The Narrative Networks Podcast

Hey! I'd love to hear your thoughts, send me a voice note.

Michael Knowles said the first part of the Trump administration delivered "$17 trillion, new investment in America." We rate that false. Truth score 0.1, confidence 0.9.

The White House's own list of investment announcements totalled $8.8 trillion the month that number was first said publicly, and BEA measures $232.2 billion in actual new foreign direct investment for all of 2025.

The Michael Knowles Show, Ep. 2043, 21 August 2026, at 36:09. The line arrives as an appositive inside a sentence about midterm enthusiasm — no source, no date, no qualifier.

The steelman is real. Announcement volume did surge, and the White House list carries named, dated corporate commitments: Apple $600B, Meta $600B, Stargate $500B, NVIDIA $500B. TSMC's Arizona project has grown to $265 billion, the largest greenfield foreign direct investment project in US history. 2025 FDI rose 49.5% over 2024, the first rebound after three years of decline. Commerce Secretary Howard Lutnick told Senate appropriators in April 2026 that twenty trade deals total $9.94 trillion in commitments. "Massive new investment in America" is directionally defensible.

The figure is not. When Trump first said "$17 trillion" in September 2025, the administration's own tally was $8.8 trillion; by late November 2025 it was about $9.6 trillion while he claimed $18-22 trillion. Bloomberg Economics found only about $7 trillion of that list qualifies as real investment pledges. Reviewers found India's $500B bilateral trade goal, Qatar's $1.2T "economic exchange," Japanese loans and guarantees, and JERA's $200B LNG purchase — a Japanese import — counted as investment in America. Over $250 billion of the listed pledges predate January 2025. Meanwhile Census data show manufacturing construction spending peaked in Q3 2024 and was roughly 32% below that peak by June 2026.

What we don't know: how much of the ~$9.6 trillion will ever be spent (there is no public tracker, no completion reporting, no auditor), what the February 2026 Supreme Court tariff ruling did to pledges negotiated under tariff threat, and what Knowles's source was — he cited none, and we won't assume.

Hey! I'd love to hear your thoughts, send me a voice note.

Michael Knowles said Democrats wanted mass migration to manufacture a permanent electoral majority and have "mostly been proven right." We rated it false — truth score 0.15, confidence 0.82.

The prediction was real. The program was not — and in 2024 the naturalized-immigrant vote swung to Trump.

On Ep. 2044 of The Michael Knowles Show (24 August 2026), at 29:27, Knowles packed two claims into one sentence: a motive claim (Democrats pursued migration in order to build a majority) and an outcome claim (it worked). The line sits inside a defense of J.D. Vance, whose 21 August speech in Middletown, Ohio attacked the Wall Street Journal over its 1984 "There shall be open borders" editorial. Note that Vance's target was capital and offshoring; the Democrats-manufacturing-voters clause is Knowles's addition.

The steelman is genuine. Judis and Teixeira's 2002 The Emerging Democratic Majority did predict a durable coalition built on demographic change. In April 2013 the Center for American Progress published Teixeira and Halpin explicitly forecasting electoral gains from that change. And restrictionist analysts at CIS documented real effects — immigrant partisan lean, marginal counties turning safely Democratic, apportionment shifts. Those are defensible claims about effects.

The causal half is where it fails. The major expansions were bipartisan and employer-driven, and the mechanism doesn't run: unauthorized immigrants cannot vote in federal elections and cannot naturalize without first obtaining lawful permanent residency; even for legal immigrants, LPR status plus five years plus processing is the floor, and one study found 30% of LPRs remain LPRs for over 20 years. Democratic administrations ran the largest removal operations on record. And "proven right" collides with 2024: a Republican trifecta, Trump winning the popular vote by about 1.5 points, naturalized immigrants swinging from Biden +27 to Trump +1 by Cato's estimate, and Trump taking about 46% of the Latino vote. Both authors of the 2002 thesis have abandoned it.

What we do not claim: motive is genuinely hard to adjudicate. Some Democratic-aligned analysts clearly argued electoral benefit from demographic change. Whether that drove policy is not established, and one election cycle is not a trend.

Hey! I'd love to hear your thoughts, send me a voice note.

