A half-trillion-dollar industry exists for people who voluntarily seek mortal danger—and it's not who you think it is. We explore the psychology of sensation-seeking, from the dopamine science that explains why some brains crave extreme risk, to the economics of adventure tourism that's growing at 15-18% annually. Spoiler: your dad is the adrenaline junkie, and he's loaded.

0:00 - The Hook: What Would You Pay to Almost Die?

1:15 - The Adventure Market: $415-552B Annually

3:45 - Demographics Shock: It's Not Young People

6:30 - The Adrenaline Myth: Marvin Zuckerman's Dopamine Discovery

10:00 - The Four Factors of Sensation-Seeking

12:15 - Who Profits: Space Tourism and Extreme Segments

15:30 - Why Your Brain is Wired Differently

---

Sources & further reading:

• Marvin Zuckerman — Sensation Seeking and Risky Behavior (APA, 2007)

• Stephen Lyng — "Edgework: A Social Psychological Analysis of Voluntary Risk Taking" — American Journal of Sociology, Vol. 95 No. 4, 1990: https://www.journals.uchicago.edu/doi/abs/10.1086/229379

• Eric Brymer & Robert Schweitzer — "Extreme sports are good for your health" — Journal of Health Psychology, 2013: https://pubmed.ncbi.nlm.nih.gov/22689592/

• Alan Arnette — "How Much Does It Cost to Climb Everest: 2025 Edition": https://www.alanarnette.com/2025/01/13/how-much-does-it-cost-to-climb-everest-2025-edition/

• Kai Bai & Rufeng Chen — "Eliciting Mortality Salience in Dark Tourism and its Influence" — Journal of Travel Research, 2025: https://doi.org/10.1177/00472875231223659

• Chauncey Starr — "Social Benefit versus Technological Risk" — Science, 1969

• Gerald Wilde — "The Theory of Risk Homeostasis" — Risk Analysis, 1982: https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1539-6924.1982.tb01384.x

• Philip Stone — Institute for Dark Tourism Research, University of Central Lancashire — mortality mediation framework

• Frank Farley — "The Type T Personality: Born to Be Wild?" — The Washington Post, 1986: https://www.washingtonpost.com/archive/lifestyle/wellness/1986/05/21/the-type-t-personality-born-to-be-wild/

• Jeff Greenberg, Sheldon Solomon, Tom Pyszczynski — Terror Management Theory: https://en.wikipedia.org/wiki/Terror_management_theory

• USCG — Titan submersible investigation report, August 2025: https://www.ntsb.gov/investigations/AccidentReports/Reports/MIR2536.pdf

• Global Summit Guide — "Sherpa Wage Economy: Everest Expedition Payments": https://globalsummitguide.com/sherpa-wage-economy-everest-expedition-payments/

• Grand View Research — Adventure Tourism Market Report: https://www.grandviewresearch.com/industry-analysis/adventure-tourism-market-report

• NPR — "One-Third of Everest Deaths Are Sherpa Climbers": https://www.npr.org/sections/parallels/2018/04/14/599417489/one-third-of-everest-deaths-are-sherpa-climbers

• Journal of Travel Medicine — Selfie-related deaths analysis, 2022

• ScienceDirect — Whakaari/White Island regulatory failure analysis, 2025: https://www.sciencedirect.com/science/article/pii/S0377027325001970

• National Geographic — "How Instagram Is Changing Travel": https://www.nationalgeographic.com/travel/article/how-instagram-is-changing-travel

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Fanatics doesn't just sell jerseys—they *are* the jersey. We break down how a single company conquered official merchandise, trading cards, and sportsbooks across every major U.S. league, and why the closest historical parallel is the British East India Company's monopoly through vertical integration and exclusive royal charters.

00:00 - Intro: How much did you really spend on sports gear last year?

