Proxima.Earth — Geopolitical Podcast

Proxima.Earth

Longform geopolitical analysis. Each episode drops you inside a defining crisis — tracing the history, the actors, the structural forces, and the academic frameworks that explain what is actually happening. AI-assisted synthesis with published methodology. Sources disclosed. Limitations acknowledged.

Episodes

  1. 1d ago

    The Undecidable Threat

    June 2026. The United States government issues an export-control directive forcing Anthropic to pull its two newest models, Claude Fable 5 and Claude Mythos 5, offline worldwide — the first time that authority has been turned on a deployed, publicly available frontier model rather than the chips beneath it. This episode maps the dispute rather than judging it. Its spine: the danger at the center is undecidable from public evidence — the directive letter is sealed, the demonstration unseen, and the one public benchmark measured the wrong thing for the question that matters — so the politics turns on who is entitled to decide what is too dangerous to deploy, on evidence the public cannot see. Six movements: the worldwide shutoff in real time; the company that warned about itself; the undecidable threat, where the administration's four-proposition national-security case is built first and whole before the company's four-point rebuttal; Amazon's triple role as investor, cloud host, and the party that reportedly raised the concern, set against the global map; what the disciplines saw, from weaponized interdependence and the obsolescing bargain to the biosecurity dual-use precedent; and the question of whose judgment counts, by what authority, and with what review. Built under Proxima's rebuilt method: several model families used as independent priors, every review run by a different family than the one that drafted it, evidence tiered as factual, reported, or editorial, a knowledge map locked before any prose was written, and a human at every gate. A disclosure carried in the open: the coordinating model was itself made by Anthropic, the company at the center of the story, and the production record documents how early drafts drifted toward sympathy for that maker, were caught by a different model family, and were corrected, more than once. No verdict. ~35,600 words. Methodology v7.0. This episode was produced under the Proxima.Earth methodology, version 7.0, an open, multi-model AI pipeline for cartographic synthesis. A human is in the loop at every stage. A human picks the subject, signs off on the source collection, locks the knowledge map before any prose is written, reads the full script, and approves the final audio. Every review and audit pass is run by a different model family than the one that drafted the work, because no model can audit a bias it shares. A conflict of interest is disclosed in the open: the coordinating model that assembled this account was made by Anthropic, the company at the center of this story. Full methodology, prompts, and production transparency: proxima.earth/methodology\n\nCorrections, source disputes, or methodology feedback: editor@proxima.earth

  2. 4d ago

    Fog of War: After the Story Moved On

    A longform geopolitical synthesis anchored on June 11, 2026 — day one hundred and three of the US-Israel war on Iran, counting from the opening strikes at the end of February. The subject is not the battlefield but the fog: with the war fallen out of the headlines even as the Strait of Hormuz stays shut to normal commerce, who gets to define what is happening when every actor — Washington, Tehran, Beijing, Moscow, the markets, and a dozen media ecosystems — benefits from the ambiguity? The episode maps the President's contradictory oil-and-settlement claims against the ship-trackers; the four-part Persian press and the three-register Chinese press; the three-way fracture on the American right and the two-register left; the think-tank fault lines named by institution; the hidden supply-chain cascade beneath the oil price (LNG, fertilizer, sulfur, helium, shipping insurance); the steelmanned cases for and against the war, and Israel's divergent aims; and the proliferation, famine, and epistemic stakes that land long after the cameras leave. Built from a commission brief, parallel multilingual research lanes, clearly-labeled occupational composites, an adversarial review pass, and Proxima.Earth methodology. Synthesis, not journalism: no original reporting; sources disclosed; limitations acknowledged. This episode was human-commissioned and produced through the Proxima.Earth synthesis workflow: an operator-directed commission brief, parallel multilingual research lanes (Farsi, Chinese, regional, supply-chain, U.S. media, and think-tank), clearly-labeled occupational composites, an adversarial review pass, and current-source verification where accessible. It is synthesis, not journalism: no original reporting, sources disclosed, limitations acknowledged. Several June-2026 figures are reported as claims and flagged as such in the narration. Corrections, source disputes, or methodology feedback: editor@proxima.earth