Matt Walsh said every claimed lynching "turns out to be a completely fake story." We rated it false — truth score 0.05, confidence 0.93.

The cases he's generalizing from are real. The number he needs is 100%. The number the evidence supports is closer to 2%.

On Ep. 1823 of The Matt Walsh Show (17 August 2026), responding to Ibram X. Kendi's essay "The Media Lynched Jason Arday," Walsh made two moves. The first — that calling press scrutiny a lynching is a category error — is a critique shared by writers at spiked and Reason. The second converts that critique into a universal claim about the historical and legal record, and uses it as grounds to reopen documented lynchings as possibly justified. We rated the second move.

The steelman is genuine: Jussie Smollett staged a 2019 attack and was convicted of lying to police (conviction later overturned on procedural grounds, which the special prosecutor said had nothing to do with innocence). The FBI determined the Bubba Wallace garage rope was a pre-existing door pull. Wilfred Reilly documented 409 debunked allegations, 2010-2017. Fabricated incidents are real.

The record they're set against: roughly 6,500 racial-terror lynchings documented by the Equal Justice Initiative from contemporaneous newspapers, court records and coroners' reports. Michael Donald's killers — one executed, one imprisoned for life. James Byrd Jr. — three convicted, two executed. Ahmaud Arbery — federal hate-crime convictions upheld on appeal. Six Rankin County, Mississippi deputies sentenced to a combined ~132 years in 2024. The Emmett Till Antilynching Act passed 422-3 in the House and unanimously in the Senate. And in 1892, Ida B. Wells examined 728 lynchings specifically to refute the claim that the victims had it coming.

Hey! I'd love to hear your thoughts, send me a voice note.

Matt Walsh told his audience that anonymous surveys show more than 80% of experts agree genes play a non-trivial role in average race differences in intelligence.

One survey does. It is 86 answers out of roughly 1,237 researchers contacted - and the man who compiled the survey table reports the real range as 14% to 84%, average 45%. Verdict: mostly-false, truth score 0.25, confidence 0.85.

The claim: The Matt Walsh Show, Ep. 1823, "Cambridge's DEI Superstar Was A Serial Liar, Plagiarist & Hate Hoaxer," published 17 August 2026, at 56:42.

The strongest version of Walsh's side is real and we give it room. Rindermann, Becker and Coyle (2020, Intelligence) found roughly 84% of the 86 researchers who answered their US Black-White IQ gap item attributed at least some of the gap to genes, mean estimate about 47-49% genetic. A 2016 companion paper from the same fielding reports about 90% allowing genes some influence on cross-national differences. Kruger et al. (2022) found 74.4% among 581 evolutionary scientists. Among intelligence researchers and evolutionary behavioural scientists, majorities do accept some genetic contribution, and that is not the impression most of the public has been given.

What breaks is the word "surveys" and the words "experts in the field." The 84% comes from 86 answers - under 8% of the researchers contacted on that item - drawn largely from International Society for Intelligence Research circles, with fewer than half of respondents who named a field working in intelligence-related disciplines and only four identifying as geneticists. Every other survey of a relevant expert population lands far lower: 28% (APA members, 1973), 45% (Snyderman and Rothman, 661 scholars, 1987), 26.4% (Society of Experimental Social Psychology, 2017), 14% (US academic anthropologists, 2019). That table is Noah Carl's own compilation, and Carl's own headline is 14% to 84%, average 45%. Walsh's number is the ceiling of his own side's range, presented as its centre.

We also cover what we cannot resolve: which survey he meant, whether "experts in the field" was intended narrowly, the direction of non-response bias, and whether the six surveys are commensurable at all. Expert opinion polling is not evidence about genetics either way; admixture and polygenic-score work currently offers no positive support for a genetic component to the gap, but "no support found so far" is not "shown to be zero." On the Cambridge cases: Liverpool John Moores University did not uphold the plagiarism allegations against Jason Arday, and Cambridge's investigation into his qualifications was open at the time of this episode.

Hey! I'd love to hear your thoughts, send me a voice note.

A guest on The Benny Show said elections are stolen "every election in most of the Democratic states, if not all." We rate it false — truth score 0.08, confidence 0.88.