02:30 - The invisible hand: Fanatics runs NFL, NBA, MLB, NHL, MLS official shops

05:45 - Three empires in one: Merchandise, trading cards, and sportsbooks

08:15 - The East India Company parallel: Why chartered monopolies always vertically integrate

14:30 - The crown grants the charter, but the company builds the empire

---

Sources & further reading:

• Sacra — "Fanatics revenue, valuation & funding": https://sacra.com/c/fanatics/

• DiCello Levitt — "Landmark Antitrust Class Action Against Fanatics and Major U.S. Sports Leagues Over Trading Card Monopoly": https://dicellolevitt.com/dicello-levitt-files-landmark-antitrust-class-action-against-fanatics-and-major-u-s-sports-leagues-over-trading-card-monopoly/

• CNBC — "How Fanatics cornered the sports collectibles market": https://www.cnbc.com/2026/05/22/how-fanatics-cornered-the-sports-collectibles-market.html

• Wikipedia — "2024 Major League Baseball uniform controversy": https://en.wikipedia.org/wiki/2024_Major_League_Baseball_uniform_controversy

• Yahoo Sports — "Fanatics CEO's initial reaction to MLB uniform fiasco: 'We f***ed this up'": https://sports.yahoo.com/fanatics-ceos-initial-reaction-to-mlb-uniform-fiasco-we-fed-this-up-232147021.html

• Sportico — "Fanatics Accused of Conspiring With Leagues, Unions on High Card Prices": https://www.sportico.com/law/analysis/2025/fanatics-trading-card-antitrust-lawsuit-defenses-1234862996/

• Athlon Sports — "The Fanatics Monopoly Lawsuit Was Just Dismissed With Prejudice": https://athlonsports.com/collectibles/fanatics-monopoly-lawsuit-dismissed-2026

• CNBC — "Fanatics acquires Topps trading cards for $500 million": https://www.cnbc.com/2022/01/04/fanatics-acquires-topps-trading-cards.html

• Fanatics Inc — "Michael Rubin, Founder & CEO": https://www.fanaticsinc.com/michael-rubin

• Wikipedia — "Michael Rubin (businessman)": https://en.wikipedia.org/wiki/Michael_Rubin_(businessman

• FOX 9 — "Super Bowl fumble: Fanatics faces backlash over NFL merch quality, availability": https://www.fox9.com/news/super-bowl-fumble-fanatics-backlash-nfl-merch-quality-availability

• Grand View Research — "Licensed Sports Merchandise Market Report, 2026-2033": https://www.grandviewresearch.com/industry-analysis/licensed-sports-merchandise-market-report

• CNBC — "Fanatics CEO Michael Rubin on two sports businesses his conglomerate won't enter": https://www.cnbc.com/2026/07/18/fanatics-fanfest-michael-rubin.html

• ESPN — "Jay-Z, Kevin Durant, Tom Brady top attendees at Michael Rubin's 2026 'White Party'": https://www.espn.com/nfl/story/_/id/49244749/michael-rubin-white-party-2026-athletes-lil-wayne-karl-anthony-towns-tom-brady-kevin-durant

• Forge Global — "Fanatics IPO: Investment Opportunities & Pre-IPO Valuations": https://forgeglobal.com/fanatics_ipo/

• Sportico — "Fanatics, New Era, SeatGeek Lead List of IPO Maybes in 2026": https://www.sportico.com/business/finance/2026/ipo-boom-sports-companies-fanatics-new-era-seatgeek-1234882251/

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Your AI projects aren't failing because of the AI. They're failing because of invisible data corruption sitting in your systems like a time bomb. Master data management isn't boring enterprise talk—it's the $3.1 trillion problem hiding in plain sight, and it's why 95% of generative AI pilots delivered zero measurable returns in 2025.