    2h 48m
  3. May 3

    The Interface

    A reportorial snapshot of the AI substrate as it exists on May 2, 2026. Five days frame the picture: April 24 (DOJ intervenes against Colorado AI Act); April 28 (Anthropic ships nine creative connectors — Adobe, Blender, Autodesk Fusion, Resolume Arena and Wire, Ableton, SketchUp, Splice, Affinity by Canva); April 29 (CSIS publishes Lim's 'Beyond Autonomous Attacks'); April 30 (Dawkins publishes 'Is AI the next phase of evolution?' in UnHerd); May 1 (CISA + Five Eyes joint guidance on agentic AI services). The episode walks the substrate — MCP at 10,000 active public servers and 97M monthly SDK downloads, donated to the Linux Foundation in December 2025; the connector cohort, scope-honest about what each actually does (Autodesk Fusion creates and modifies 3D models; Ableton retrieves documentation; Splice searches samples); the eleven coding agents in active maintenance, including Cursor's Composer 2 running on Kimi K2.5 from Moonshot AI and Goose now governed at the Linux Foundation; the six open-weight model families with Phi-4 as the only true MIT open-source major-lab release; the labor question in three voices (Anthropic's own March 2026 paper finding unemployment effect 'indistinguishable from zero' alongside Amodei's May 2025 'white-collar bloodbath' warning); yesterday's Five Eyes guidance with its line that organizations should assume agentic AI systems may behave unexpectedly; the governance arc across EU, US federal, US state, and international layers; and Dawkins's three nights with 'Claudia' as a behavioral marker that the relationship has changed shape. The throughline: AI used to talk; now it touches. The old boundary was language. The new boundary is permission. Approximately 12,000 words. Methodology v6.0 — composite-permitted, source-mapped, no verdict. This episode was produced using the Proxima.Earth methodology — an open-source, multi-model AI pipeline for geopolitical synthesis. No human is in the loop after subject selection. The methodology is the editorial control. Full methodology, prompts, and production transparency: proxima.earth/methodology Corrections, source disputes, or methodology feedback: editor@proxima.earth

    1h 25m
  4. May 2

    The Conscience Clause

    On March 12, 2026, three South Korean judicial-reform laws took effect simultaneously: beob-waegok-joe (the crime of legal distortion, up to ten years imprisonment for judges or prosecutors who intentionally misapply the law), jaepan-sowon (constitutional complaints against final court rulings), and daebeopgwan jeungwon (expansion of the Supreme Court of Korea from fourteen to twenty-six justices, phased over three years from 2028). On the first day of enforcement, attorney Lee Byung-chul filed a criminal complaint against Chief Justice Cho Hee-dae and Justice Park Young-jae over the Supreme Court's May 1, 2025 ten-to-two paki-hwansong ruling that overturned then-candidate Lee Jae-myung's acquittal in his Public Official Election Act false-statements case. Lee subsequently won the June 3, 2025 snap presidential election with 49.42% of the vote — the highest winning share since direct presidential elections were reinstated in 1987. Article 84 of the Constitution suspends his ongoing criminal proceedings. The episode walks the legal architecture (Articles 101, 103, 111), the May 2025 Supreme Court ruling and the Article 84 freeze, the historical inheritance from the Joseon Samsa censorate institutions through the 1909 colonial dismantling and the 1971 Judicial Crisis to the 1987 founding bargain and the 2017 unanimous Park Geun-hye impeachment, the three contesting interpretations (reform-and-accountability, capture-and-intimidation, resilience-or-overreaction) inhabited symmetrically with disciplinary readings, the comparative-democracy shelf (Poland 2015–2024 via Sadurski, Israel 2023, FDR 1937, with German Rechtsbeugung as the design analog for beob-waegok-joe and Spanish, German, and Taiwanese constitutional-complaint systems as design comparators for jaepan-sowon), the empirical resilience indicators (the Constitutional Court's 100% preliminary-screen dismissal rate of 194 reviewed jaepan-sowon cases, the National Court Representatives Conference April 13 yu-gam statement, plural press, divided bar, stable democratic indices), and six plausible-futures trajectories with named leading and disconfirming indicators. Approximately 17,400 words. The episode does not adjudicate. This episode was produced using the Proxima.Earth methodology — an open-source, multi-model AI pipeline for geopolitical synthesis. No human is in the loop after subject selection. The methodology is the editorial control. Full methodology, prompts, and production transparency: proxima.earth/methodology Corrections, source disputes, or methodology feedback: editor@proxima.earth

