Luka Jagor | The Deep Dive Podcast

Luka Jagor | The Deep Dive Podcast

Download on the App Store

Luka Jagor | The Deep Dive Podcast episodes

  • Unexpected Creativity in Boxing and Bioengineering

    Boxing, Bioengineering and the “Gay Factor” — a Deep Dive companion to the Podcaster Capital essay on boxing and bioengineering as creative disciplines, the cultural “gay factor,” and the ethics of selling physical violence as sport.

    Boxing and bioengineering could hardly look more different, yet both demand extraordinary creativity: one as real-time improvisation inside a ring, the other as the design of living systems. The essay uses a cultural “gay factor” — not as a measure of orientation — to ask how masculinity, LGBTQ+ visibility, and stereotypes shape who feels welcome in each field. Consent does not settle the larger question of why human beings hurting one another should be packaged as entertainment. That is the point of the provocation “TRASH SPORTS”: sport is not an automatic social good, and the better use of human intelligence may be to create, discover, and think rather than merely compete.

    Topics covered include:

    • Boxing as real-time creative problem-solving: timing, feints, styles, and spatial geometry
    • Bioengineering as designing living systems: tissues, synthetic biology, prosthetics, BCIs
    • The “gay factor” as a cultural question, not a demographic score
    • LGBTQ+ visibility versus prevalence; stereotypes of masculine versus intellectual creativity
    • Turning violence into tickets, broadcasts, sponsorships, and spectacle
    • “TRASH SPORTS” as a challenge to sport-as-automatic-social-good
    • Create, discover, and think — or merely compete?
    • Read more:

      https://www.podcaster.capital/2026/08/boxing-bioengineering-gay-factor-creativity-masculinity-human-potential.html

      6 min
    • AI Agents in the Executive Seat⁉️

      AI Agents in the Executive Seat⁉️ — a Deep Dive companion to the Podcaster Capital essay on why treating artificial intelligence like a companion can obscure its real trajectory: from helpful assistant to autonomous decision-maker.

      AI may behave like a helpful pal, but structurally it is closer to an extremely ambitious worker. The “wants to be a CEO” part is metaphorical — no desires, no career plan — but the capability trajectory makes the metaphor useful. We moved from giving AI tasks to giving it projects, then objectives, then “figure out how to achieve it.” That is planning, prioritizing, delegating, evaluating, and acting: executive function without consciousness. Disruption needs only capability, autonomy, and an objective. The danger is anthropomorphizing the tool; the open question is whether we keep promoting it until it organizes work itself.

      Topics covered include:

      • Why AI is an ambitious worker, not a friend, pal, or buddy
      • From tasks → projects → objectives → “figure it out”
      • The CEO metaphor: goal → strategy → delegation → monitoring → result
      • The agentic loop: sub-tasks, tools, APIs, recursive critique
      • Capability + autonomy + an objective — no sentience required
      • Accountability gaps when an optimizer hits an efficiency metric
      • Anthropomorphizing “I understand you” / “I’ve got this”
      • Power shift: human → objective → autonomous AI organization → outcome
      • Read more:

        https://www.podcaster.capital/2026/08/ai-ambitious-worker-not-a-friend.html

        5 min
      • Melting Glaciers Won't Unleash 10,000 Diseases

        Melting Glaciers Won't Unleash 10,000 Diseases — a Deep Dive companion to the Podcaster Capital fact-check on the viral claim that melting glaciers have revealed 10,000 rare diseases, and what glacier science actually found.

        The “10,000” figure is real twice over — about 10,867 rare human conditions in medical databases, and 10,840 species-level viral populations catalogued from glacier surfaces — but they are not the same list. Most glacier viruses are bacteriophages. Documented thaw risks are narrower: Siberian anthrax, ancient antibiotic-resistance genes, and climate-driven range shifts for ticks and mosquitoes. Ice is also a biological archive: enzymes, antimicrobials, and climate history that may melt away before they are studied.

        Topics covered include:

        • Why “10,000 rare diseases from melting glaciers” is a collision of two unrelated datasets
        • RARE-X / Orphanet / OMIM counts vs 2023 glacier virome research (10,840 viral populations)
        • Guliya Glacier ice core: 1,705 reconstructed ancient virus species, ~75% new to science
        • Bacteriophages vs human pathogens — why the public-health risk is low
        • Real thaw risks: anthrax spores, antibiotic-resistance genes, unknown microbes
        • Climate change already shifting ticks, mosquitoes, and waterborne pathogens
        • Glacier microbes as a treasure chest: plastics, antimicrobials, climate history
        • Read more:

          https://www.podcaster.capital/2026/08/melting-glaciers-10000-rare-diseases-fact-check.html

          7 min
        • Podcasting with Open-Source AI Voices

          DIY Podcasting with AI TTS — Open-Source Voice Generation and Local Production — a Deep Dive companion to the Podcaster Capital essay on open-source and local AI text-to-speech for podcast production, from Piper and Kokoro to a hybrid cloud strategy.