The same segment says you can't prove it "because of the way they've done it. They set it up that way." Here is what the record actually holds.

The claim comes at 38:28 in "How Communism Will Win…By Destroying America From The Inside | This Is Their Plan...," published August 18, 2026. We steelman it before we land the verdict, because the speaker is not inventing his material. Bridgeport, Connecticut is real: a judge voided a 2023 Democratic mayoral primary after drop-box video showed 1,253 absentee ballots deposited where only about 420 people appeared, and five people were criminally charged in February 2025. New Jersey disclosed roughly 6,600 noncitizen registrations from a Motor Vehicle Commission data transfer. The 2005 Baker-Carter Commission did conclude absentee ballots are the form of voting most vulnerable to fraud.

Then the arithmetic. The Heritage Foundation's own database — the best right-of-center compilation that exists — holds about 1,500 proven cases spanning decades, including roughly 10 in-person impersonation cases across billions of ballots. Using that database, Brookings computes mail-ballot fraud at about 0.00005% of ballots cast. Bill Barr's DOJ found no fraud on a scale that could have changed an outcome. A Republican-led Michigan Senate committee found no systemic fraud in a Democratic-governed swing state. And Bridgeport was detected on video, litigated, overturned and prosecuted by the system — the opposite of a system that hides theft.

We also take apart the illustration Benny Johnson supplies. Milwaukee did have a real error: five flash drives were uploaded with audit logs instead of results, delaying about 28,000 absentee ballots by roughly 90 minutes. No votes were lost. David Crowley won 39.8% to 39.3% — about 3,000 votes, 0.4 points — and Francesca Hong, who had led public polling for weeks, conceded and declined an available recount. Nobody was "down by 30 points"; the double-digit lead belonged to the candidate who lost.

One more thing our archive shows: four tracked shows covered the Milwaukee flash-drive error within twenty-four hours, and Bridgeport — the single prosecuted case that genuinely supports the speaker's position — has never been mentioned on any show we track.

What we do not claim to know: the archived transcript identifies the guest only as a former attorney general, and we have not confirmed who he is. The precise final-poll margin in the Wisconsin primary remains unpinned. The New Jersey and Pennsylvania figures are counts of registrations, not ballots cast. And every fraud estimate measures detected fraud — the record constrains the plausible size of what is missed, it does not reduce it to a proven zero.

Hey! I'd love to hear your thoughts, send me a voice note.

Michael Knowles told his audience "we know that the transvestite community is particularly violent." We rate it FALSE — truth score 0.1, confidence 0.82.

The aside came one day after a real federal sentencing, and that is exactly how this move works: one prosecuted case, then a property assigned to millions of people.

On 19 August 2026, in Ep. 2041 of The Michael Knowles Show, the line arrives as a subordinate clause during a segment about Ryan Michael English — who goes by Riley Jane — sentenced to 73 months in federal prison for a January 2025 plot to kill Treasury Secretary Scott Bessent at the U.S. Capitol.

We steelman the other side before landing. John Lott's Crime Prevention Research Center argues that in 2024 transgender individuals committed active-shooter attacks at 12–16 times their population share, resting on two real arguments: that the Gun Violence Archive's "mass shooting" denominator mixes in gang disputes and robberies, and that the trans-identifying population skews young, exactly where mass shooters come from. The age-composition point is a legitimate criticism the fact-checks generally don't address head-on.

It still doesn't carry the claim. "Over-represented among active shooters" and "particularly violent" are different propositions; the second requires elevated general violent offending, which no dataset shows. The base is four to seven cases across eight years, wide enough that one case moves the multiplier substantially. And the most-cited study on trans criminality, Dhejne 2011, found no statistical difference in conviction rates for the later-transitioning cohort — its author has said publicly that people making claims about trans criminality are misrepresenting her findings.

What is measured points the opposite way: 86.2 violent victimizations per 1,000 transgender people versus 21.7 per 1,000 cisgender people, from the same national survey the government uses to measure crime.

What we still don't know, and say so in the episode: there is no U.S. perpetration dataset recording offender gender identity, no published age-and-sex-standardized denominator, and no rate for assassination plots by any demographic group at all.

