by commuting zone. Data: David Dorn's public China-shock data files — https://www.ddorn.net/data.htm (the ADH commuting-zone import-exposure measures are downloadable) crossed with MIT Election Data + Science Lab county returns — https://electionlab.mit.edu/data This is the scatterplot that is Autor/Dorn/Hanson/Majlesi (2020). Highlight the 206 pivot counties in a second color.

3. Deaths of despair, 1990–2024, by education group — including the 2023–24 reversal. Data: CDC WONDER (https://wonder.cdc.gov/) multiple-cause-of-death, ICD-10 codes for drug poisoning (X40–X44, X60–X64, X85, Y10–Y14), alcoholic liver disease (K70), and suicide (U03, X60–X84, Y87.0); plus NCHS provisional overdose counts (https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm). Build the chart through 2024, including the drop. A chart that stops in 2019 is a chart that will be used against the essay.

4. Opioid shipments vs. 2016 vote share, county-level. Data: the Washington Post's released DEA ARCOS database — https://www.washingtonpost.com/graphics/2019/investigations/dea-pain-pill-database/ (the Post published the pill-by-pill data for 2006–2014 and later extended it; the underlying files are downloadable from https://github.com/wpinvestigative/arcos-api) crossed with MIT Election Lab county returns. Two-panel design: the pills map and the swing map side by side, then dissolve one into the other. Then — and this is the honest move — a third panel showing the correlation shrinking from 18% to 7% of variance once you control for income, education, unemployment and disability (Goodwin et al., JAMA Netw Open 2018). Show the correlation, then show yourself deflating it. That single sequence would do more for the essay's credibility than any amount of narration.

5. The 206 pivot counties, mapped. Data: Ballotpedia's list (https://ballotpedia.org/List_of_Pivot_Counties_-_the_206_counties_that_voted_Obama-Obama-Trump) + county shapefiles from the Census TIGER/Line files. Add a turnout layer: shade the 109 counties where turnout fell. The story most people don't know is in that second layer.

6. TARP vs. HAMP, as a single bar. Data: Treasury TARP program tracker (https://home.treasury.gov/data/troubled-assets-relief-program) and the MHA/HAMP program reports (https://home.treasury.gov/data/troubled-assets-relief-program/housing/mha/hamp). Design: $700B authorized / $475B disbursed for the financial system next to $75B authorized / $10.2B spent for homeowners. Then the third bar: zero — senior bank executives imprisoned. The visual argument makes itself; the narration should get out of the way.

7. Prime-age male labor force participation, 1948–2026. Data: FRED series LNS11300060 — https://fred.stlouisfed.org/series/LNS11300060. One line, seventy years, no annotation until the end. Overlay the CEA's June 2016 report cover to make the point that the outgoing administration had already published this.

8. Piketty's education-gradient reversal. Data: WID.world, Piketty 2018 WP — https://wid.world/ — the charts showing left-party vote share by education decile flipping sign between 1948 and 2017 in France, the UK and the US. These charts already exist and are reproducible from published data. Three small multiples, three countries, same reversal — this is the strongest visual argument in the domain that the abandonment was structural and not an American accident.

9. Rural hospital closures, cumulative, 2005–2026. Data: UNC Sheps Center — https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/ . A dot map with a running counter to 197. Optional second layer: Medicaid-expansion status by state.

10. Life expectancy by county, high vs. low. Data: IHME US Health Map (https://www.healthdata.org/research-analysis/health-by-location/united-states) and County Health Rankings (https://www.countyhealthrankings.org/). The 20-year gap between the best and worst counties, shown as two human lifespans side by side.

11. The IRA in proportion. Data: SSCI Vol. 2 figures. $100,000 of IRA Facebook ads against $1.4 billion in 2016 social digital ad spend. One pixel against a screen. This visual is the anti-overclaiming device and it should be built, because it is the moment the essay earns the right to be believed by a skeptical viewer.

10.3 Graphics & diagrams

• A split-screen timeline: "what was happening to them" vs. "what was being built." Left column 2008–2015: foreclosures, plant closures, overdose deaths, hospital closures. Right column: Nov 12 2012 trademark signature, Nov 2012 domain registrations, the Apprentice seasons, the birther tour. Two clocks running at once.
• The USPTO document itself, on screen. Pull the actual filing from TSDR. The signature and the date are the beat; let the viewer read it.
• A "splice" diagram for the escalator speech: three grievances named in one paragraph — China (real, documented), the death of the American dream (real, documented), Mexican immigrants (the substitution). Show the first two connected by evidence and the third connected by nothing.
• The Piketty "two elites, no representation" diagram — a simple 2x2 of education and income against party, with the empty quadrant highlighted. This is the single clearest way to explain why 2016 was possible.
• The Tea Party → MAGA loop: grassroots anger → professional capture → monetization → delivery to donor-class ends → grassroots anger. Drawn once for the Tea Party in Act I, then re-drawn identically for MAGA later. If the essay uses the enshittification frame, this is the frame's diagram.