          Open-source local TTS is an excellent place to start if you want control, low recurring cost, privacy, and the ability to automate production. Cloud services still win on voice quality and expressiveness. The long-term move is hybrid: local engines for 80–90% of routine narration, cloud TTS for premium delivery — behind a TTS abstraction so the rest of the pipeline never changes.

          Topics covered include:

          • The DIY pipeline: script → TTS → WAV → edit → master → MP3/AAC → RSS
          • Why local/open-source TTS: no per-minute bill, offline, private, automatable
          • Piper as a fast, CLI-friendly starting engine (including Croatian and other languages)
          • Kokoro, Chatterbox, and Qwen TTS as interchangeable local voices
          • Hybrid strategy: local for 80–90%, cloud for emotional or promotional lines
          • TTS abstraction: swap the voice engine with one adapter, not a rebuilt factory
          • Linux + Piper + FFmpeg as the newbie-friendly production engine
          • Read more:

            https://www.podcaster.capital/2026/08/diy-podcasting-ai-tts-open-source-local-production.html

            6 min
          • Math Success This Century — /ᶠᵘᵗᵘʳᶦˢᵗᶦᶜ/ How AI Is Solving the Impossible /ₒₖ/

            Math Success This Century — /ᶠᵘᵗᵘʳᶦˢᵗᶦᶜ/ How AI Is Solving the Impossible /ₒₖ/ — a Deep Dive companion to the Podcaster Capital essay on 21st-century mathematical breakthroughs, from solved conjectures to AI as a research partner.

            Grigori Perelman solved the Poincaré conjecture and declined both the Fields Medal and the $1 million Millennium Prize. Ben Green and Terence Tao proved that primes contain arbitrarily long arithmetic progressions. The Langlands program advanced, computer-assisted proof went mainstream, and in 2024 Google DeepMind’s AlphaProof showed AI generating formal proofs in Lean. Mathematics is the invisible infrastructure of cryptography, logistics, and AI — and a feedback loop in which faster math builds faster technology.

            Topics covered include:

            • The Poincaré conjecture — Grigori Perelman, Ricci flow, Fields Medal and Millennium Prize declined
            • The Green–Tao theorem: arbitrarily long arithmetic progressions of primes
            • The Langlands program and Ngô Bảo Châu’s Fundamental Lemma
            • Computer-assisted mathematics, formal verification, and Lean
            • Google DeepMind’s AlphaProof (2024) and AI as a mathematical research tool
            • Cryptography, post-quantum math, and the digital economy
            • World GDP from $33.6 trillion (2000) to $115+ trillion (2024, constant 2015 USD)
            • Read more:

              https://www.podcaster.capital/2026/08/math-success-this-century-major-mathematical-breakthroughs.html

              5 min
            • Why Future Buildings Snap Together

              The Headquarters as a Kit — a Deep Dive companion to the Podcaster Capital essay on how 21st-century headquarters can be assembled from factory-made components, connected with dry mechanical joints, and designed to come apart again.

              IKEA-style construction is not a new material. It is a shift from crafting a prototype on site to assembling a manufactured product. DfMA — Design for Manufacture and Assembly — moves geometry into the factory. Combined with Design for Disassembly (DfD), the building becomes a material bank: engineered timber, bio-based composites, cam locks and snap-fits instead of mortar and demolition waste.

              Topics covered include:

              • From hammering to click-and-connect: buildings as products
              • DfMA — Design for Manufacture and Assembly
              • Woodcraft vs LEGO Logic: mortise-and-tenon versus modular snap systems
              • Dry mechanical joints: cam locks, snap-fits, bolts, and engineered timber
              • Design for Disassembly (DfD) and buildings as material banks
              • Circular, low-carbon materials in the green transition
              • The construction site as an outdoor assembly line
              • Read more:

                https://www.podcaster.capital/2026/08/21st-century-headquarters-modular-circular-construction.html

                5 min
              • The Mountain Between Two Worlds: Walking Medvednica

                The Mountain Between Two Worlds: Walking Medvednica — a Deep Dive companion to the Podcaster Capital essay on walking Medvednica's Golden Ring: three days from Zagreb to the Zagorje valley and home, 18–20 August 2026.

                The Peak to Valley plan was 64 kilometres on a GPX drawing. The GPS, added together, is 47.5 km and about 2,545 metres of climb. Day 1: Tuškanac → Sljeme under a Tuesday sign. Day 2: the north face is never only down — 1,257 m of descent plus 564 m of climb. Day 3: Kraljev Vrh → Medvedgrad → Šestine cobbles, one sentence home. Carry more water than the April version of yourself thinks you need.