Hey! I'd love to hear your thoughts, send me a voice note.

Timcast IRL, 18 August 2026, at 31:38: "far more blood was shed in like literally the 10 years of decolonization in Africa than like the 100s of years of colonial rule."

We rate that false — truth score 0.10, confidence 0.75. The ten-year window the claim names holds roughly 120,000 deaths; the events summoned to fill it span 1959 to 1994.

This is a magnitude claim made at the end of a five-and-a-half-minute block, and eleven seconds later the show has moved on to something else. No number is named. No source is given.

We steelman it first. Post-independence Africa's cumulative violent toll is real and under-discussed — 2 million of 3.9 million global conflict deaths from 1989 to 2024 were in Africa — and the colonial figures are genuinely contested, with Sanderson revising Congo Free State losses down to 1-5 million against Hochschild's 10 million. The broad version of the argument is not stupid. The version actually said is a different claim, and it fails three ways at once: the window (Biafra 1967-70 and the 1994 genocide are outside the decade it names), the denominator (formal European rule over most of Africa ran about 70-80 years, not hundreds), and the numerator (the Congo Free State floor alone, 1.2 million in one colony, is ten times the whole decade). The demography is the cleanest test: Africa's population sat flat near 140 million from 1850 to 1920, reached about 280 million by 1960, and sub-Saharan life expectancy rose from 43-45 years in 1960 to 62.6 in 2023.

We do not name the speaker. Our transcript diarizes the line without name-level attribution, and we will not put words in a named mouth.

Sources

https://en.wikipedia.org/wiki/Atrocities_in_the_Congo_Free_State

https://en.wikipedia.org/wiki/Herero_and_Nama_genocide

https://blackpast.org/global-african-history/congo-civil-war-1960-1964/

https://www.britannica.com/topic/Hutu-revolution

https://www.britannica.com/topic/Nigerian-civil-war

https://www.britannica.com/event/Rwanda-genocide-of-1994

https://www.aehnetwork.org/wp-content/uploads/2016/02/ReconstructingAfricanPopulation.COMPLETE.pdf

https://data.worldbank.org/indicator/SP.DYN.LE00.IN?locations=ZG

https://ourworldindata.org/conflict-deaths-breakdown

https://archive.org/details/la-demographie-du-congo-sous-la-colonisation-belge_202201

Corrections policy: if we get something wrong, tell us. We correct it on air and in the episode brief, and we say what changed.

Hey! I'd love to hear your thoughts, send me a voice note.

A guest on The Benny Show said transgender people are "mentally ill," "not born that way," and that this is "what got Charlie Kirk killed."

We rate it false — truth score 0.12, confidence 0.86. Here is the strongest version of that argument, and exactly where it breaks.

The claim lands at 54:22 in the 13 August 2026 episode "James O'Keefe Wears HIJAB, Goes Undercover to EXPOSE Mass Muslim Voter Fraud: They're Going to Jail." The transcript does not name the speaker, so we attribute it to an unnamed guest, not to host Benny Johnson.

Three assertions are stacked into one sentence: that trans people are mentally ill, that the identity is produced by trauma, and that this caused Charlie Kirk's death. The steelman is real — trans people do report higher rates of adverse childhood experiences and distress, the UK's Cass Review found the youth gender-medicine evidence base weak, and the etiology question is genuinely open in both directions. But the DSM-5 says gender non-conformity "is not in itself a mental disorder," ICD-11 moved gender incongruence out of the mental-disorders chapter, twin registries point away from a purely experiential origin, and no study has shown trauma to produce a transgender identity. Charging documents record Tyler Robinson's stated reason as Kirk's "hatred" and "evil"; Robinson is not transgender, and his partner Lance Twiggs told investigators Robinson "did not talk much about gender issues or LGBTQ+ rights."

We also cover the fifty-year history behind "something horrible happened," the 48 hours after 10 September 2025 that built the trans-shooter frame, and the archive finding we think is sharper than the debunk: across 95 logged Charlie Kirk claims, the motive slot has now been filled four mutually exclusive ways while the prosecutors' documented account sits verified at 0.95.

What we still don't know: Robinson's full motive — the trial is ongoing as of August 2026 — the direction of the trauma association, and whether Twiggs' gender identity figured in Robinson's thinking at all.