10.4 Iconic imagery

• The escalator descent, June 16, 2015 — the most-reproduced image of the entire story and the essay's obligatory opening rhyme.
• Reagan in 1980 wearing a "Let's Make America Great Again" button — the photograph exists and is the visual proof of the borrowing.
• The red hat, first appearances 2015 — worth tracking down the earliest documented photograph, since the hat's transformation from campaign merch to identity object is the enshittification story's central object.
• Lehman Brothers employees carrying boxes, Sept 15, 2008 — the shorthand image of the crisis. Getty/AP.
• Empty Lordstown parking lot, 2019.
• Rick Santelli on the CBOT floor, Feb 19, 2009 — traders cheering a rant against helping homeowners. That image is the thesis.
• Case & Deaton's original 1999–2013 mortality chart — the one line going up while every other rich country's goes down. It is already an icon in policy circles and it reproduces cleanly on screen.

11. Corrections and Cautions Against research-foundation.md

The provided foundation document is a good map of this domain. Five items in it need correction, softening, or added caveat before they reach the script.

One thing the foundation gets exactly right and that should be preserved verbatim in spirit: the "crucial honesty nuance" that the authors who documented the China shock reject protectionism as the remedy. That is the intellectual center of gravity of this whole domain.

12. Source Inventory

Primary sources / government documents

• Senate Select Committee on Intelligence, Russian Active Measures Campaigns and Interference in the 2016 U.S. Election, Volume 2: Russia's Use of Social Media — https://www.intelligence.senate.gov/sites/default/files/documents/Report_Volume2.pdf (full text extracted and verified for this brief)
• Mueller, Report On The Investigation Into Russian Interference In The 2016 Presidential Election, Vol. I — https://www.justice.gov/archives/sco/file/1373816/download
• United States v. Internet Research Agency, et al., Case 1:18-cr-00032-DLF (D.D.C., Feb. 16, 2018)
• ODNI, Assessing Russian Activities and Intentions in Recent US Elections (ICA), Jan 6, 2017 — https://www.dni.gov/files/documents/ICA_2017_01.pdf
• House Oversight Committee, Hearing with Michael Cohen, Feb 27, 2019 — https://www.congress.gov/116/chrg/CHRG-116hhrg35230/CHRG-116hhrg35230.pdf ; opening statement text https://edition.cnn.com/2019/02/27/politics/cohen-testimony-read/
• USPTO TSDR — "Make America Great Again" service mark record — https://tsdr.uspto.gov/ (NOT YET PULLED — see gaps)
• US Treasury, TARP program data — https://home.treasury.gov/data/troubled-assets-relief-program
• US Treasury, HAMP / Making Home Affordable — https://home.treasury.gov/data/troubled-assets-relief-program/housing/mha/hamp ; March 2014 MHA report — https://home.treasury.gov/system/files/initiatives/financial-stability/reports/Documents/March%202014%20MHA%20Report%20Final.pdf
• Council of Economic Advisers, The Long-Term Decline in Prime-Age Male Labor Force Participation, June 2016 — https://obamawhitehouse.archives.gov/sites/default/files/page/files/20160620_cea_primeage_male_lfp.pdf
• Trump campaign announcement transcript, June 16, 2015 — https://time.com/3923128/donald-trump-announcement-speech/

