Dwarkesh PodcastPatrick McKenzie — Money laundering, big tech censorship, SBF & Japan
CHAPTERS
- 0:00 – 2:20
VaccinateCA’s origin story: a Discord hackathon becomes national vaccine search
Patrick recounts how a late-night tweet turned into a Discord-led sprint to centralize vaccine availability information. What started as a volunteer hackathon quickly became a de facto public-private clearinghouse used nationwide.
- •People were making dozens of calls to find available vaccine appointments
- •A volunteer team spun up a working plan and site overnight via Discord
- •VaccinateCA scaled from a California effort into a national information layer
- •Patrick unexpectedly became the organizational lead/CEO
- •Core lesson: society shouldn’t depend on ad-hoc volunteer heroics as Plan A
- 2:20 – 5:52
Why government and pharmacies didn’t solve it: ownership gaps and software abdication
They unpack why no institution claimed responsibility for the end-to-end logistics and information problem. Patrick argues the federal government learned the wrong lesson from Healthcare.gov and treats software as someone else’s problem.
- •No actor in the system felt clear responsibility for “shots in arms” logistics
- •Complex, fragmented supply chains and incompatible databases blocked visibility
- •Government IT procurement dynamics repeatedly failed to produce integration
- •Healthcare.gov backlash made leaders reluctant to bet politically on software
- •Modern competence requires software competence, but government has abdicated it
- 5:52 – 8:42
The “memory hole” after COVID: incentives, politics, and lack of accountability
Dwarkesh presses on why the vaccine rollout failures aren’t a dominant scandal. Patrick attributes it to election-year incentives and a broader institutional tendency to avoid re-litigating embarrassing decisions.
- •Pandemic failures are being intentionally or passively forgotten
- •Political incentives discourage revisiting contradictions (e.g., shifting mask guidance)
- •Controversial allocation policies (e.g., geofencing/redlining) went largely unchallenged
- •Iterated games: both parties find it convenient not to make it salient
- •Broader claim: lack of seriousness across institutions about “winning” the pandemic
- 8:42 – 15:29
Big Tech’s political risk calculus: January 6, “don’t get in the newspaper,” and the Copenhagen principle
Patrick explains why major tech companies didn’t step in to build an inventory-style solution, despite having the capability. After January 6, internal policy and PR incentives pushed companies to minimize visibility and avoid appearing more competent than government.
- •Post–Jan 6 political pressure led firms to keep heads down
- •Public-health teams inside big tech were often overruled by policy/PR
- •Fear: “embarrassing government” would be punished via regulation or public backlash
- •Copenhagen principle: whoever builds becomes blamed for all downstream failures
- •Cambridge Analytica narrative contributed to a desire to clip tech’s wings
- 15:29 – 32:23
How politics crippled common-sense distribution: byzantine tiers and equity bikeshedding
They drill into how politically driven prioritization created operational complexity that reduced vaccination speed. Patrick argues the tiering system optimized for coalition management and symbolic equity over lives saved and throughput.
- •Tier systems (1A/1B/1C) varied by place and changed frequently
- •Lobbying shaped eligibility (e.g., veterinarians, teachers) over medical need
- •Complex rules forced burdensome eligibility verification, excluding vulnerable groups
- •Bikeshedding: “equity strategy” dominated conversations over logistics reality
- •California reportedly administered only ~25% of allocated doses early on
- 32:23 – 39:01
“End-of-day shots” and the madness of throwing vaccines away
A concrete example of bad epistemics: expiring doses should go to anyone rather than the trash. Patrick contrasts Israel’s pragmatic approach with U.S. policies that protected tier integrity even at the cost of waste and intimidation of providers.
- •Multi-dose vials expire quickly once opened (regulatory + practical constraints)
- •Israel: walk outside and offer shots to anyone to prevent waste
- •U.S.: policies and public threats discouraged end-of-day shots
- •Quiet policy reversals left pharmacies uncertain and risk-averse
- •Some pharmacists quietly advised people to “say what you need” to get vaccinated
- 39:01 – 50:46
Fundraising for VaccinateCA: why raising $1.2M was harder than a seed round
Patrick details the fundraising reality: fast decisive donors existed, but many funders had slow processes and shifted attention once they personally got vaccinated. At the margin, the effort relied on personal financial risk—down to Patrick advancing payroll and tapping family savings.
- •Total budget about $1.2M—small relative to stakes and tech wealth
- •Some leaders wrote large checks immediately; others took weeks of process
- •By May, many funders perceived the problem as “solved” or merely hesitancy
- •Funding ultimately included Patrick’s personal money (including college fund)
- •Charity virtues (minimizing waste/process) can conflict with speed and saving lives
- 50:46 – 59:30
What tech should learn about government (and vice versa): engagement, strategy, and software realism
They move from diagnosis to prescriptions: tech needs higher-fidelity models of how Washington works and should engage strategically rather than only complaining. Government must treat software as core state capacity and stop treating tech as the enemy.