Hey! I'd love to hear your thoughts, send me a voice note.

In 2008, Al Franken beat Norm Coleman by 312 votes out of nearly 2.9 million. A Daily Wire roundtable called it a theft — and in the same breath declined to argue it.

We checked the claim: mostly false. Truth score 0.12, confidence 0.90.

On the 14 August 2026 "Friendly Fire" episode carried on The Michael Knowles Show feed, a speaker sent off Minnesota Senate candidate Michele Tafoya by noting the state had Republican senators "before the Democrats stole it," said "a dispute for another time," then asserted "they did totally steal that" anyway. Our diarization does not resolve which of the four hosts spoke, so we don't name one.

The strongest case for the claim is arithmetic: Minnesota Majority alleged at least 341 ineligible felons voted in two counties, and Byron York later escalated that to 1,099 statewide — both larger than 312. The administrative failures were real too: 133 ballots lost in Minneapolis and never found, roughly 150 alleged double-counts that a Republican board member called likely, and absentee rejection standards that varied county to county.

What the record shows: the canvassing board, with two Pawlenty-appointed justices, voted unanimously. The Minnesota Supreme Court affirmed 5-0 and rejected Coleman's Bush v. Gore equal-protection theory. Coleman conceded and Pawlenty signed the certificate. Coleman's public complaint was about counting, not fraud. The 933 wrongly rejected absentees (some accounts say 953 — we have not resolved which) split 52-48 for Franken. Of roughly 1,359 suspected ineligible felons referred to prosecutors, most were withdrawn, unsubstantiated or wrong; total convictions ran to 26 for felon voting and 12 for felon registration without voting.

What we still don't know: how many ineligible ballots were actually cast (26 is a floor, not a count), and how any of them broke — ballots are secret, so every version of this claim assumes its own conclusion. Whether a uniform statewide absentee standard would have produced a different winner is genuinely arguable. That is not the claim that was made.

Hey! I'd love to hear your thoughts, send me a voice note.

Ben Shapiro's show played a line from Ben Shapiro's movie: "three ways Islam will conquer the West." We rate it false. Truth score 0.1, confidence 0.85.

The cleric quotes behind it are real. The demographics are not — and "the wombs of our women" comes from a 1974 UN speech that never contained it.

On 12 August 2026, at 1:11:53 in Ep. 2484 ("Capitalism Is Under Attack. Here's Why."), The Ben Shapiro Show ran an audio roll of dialogue from Run Hide Fight: Infidels — a Daily Wire feature releasing 16 September 2026 that Shapiro himself produced. The line arrives with no rebuttal and no frame.

We steelman it first. Yusuf al-Qaradawi did tell an audience in Toledo, Ohio in 1995 that "we will conquer Europe, we will conquer America, not through the sword but through dawah," and repeated it in a 2003 fatwa and on Qatar TV in 2007. Gaddafi said much the same at Timbuktu in 2006. The 1991 "Explanatory Memorandum" does contain the phrase "civilization-jihadist process" (its reading is contested; the trial judge in 2009 declined to treat it as supporting evidence of a conspiracy). Europe's Muslim population genuinely is growing faster than its non-Muslim population.

The claim fails on the move from "some Islamists have said this" to "Islam will do this." Pew puts Muslims at 4.9% of Europe in 2016 (25.8 million) and projects 7.4% by 2050 with zero further migration, 11.2% at medium migration, 14.0% at high — one European in seven at the maximum. US projections run near 2.1%. Grant the Middle East Forum's undercount critique in full and 2050 moves from roughly one in fourteen to roughly one in six. The Muslim-to-non-Muslim fertility gap in Europe (2.6 vs 1.6) is projected to narrow to 0.7 by 2045–50, with age structure (median 30.4 vs 43.8) doing as much work as births. And the "wombs of our women" phrasing traces to a quotation long misattributed to Houari Boumédiène's 1974 UN address; the speech is public, concerns commodity prices, and contains nothing of the kind.

Open questions we have not closed: who speaks the line in the finished film — villain or protagonist — and whether Shapiro voiced or endorsed the proposition himself. Neither changes the verdict.