Peer-reviewed research

• Autor, Dorn & Hanson, "The China Syndrome," AER 2013 — https://www.nber.org/papers/w18054
• Acemoglu, Autor, Dorn, Hanson & Price, "Import Competition and the Great US Employment Sag of the 2000s," JOLE 2016 — https://www.nber.org/papers/w20395
• Autor, Dorn & Hanson, "The China Shock," Annual Review of Economics 2016 — https://www.nber.org/papers/w21906
• Autor, Dorn & Hanson, "On the Persistence of the China Shock," NBER WP 29401 / BPEA 2021 — https://www.nber.org/papers/w29401 ; https://www.brookings.edu/articles/on-the-persistence-of-the-china-shock/
• Autor, Dorn, Hanson & Majlesi, "Importing Political Polarization?", AER 2020 — https://www.nber.org/papers/w22637
• Caliendo, Dvorkin & Parro, "Trade and Labor Market Dynamics," Econometrica 2019
• Case & Deaton, "Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century," PNAS 2015 — https://www.pnas.org/doi/10.1073/pnas.1518393112
• Case & Deaton, comment on Ruhm, Jan 2018 — https://www.princeton.edu/~accase/downloads/Case_and_Deaton_Comment_on_CJRuhm_Jan_2018.pdf
• Novosad, Rafkin & Asher, "Mortality Change Among Less Educated Americans" — https://paulnovosad.com/pdf/novosad-rafkin-asher-mortality.pdf
• Chetty et al., "The Association Between Income and Life Expectancy in the United States, 2001–2014," JAMA 2016 — https://jamanetwork.com/journals/jama/fullarticle/2513561
• Goodwin et al., "Association of Chronic Opioid Use With Presidential Voting Patterns in US Counties in 2016," JAMA Netw Open 2018 — https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2685627
• Mutz, "Status threat, not economic hardship, explains the 2016 presidential vote," PNAS 2018 — https://www.pnas.org/doi/10.1073/pnas.1718155115
• Morgan, "Status Threat, Material Interests, and the 2016 Presidential Vote," Socius 2018 — https://journals.sagepub.com/doi/full/10.1177/2378023118788217
• Mutz, "Response to Morgan," Socius 2018 — https://journals.sagepub.com/doi/10.1177/2378023118808619
• Morgan, "Correct Interpretations of Fixed-effects Models... A Response to Mutz," Socius 2018 — https://journals.sagepub.com/doi/full/10.1177/2378023118811502
• Eady, Paskhalis, Zilinsky, Bonneau, Nagler & Tucker, Nature Communications 14:62 (2023) — https://www.nature.com/articles/s41467-022-35576-9
• Piketty, "Brahmin Left vs Merchant Right," WID.world WP 2018/7 — https://wid.world/

Books

• Case & Deaton, Deaths of Despair and the Future of Capitalism (Princeton, 2020) — https://press.princeton.edu/books/hardcover/9780691190785/deaths-of-despair-and-the-future-of-capitalism
• Sides, Tesler & Vavreck, Identity Crisis (Princeton, 2018) — https://press.princeton.edu/books/hardcover/9780691174198/identity-crisis
• Tesler, Post-Racial or Most-Racial? (Chicago, 2016)
• Frank, Listen, Liberal (Metropolitan, 2016)
• Skocpol & Williamson, The Tea Party and the Remaking of Republican Conservatism (Oxford, 2012)
• al-Gharbi, We Have Never Been Woke (Princeton, 2024)
• Macy, Factory Man (2014) and Dopesick (2018)
• Barofsky, Bailout (2012)
• Eberstadt, Men Without Work (AEI, 2016)
• Rid, Active Measures (2020)
• Wolff, Fire and Fury (2018) — reliability caveat
• Lewandowski & Bossie, Let Trump Be Trump (2017) — self-interested

Data sources

• BLS QCEW — https://www.bls.gov/cew/
• Census County Business Patterns — https://www.census.gov/programs-surveys/cbp.html
• David Dorn's China-shock data files — https://www.ddorn.net/data.htm
• MIT Election Data + Science Lab — https://electionlab.mit.edu/data
• CDC WONDER — https://wonder.cdc.gov/ ; NCHS provisional overdose data — https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm
• Washington Post DEA ARCOS database — https://www.washingtonpost.com/graphics/2019/investigations/dea-pain-pill-database/ ; API/files — https://github.com/wpinvestigative/arcos-api
• UNC Sheps Center rural hospital closures — https://www.shepscenter.unc.edu/programs-projects/rural-health/rural-hospital-closures/
• FRED LNS11300060 (prime-age male LFPR), MEHOINUSA672N (real median household income) — https://fred.stlouisfed.org/
• IHME US Health Map — https://www.healthdata.org/research-analysis/health-by-location/united-states
• Ballotpedia Pivot Counties — https://ballotpedia.org/Pivot_Counties (+ demographics, turnout, list, and 2020 sub-pages)
• Voter Study Group / Democracy Fund — https://www.voterstudygroup.org/publication/the-five-types-of-trump-voters
• EPI productivity–pay gap — https://www.epi.org/productivity-pay-gap/