00:00 - Minecraft save corruption as data quality metaphor

03:45 - The moment you realize the damage happened long ago

06:30 - $3.1 trillion annually: the true cost of bad data

09:15 - Why AI pilots fail: 95% return zero value, 42% of companies abandoned initiatives

11:20 - Master data management: the unsexy foundation

14:30 - Building data quality before building AI systems

16:45 - Data first, AI second—the only path that works

---

Sources & further reading:

• Tadhg Nagle, Thomas C. Redman, David Sammon — "Only 3% of Companies' Data Meets Basic Quality Standards": https://hbr.org/2017/09/only-3-of-companies-data-meets-basic-quality-standards

• Thomas C. Redman — "Data Quality Should Be Everyone's Job" (HBR, 2016) — (paywalled HBR archive)

• Richard Y. Wang et al. — "Beyond Accuracy: What Data Quality Means to Data Consumers" (Journal of Management Information Systems, 1996)

• Thomas C. Redman — "Data's Credibility Problem" (HBR, 2013): https://hbr.org/2013/12/datas-credibility-problem

• IBM — "The True Cost of Poor Data Quality": https://www.ibm.com/think/insights/cost-of-poor-data-quality

• Gartner — "Lack of AI-Ready Data Puts AI Projects at Risk" (2025): https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk

• John Sviokla — "Stop Cleaning Your Data. Use AI to Figure Out Which Info Matters" (Forbes, 2026): https://www.forbes.com/sites/johnsviokla/2026/04/22/stop-cleaning-your-data-start-finding-the-signal/

• TDAN — "Mars Orbiter Disaster: A Failure of Data Management": https://tdan.com/mars-orbiter-disaster-a-failure-of-data-management/5128

• The Register — "What Happened with England's COVID-19 Excel Spreadsheet Blunder" (2020): https://www.theregister.com/2020/10/05/excel_england_coronavirus_contact_error/

• Monte Carlo — "5 Examples of Bad Data Quality": https://montecarlo.ai/blog-bad-data-quality-examples

• Profisee — "What is a Golden Record in MDM?": https://profisee.com/blog/what-is-a-golden-record/

• Tahir Khan — "From Golden Record to Golden Context" (Medium, 2026): https://medium.com/@Tahir-Khan/from-golden-record-to-golden-context-redefining-master-data-for-ai-agent-consumption-2349eec0840b

• Landbase — "Duplicate Record Rate Statistics": https://www.landbase.com/blog/duplicate-record-rate-statistics

• Matillion — "The 1-10-100 Rule of Data Quality: A Critical Review": https://www.matillion.com/blog/the-1-10-100-rule-of-data-quality-a-critical-review-for-data-professionals

• Global Growth Insights — "MDM Market Size Forecast 2025–2033": https://www.globalgrowthinsights.com/market-reports/master-data-management-market-100904

• Stanford Law (via arXiv) — "Data Quality Challenges in Retrieval-Augmented Generation": https://arxiv.org/pdf/2510.00552

• NBC New York — "CFPB Fines Equifax $15 Million Over Credit Report Errors": https://www.nbcnewyork.com/news/business/money-report/cfpb-fines-equifax-15-million-for-errors-on-credit-reports/6111975/

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Why 2026 is genuinely a terrible year to buy a computer—and why a four-year-old used graphics card is actually the smart move. This episode dives into the AI-driven memory shortage that shattered consumer GPU pricing, explains why price-per-gigabyte is the metric that actually matters, and tackles the critical question five research passes still haven't answered.

00:00 - Intro & The Turning Point Episode

02:45 - The Broken Market: AI Data Centers & Memory Shortage

05:30 - Graphics Card Pricing Collapse Explained

08:00 - Price Per Gigabyte: The Real Shopping Metric

11:15 - The 16GB Ceiling Problem & Used Hardware Economics

15:00 - RTX 3090 Pricing: Why Old GPUs Make Sense Now

17:30 - The Unanswered Question

---

Sources & further reading:

• ComputerBuild repo — docs/build-recommendation.md, section 1 (market pricing, the leak-chain hedge, tracker independence caveat)