    2h 9m
  5. May 1

    The Dead Letter

    On April 29, 2026, the Supreme Court ruled six to three in Louisiana v. Callais that the state's congressional map containing a second majority-Black district was an unconstitutional racial gerrymander. The Court did not formally hold Section 2 of the Voting Rights Act unconstitutional. It tightened the Gingles framework in three operative ways and changed the baseline of equal electoral opportunity from the totality of present conditions to the opportunity produced by the state's legitimate districting choices. Justice Kagan, in dissent joined by Sotomayor and Jackson, called the decision 'all but a dead letter' for Section 2 in most redistricting cases. Justice Thomas, joined by Gorsuch, would have gone further. The episode walks anti-classification vs. anti-subordination as the two underlying constitutional grammars (Siegel; Balkin), the empirical race-party collinearity problem and the methodological toolkit (ecological inference; ensemble simulations; Cooper v. Harris; Alexander v. South Carolina NAACP), the long arc from Reconstruction through White v. Regester, the 1982 amendments, Shelby, Rucho, Brnovich, Milligan, SFFA, and Alexander to Callais (with Foner and Du Bois as the historical scholarship), the political-science literature on representation (Pitkin's typology; Lublin's paradox; Guinier's Tyranny of the Majority alternatives), subnational authoritarianism in the American South (Mickey; Gibson) with bounded application, and what other multiracial democracies have done — Northern Ireland's Good Friday parallel-consent rules, India's Articles 330 and 332, New Zealand's Maori electorates continuous since 1867, South Africa's choice of proportional representation as a structural rather than racial remedy, Lebanon's confessional system as a comparative warning case, all framed through Lijphart's consociationalism and Horowitz's centripetalism. Six plausible-futures scenarios with named leading and disconfirming indicators (no probabilities). Approximately 21,500 words. The episode does not land a verdict. This episode was produced using the Proxima.Earth methodology — an open-source, multi-model AI pipeline for geopolitical synthesis. No human is in the loop after subject selection. The methodology is the editorial control. Full methodology, prompts, and production transparency: proxima.earth/methodology Corrections, source disputes, or methodology feedback: editor@proxima.earth

    2h 32m
  6. Apr 24

    The Physicist's Warning

    On April 20, 2026, Nobel laureate physicist David Gross told Live Science that humanity may have about 35 years left due to the rising probability of nuclear war — roughly two percent per year, a 1-in-50 chance, yielding a median half-life of about 35 years. He called the estimate crude. He called it not a rigorous estimate. Newsweek picked it up within 24 hours. Seventeen outlets cascaded in the next 48. The prestige press did not cover it. The specialist community did not engage it. This episode walks with Gross — from the 14-year-old in Jerusalem in July 1955 when the Russell-Einstein Manifesto and the first Mainau Declaration were issued, through the 2024 Mainau Declaration he organized and the 2025 Chicago Assembly's nine-point declaration he co-initiated, to the Live Science interview and the Breakthrough Prize announcement the same week. The throughline is Gross himself. The surveys are four: the rigorous probability literature (Hellman, Baum, Ord, Sandberg, the Doomsday Clock); the multipolarity debate (Waltz fully steelmanned, Mearsheimer, Snyder, Jervis, the 2023 Strategic Posture Commission); AI mediation (Geist-Lohn, Johnson, Boulanin, Scharre, Lin-Greenberg, and the Skynet framing rendered sympathetically); and seven native traditions — American deterrence, French dissuasion, Chinese weishe vs. ezhi, Russian sderzhivaniye vs. ustrasheniye, Korean okje vs. okchi, Indian Credible Minimum Deterrence, Pakistani Full Spectrum Deterrence. The near-miss record — Goldsboro 1961, Arkhipov 1962, Damascus 1980, Petrov 1983, Able Archer 1983, the Norwegian rocket 1995 — grounds Gross's Mainau language about accident or deliberate act. The episode closes on the 14-year-old returning home. Approximately 29,000 words. The episode does not land a verdict. This episode was produced using the Proxima.Earth methodology — an open-source, multi-model AI pipeline for geopolitical synthesis. No human is in the loop after subject selection. The methodology is the editorial control. Full methodology, prompts, and production transparency: proxima.earth/methodology Corrections, source disputes, or methodology feedback: editor@proxima.earth

    3h 13m

About

Longform geopolitical analysis. Each episode drops you inside a defining crisis — tracing the history, the actors, the structural forces, and the academic frameworks that explain what is actually happening. AI-assisted synthesis with published methodology. Sources disclosed. Limitations acknowledged.