Opposition / counterargument sources

• Cato Institute, "The 'China Shock' Demystified" — https://www.cato.org/publications/china-shock
• Yglesias, "The 'Deaths of Despair' narrative is wrong" — https://www.slowboring.com/p/the-deaths-of-despair-narrative-is
• Gelman, Statistical Modeling blog on Mutz — https://statmodeling.stat.columbia.edu/2018/05/14/status-threat-explain-2016-presidential-vote/ and .../2018/07/01/...
• Forrest, "What's Wrong with the 'Great Awokening'?" — https://publicseminar.org/essays/whats-wrong-with-the-great-awokening/
• The Nation, "What Political Scientists Get Wrong About 2016" — https://www.thenation.com/article/archive/identity-crisis-and-the-roots-of-2016-book-review/
• Dissent, "Beyond the Backlash" — https://www.dissentmagazine.org/article/beyond-the-backlash/
• Noah, "Popularism v. Deliverism" — https://timothynoah.substack.com/p/popularism-v-deliverism
• The New Republic, "Doing Popular Things Won't Save the Democratic Party" — https://newrepublic.com/article/164144/popular-things-wont-save-democratic-party
• al-Gharbi, "The 'Great Awokening' Is Winding Down" — https://musaalgharbi.com/2023/02/08/great-awokening-ending/
• Rakoff, "The Financial Crisis: Why Have No High-Level Executives Been Prosecuted?", NYRB 2014 — https://www.nybooks.com/articles/2014/01/09/financial-crisis-why-no-executive-prosecutions/

13. Research Gaps — What Still Needs Doing

Must be closed before script lock:

1. The USPTO record has not been pulled. Get the serial number, filing date, signature date, registration number, registration date, and owner of record from https://tsdr.uspto.gov/ for both "Make America Great Again" and "Keep America Great." The Nov 19, 2012 filing date and the "before inauguration" KAG claim both rest on secondary sources right now. This is a public record and it should be on screen.
2. The 59.3% figure. Either locate it in the body of NBER WP 29401 with its exact referent, or strike it. Currently unsupported as the foundation states it.
3. Ballotpedia's 2020 pivot-county outcomes (retained vs. boomerang). Page fetch failed at research time.
4. 2016 exit poll income crosstabs. Needed to keep the essay from the "MAGA is the poor" error. Roper/CNN exit poll archives.
5. NAFTA and PNTR roll-call tallies — verify against clerk.house.gov and senate.gov rather than memory.
6. The foreclosure total. Pin to one series (CoreLogic completed foreclosures, or RealtyTrac filings, or Fed/Census) and state which. The "6 million / 10 million homes" figures circulate loosely.
7. The "foam the runway" quote. Barofsky's account, Geithner's denial, and the exact wording. This is a strong beat and it will be attacked.
8. The paid-actors story from June 16, 2015 — get the original Hollywood Reporter piece and the follow-up on whether Extra Mile Casting was paid.
9. The Phoenix rally attendance figure (July 11, 2015) and the sourcing for "the campaign realized here."
10. Kermit, WV pill numbers — pull from the Post's ARCOS database directly, not from circulating summaries.
11. Current CDC final (not provisional) 2024 mortality figures for the deaths-of-despair reversal, and any 2025 data now available.

Known limits of this brief:

• No first-person testimony has been sourced. The essay's strongest emotional device — a named Obama-to-Trump voter who will speak on camera — does not exist in this research yet. This is the biggest single gap. Candidates: the Voter Study Group panel is anonymized; local journalism in Trumbull County (OH), Erie County (PA), Macomb County (MI) and Howard County (IA, a pivot county) is the place to look. Consider whether the production can conduct original interviews.
• The Autor-on-camera clip has not been located. Getting him saying "tariffs won't fix this" in his own voice is worth a dedicated search of C-SPAN, Brookings event video, EconTalk, and Planet Money archives.
• Rate-limited. Roughly ten web searches were available for this domain. Several claims above are flagged VERIFY rather than confirmed; each flag is deliberate and each is a specific, closeable task.

A note on what this domain cannot deliver, and shouldn't pretend to: The evidence establishes that the pain was real, widespread, geographically concentrated, and bipartisan in its causes and its neglect. It does not establish that economic pain caused the 2016 vote at the individual level — the best individual-level data point at racial and status attitudes, and the best place-level data point at trade shocks, and those are not the same finding. The essay's steelman is strongest when it makes the narrower, unimpeachable claim: these people were not imagining it, nobody was coming, and a man with a registered trademark showed up.