• ComputerBuild repo — docs/build-recommendation.md, section 4 (buying a used 3090; sold-comp pricing; the $995–1,504 spread)

• ComputerBuild repo — decisions/decision-log.md, P1 and P11 (used-card validation unanswered after five passes; the ceiling now consumed)

• ComputerBuild repo — decisions/decision-log.md, D2 (used 3090 decision, accepted risks, reversal condition)

• ComputerBuild repo — docs/build-recommendation.md, chassis and rack section (rack pricing gap)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Building an AI computer from scratch forces a question: why not just rent compute from a cloud provider? This episode reveals the answer—a research finding that reshapes everything. VEC2 text can reconstruct up to 92% of your original input from numerical embeddings alone, proving your data isn't as hidden as you thought. Privacy just became non-negotiable. // 00:00 - Why rent when you can buy? // 02:30 - Embeddings decoded: How text becomes coordinates // 07:15 - The VEC2 text bombshell: Your embeddings leak secrets // 13:45 - Why this changes the entire AI equation

---

Sources & further reading:

• ComputerBuild repo — decisions/decision-log.md, D8 (embedding inversion, single security boundary, manually-unlocked volume)

• ComputerBuild repo — workloads/workload-map.md, section E (threat model, security architecture, autonomy ramp)

• Morris, Zhao, Chiu, Shmatikov et al. — vec2text and embedding-inversion research (arXiv 2406.10280, 2411.05034, 2507.07700): https://arxiv.org/abs/2406.10280

• ComputerBuild repo — decisions/decision-log.md, D4 and P16 (keeping the home server separate; plaintext transcripts still exposed)

• ComputerBuild repo — docs/build-recommendation.md, section 5 (disk-encryption performance, and the warning against quoting the worst case)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Five episodes deep into building a machine, the hosts finally discover the answer to a question left blank in the project spec: what is this actually for? A private assistant designed to sit atop someone's life records—diary material, financial data, therapy-adjacent notes—to surface stressors and clarify what matters. This episode explores the awkward reality of building something before knowing why, and how a completed spec can be the moment everything clicks into place.

00:00 - The Five Episodes of a Machine Without a Job

03:30 - The Empty Section Nobody Noticed

06:45 - Introducing the Real Purpose

09:15 - What "Life Records" Actually Means

13:00 - Why Architecture Matters When You're Building for One Person

18:30 - The Moment a Project Stops Being Abstract

---

Sources & further reading:

• ComputerBuild repo — workloads/workload-map.md, section E (the application, retrieval design, resource requirements)

• ComputerBuild repo — corpus/plan.md (measured corpus size, the 13.7% signal asymmetry, open design questions)

• Ovadia et al. — Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs (arXiv 2312.05934, EMNLP 2024): https://arxiv.org/abs/2312.05934

• ComputerBuild repo — decisions/decision-log.md, D5 and P8 (model choice, the capability gap, and why it is a values call)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Episode 5 of the build series confronts the central question: what does this component do that nothing you already own can do? We introduce the workload map—a file that justifies every hardware purchase by tracing each part back to a real workload. The philosophy is ruthless: requirements come from code, logs, or actual instrumentation. Every part must survive the question, or it's just a purchase looking for a reason.

00:00 - Introduction & Episode Overview

02:45 - The Build Series Context

04:30 - The Central Philosophy: A Build is Only Defensible if it Traces

08:15 - Shopping Carts vs. Arguments

11:00 - The Workload Map File

13:45 - Requirements, Measurements & the Legend

16:20 - Building with Intention

---

Sources & further reading:

• ComputerBuild repo — workloads/workload-map.md (the stated purpose, the (measured) legend, sections A and D)

• ComputerBuild repo — inventory/machines.md, cross-cutting section ("a GPU wouldn't accelerate an existing local practice; it would create one")

• ComputerBuild repo — decisions/decision-log.md, D3 (one card not two; refuted dual-GPU benchmarks; the honest "unanswered" framing)

• ComputerBuild repo — decisions/decision-log.md, P11 (the $3,000 ceiling, still open)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Storage is the pantry, RAM is the prep counter, and VRAM is the graphics card's secret kitchen—and they're all wildly different things. In episode four of the Computer Build series, the hosts untangle why computers need three completely different types of memory, each with its own speed, size, and purpose. From a one terabyte pantry to a 32GB prep counter and a graphics card's jealous 24GB private station, discover why your computer's line cook won't let anyone else touch their equipment.