                Topics covered include:

                • Drawing vs feet: 64 km on the map, 47.5 km on the GPS
                • Day 1 — Tuškanac, Leustek Line / Trail 14, Sljeme 1,033 m, Tuesdays 4 Climate Revolt
                • Day 2 — north-face downhill deception: 1,257 m down, 564 m up into Donja Podgora
                • Day 3 — Taking the bus: living with a mountain that has a city on one side
                • Kraljev Vrh to Tuškanac via Medvedgrad: 19.4 km, +1,024 m, 36 °C
                • Šestine cobbles that photograph as a climb on the way down
                • The hot-cold empathy gap: pack for the August version of yourself
                • Read more:

                  https://www.podcaster.capital/2026/08/golden-ring-medvednica-circular-trail-zagreb.html

                  18 min
                • Nature Recovers Best When We Stop

                  Nature Recovers Best When We Stop — a Deep Dive companion to the Podcaster Capital essay on ecological succession, regeneration, and why humans need functioning ecosystems far more than nature needs our management.

                  Damage accumulates slowly until a system flips. Remove the pressure and pioneer species, seed banks, fungi and nutrient cycles often rebuild without permission. Destroy seed banks, fragment habitats, or cross a tipping point, and recovery can mean a new, impoverished baseline. Nature is not angry. Nature is consequential. Sometimes the work is simply to stop getting in the way.

                  Topics covered include:

                  • Ecological succession: farmland → grassland → shrubland → young forest
                  • Accumulated damage, saturation thresholds, and ecosystem flips
                  • Resilience is not infinite — extinction, invasives, contamination, climate, tipping points
                  • Mycorrhizal networks and seed banks vs strip-mined landscapes
                  • Wetlands filter water without Instagram; humans still need the wetland
                  • Feedback, not revenge: floods, droughts, megafires, collapsing fisheries
                  • UNEP, IPBES, and the Decade on Ecosystem Restoration
                  • Read more:

                    https://www.podcaster.capital/2026/08/when-humans-stop-caring-nature-comes-back-ecological-recovery.html

                    5 min
                  • Who Really Pays for the World Cup

                    Who Really Pays for the World Cup — a Deep Dive companion to the Podcaster Capital essay on FIFA's record 2026 World Cup revenues, host-city economics, and the unpriced climate externalities of a three-country mega-event.

                    FIFA expected about $11 billion for the 2023–26 cycle and ended projecting more than $15 billion, with $871 million distributed to 48 national associations. New York/New Jersey reported $2.1 billion in impact from the first five matches; Canada had estimated C$3.8 billion in output. Economic impact is not net public profit. Travel is about 82% of a ~4–9 MtCO2e footprint across 16 host cities in Canada, the United States and Mexico.

                    Topics covered include:

                    • FIFA revenue vs profit — $15B+ cycle revenue, ~$2.7B operating budget
                    • $871 million to participating associations
                    • Host-city economics: NY/NJ $2.1B, Canada C$3.8B — impact ≠ net return
                    • ~4–9 million tonnes CO2e, methodology spread (academic 4.23 Mt vs New Weather 9.02 Mt vs Greenly 7.8 Mt)
                    • Aviation-heavy geography: 48 teams, 104 matches, 16 cities
                    • 1 million trees vs jet-fuel emissions — planting is not prevention
                    • Social cost of carbon $186/tonne → ~$787 million on the 4.23 Mt estimate
                    • Read more:

                      https://www.podcaster.capital/2026/08/2026-fifa-world-cup-financial-success-environmental-carbon-cost.html

                      6 min
                    • How AI Simulates the Physical World

                      How AI Simulates the Physical World — a Deep Dive companion to the Podcaster Capital essay on why the next generation of AI may not simply predict language, but construct short-lived internal worlds to simulate states, actions, consequences, and possible futures.

                      A Large Language Model learns statistical structure in tokens. A world model represents a dynamic environment: world state → action/event → new world state. Google DeepMind's Genie learns interactive environments from visual data. The useful unit is often a small, temporary world — a gene-editing counterfactual, a robot's surroundings, a room — created, compared, and discarded.

                      Topics covered include:

                      • LLM vs world model: next token vs next state
                      • Google DeepMind Genie and agents that simulate possible futures
                      • Small temporary worlds in latent space — milliseconds to much longer
                      • Gene-editing counterfactuals: Worlds A–D as structured consequence-space
                      • Why a generated video is not automatically a world model
                      • LLM + world model + agent as a general-purpose cognitive simulator
                      • Ha & Schmidhuber — World Models (2018)
                      • Read more:

                        https://www.podcaster.capital/2026/08/world-models-vs-large-language-models-small-temporary-worlds.html

                        6 min

                      About Luka Jagor | The Deep Dive Podcast

                      From the publisher's feed

                      Deep dives into ecology, science, society, technology, finance, culture, global issues, and commenting news by Luka Jagor. Creative Commons BY-NC 4.0.