- •Don’t “discount government to zero”—the system achieved a lot but underperformed
- •Tech must learn to interface with government and be strategic to stay “in the room”
- •Government needs to demystify software and build durable competence
- •Historic model: deputize private-sector expertise (“make them a colonel”)
- •After-action reports often celebrate coalition-maintenance over operational truth
- 59:30 – 1:07:21
What is crypto good for? Market cap vs real costs and realized utility
Dwarkesh asks what a $3T crypto peak represented; Patrick reframes the harm as rivalrous resources burned (engineering talent, GPUs, energy). He uses Vitalik Buterin’s ‘have we earned this?’ framing and challenges crypto to show proportional real-world value.
- •Market cap isn’t the same as social cost; the cost is resources diverted to build crypto
- •Estimated real resource spend: tens to hundreds of billions
- •Vitalik’s questions: unbanked banked, meaningful apps, real utility
- •Stablecoins are a real use case, but small relative to investment and hype
- •Prediction: the industry keeps saying ‘still early’ unless value materializes
- 1:07:21 – 1:13:34
Financial surveillance and civil liberties: KYC/AML, banks as policy arms, and the LLM future
Patrick agrees with concerns that the banking system functions as a de facto extension of government, enabling transaction-level surveillance. They explore how the Bank Secrecy Act framework scaled into an ‘intelligence community’ inside banks—and how AI could supercharge it.
- •Banks run massive alert-review operations; most SARs are likely never read
- •KYC/AML evolved from limited state capacity into pervasive surveillance-by-default
- •Privacy risk grows sharply if LLMs can analyze every transaction holistically
- •Crypto’s privacy rhetoric often yields to ‘number go up’ incentives
- •Underlying question: should tech (like finance) be treated as a branch of government?
- 1:13:34 – 1:25:05
How the US government leverages big tech to violate rights: Missouri v. Biden and ‘cats’ paws’
Patrick summarizes allegations and evidence from Missouri v. Biden: government actors pressuring platforms to censor speech, including direct, granular takedown demands. He argues this is constitutionally outrageous and signals unresolved power struggles between state and platforms.
- •Claim: government used tech companies as intermediaries to skirt First Amendment limits
- •Examples include takedown requests of public meeting/town hall recordings
- •Paper trail allegedly includes explicit awareness of constitutional concerns
- •Expectation: partial legal walk-backs, but the power contest will recur
- •Prescription: tech may need more risk tolerance and public advocacy to resist overreach
- 1:25:05 – 1:27:09
Can the US have nice things like Japan? Incentives, ‘culture’ as an error term, and choosing quality
They pivot to comparisons with Japan and the idea that societies can choose operational excellence. Patrick cautions against attributing everything to culture, arguing incentives and institutional choices explain a lot—and the U.S. can choose to build systems that work.
- •‘Culture’ is often a catch-all for unexplained incentives
- •Quality and ‘nice things’ can be achieved without extreme overwork at the margin
- •Japan vs U.S. differences aren’t monocausal; institutions and incentives matter
- •Normative claim: the U.S. should choose to have nice things and not be embarrassed
- •Connects back to state capacity and operational seriousness
- 1:27:09 – 1:33:14
Financial plumbing as an investigative tool: metadata trails, Mt. Gox, and why reporting fails
Patrick explains how understanding banking procedures helps verify claims via external traces—useful for uncovering fraud. He uses Mt. Gox as an example where basic diligence (e.g., asking the bank if wires are working) could have surfaced truths earlier.
- •Fraud often asserts checkable facts that create ‘metadata exhaust’ elsewhere
- •Reporters can validate claims by knowing what filings/calls/logs must exist
- •Mt. Gox story: implausible claims about disabling a major Japanese bank’s wires
- •Simple outreach (fax/phone/letters) to institutions can clarify reality
- •Short-selling and journalism have incentives but still miss major frauds
- 1:33:14 – 1:38:07
Money laundering and SBF: sophistication, regulatory cover, and everyday ‘cleaning’ mechanisms
They discuss the spectrum of money laundering sophistication and Patrick’s contrarian take that SBF/FTX were highly sophisticated in navigating power and regulation. The conversation broadens into common laundering pathways like real estate and the blurry boundary between normal commerce and laundering.
- •Money laundering spans from crude tactics to highly structured global operations
- •Patrick argues SBF’s ecosystem excelled at licenses, regulators, and influence—until collapse
- •Alameda’s cross-border banking/arb origin story raises unresolved questions
- •Real estate + shell entities can ‘clean’ money through legitimate rent streams
- •Key idea: laundering often looks like normal economic operation plus concealment intent
- 1:38:07 – 2:02:17
Maximizing your value and thinking well: negotiation, distractions, and blameless post-mortems
In the closing stretch, Patrick covers why some people negotiate better, how status games distract founders, and how video games can become an escapist substitute for agency. He ends with a governance prescription: adopt blameless postmortems to build learning loops and real accountability after crises.
- •Negotiation gaps come from culture, class signaling, and learned scripts
- •Founders over-optimize status, conferences, and ‘playing house’ vs user impact
- •Games can provide a substitute ladder for agency; can also feed depressive spirals
- •Meta-lesson: avoid ‘LARPing’ rigor—build real rigor and seek grounded models
- •Policy fix: conduct dispassionate, blameless postmortems to prevent repeat failures