00:00 - Welcome Back to Computer Build Series

01:30 - Why Three Memory Types Confuse Everyone

04:00 - Storage: The Pantry (1TB, Big & Slow)

07:45 - RAM: The Prep Counter (32GB, Fast & Volatile)

11:30 - VRAM: The Graphics Card's Private Station (24GB)

15:00 - The Line Cook Metaphor & Why It Matters

18:45 - What's Coming Next

---

Sources & further reading:

• ComputerBuild repo — workloads/workload-map.md, section E (VRAM budget table, KV cache, the reranker on CPU)

• ComputerBuild repo — decisions/decision-log.md, D9 (32 GB not 64 GB; per-gigabyte pricing; the four-DIMM caveat)

• ComputerBuild repo — decisions/decision-log.md, D6 and P3 (1 TB sizing; usable context length still open)

• ComputerBuild repo — workloads/workload-map.md, section D (measured storage footprint)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

Episode 3 of our 10-part series on building an AI computer from scratch. Today we tackle the strangest thing in the whole build: graphics cards are the artificial intelligence component. This seems backwards until you realize graphics rendering and neural networks solve identical problems—millions of trivial, independent calculations that happen simultaneously. The GPU wasn't repurposed for AI. It was always built for this shape of problem.

00:00:00 - Intro: Three of ten episodes in the build

00:02:30 - The bizarre core fact: graphics cards run AI

00:06:15 - Why this seems backwards

00:08:45 - The revelation: pixels and neurons are structurally identical

00:12:30 - How parallel processing works

00:16:00 - Why nobody repurposed anything

---

Sources & further reading:

• ComputerBuild repo — decisions/decision-log.md, D1 (NVIDIA/CUDA only; the 2024 ROCm failure; reversal condition)

• ComputerBuild repo — workloads/workload-map.md, section A (Kaggle-offloaded workloads and their measured VRAM ceilings)

• ComputerBuild repo — inventory/machines.md (laptop specs; the Radeon iGPU as origin of D1)

• ComputerBuild repo — docs/build-recommendation.md, section 7 (thermals, noise and idle power as unresearched)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.

A spec sheet looks like alphabet soup until you understand what each part actually does. This episode explains every major computer component using a working restaurant kitchen as a blueprint: the chef (CPU), prep station (RAM), walk-in cooler (storage), and more. By the end, you'll decode hardware specs without drowning.

00:00 - Intro & The Motivation Problem

01:00 - Welcome to the Kitchen Metaphor

02:30 - The Chef: Understanding Your CPU

04:30 - The Prep Station: RAM and Memory

07:00 - The Walk-In Cooler: Storage Drives

09:30 - Extra Chefs in the Kitchen: Your GPU

12:00 - Power and Ventilation: Cooling and Power Supply

15:00 - Reading the Menu: Decoding Spec Sheets

17:30 - Outro

---

Sources & further reading:

• ComputerBuild repo — docs/build-recommendation.md, section 3 (parts, prices, and the unverified-pricing list)

• ComputerBuild repo — docs/build-recommendation.md, chassis and rack section (4U clearances, the 155mm cooler, GPU length)

• ComputerBuild repo — decisions/decision-log.md, D7 (rack is not server silicon; ATX throughout; socket lifespan)

• ComputerBuild repo — decisions/decision-log.md, P12 (3090 physical fit, unverified)

This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.