Off the Clock #7: Getting on the Crazy Train with Chi Nguyen

Guest and Context

• Special guest is Chi Nguyen, a researcher focusing on "evidential cooperation in large worlds" (ECL).
• The podcast was recorded on July 4th (US Independence Day) and coincided with the UK general election.
• The episode includes a game show segment ("Is it Trump or AI?") and a "Would You Rather" scenario.

Evidential Cooperation in Large Worlds (ECL)

• ECL is a decision theory proposing that agents should cooperate with others even if they cannot causally affect them, such as entities in distant parts of the universe.
• Chi Nguyen credits Casper Österheld as the originator of the theory and her mentor; she notes she is not the "most cooperative German" due to him.
• Nguyen entered the effective altruism (EA) field in 2017 via the Stiftung für Affective Altruism (now Center on Long-term Risk).
• The organization originally focused on "S-risk" (suffering risk), specifically long-term suffering in the universe, alongside "wacky" ideas involving AI and string theory.
• Nguyen describes her research path as "historical path dependency" driven by a willingness to follow logic into counterintuitive conclusions, despite feeling some tension between rigorous logic and pragmatic constraints.
• She expresses skepticism about "wager-based reasoning" (betting on extreme outcomes like transformative AI in 5 years) outside of the EA community, though she acknowledges its prevalence within the movement.
• Funding for ECL research is described as "okay for the next half-year" but "dire" subsequently, with multiple large donors withdrawing support.
• Nguyen states there are approximately five people currently working on ECL and invites institutional donors to contact her for connection.

Voting and Political Decision Making

• Nguyen skipped voting in the UK election because she resides in Keir Starmer's district, which is a "safe Labour seat" where her vote would not be consequential.
• The group discusses Rob Wiblin's argument that voting has high expected value due to the possibility of swinging an election and the magnitude of potential impacts (e.g., trillions in budget allocation or nuclear war risks).
• Huynh (host) remains skeptical of the specific probability calculations used in Wiblin's articles, citing reliance on Monte Carlo analysis that may not reflect reality.
• The conversation touches on the "meta" nature of voting, where one must consider how other voters' perceptions of a candidate's age or competence (specifically Joe Biden) might influence election outcomes.
• A discussion references a LessWrong post titled "Politics is far too meta," which argues voters should be honest about their true preferences rather than strategically aligning with perceived public sentiment.
• Nguyen defends her decision to vote as a "non-galaxy brain" act of civic duty to prevent a "race to the bottom" where altruists disengage from normie politics, even if the direct causal impact is low.
• The group debates whether society benefits more from increased altruism (people acting on their values) or increased effectiveness (people having better beliefs/tools), concluding that marginally more altruism is preferable.

Media, Philosophy, and Historical Incidents

• Chi Nguyen appeared on BBC Radio 4 to debate the defunding of art; she argued for opportunity costs, suggesting that talent drawn to art might be better deployed solving global problems.
• The panel framing described her position as having a "hostile attitude to art" and forced her to justify moral trade-offs, which she felt was an unfair burden given the presumption that art funding is inherently good.
• The transcript notes a recent shift in Warren Buffett's philanthropy: he is no longer directing the majority of his wealth to the Bill & Melinda Gates Foundation.
• Buffett's new strategy involves distributing funds to foundations run by his three children, focusing on "art and culture in Nebraska," "earth-centered healing," and "border patrolling/countering human trafficking."
• Nguyen argues Buffett's decision is "unjustifiable" because he explicitly understood the life-saving impact of the Gates Foundation model and reneged on a moral commitment, even if not a legal one.
• The group observes that society holds no standard for billionaires to be "good" donors, only to avoid being "bad," and that shouting at philanthropists often discourages giving rather than improving allocation.
• Chi Nguyen references a past "game show" attempt to derail the podcast, which was unsuccessful, and a segment on the "manifest drama" on the EA forum.

Personal Anecdotes and Observations

• Nguyen notes that Berlin's economy reportedly improves when the capital is excluded from data, attributing the city's vibe to "lazy nightclubbing artists" rather than poverty.
• She contrasts the US "hustle" culture with continental Europe, where people work fewer hours and report higher life satisfaction, though she concedes this might be detrimental to innovation.
• Huynh recounts a past behavior of rigorously comparing grocery options across multiple stores for "organic," "fair trade," and "sustainability," which he now views as inefficient altruism.
• The "Would You Rather" segment asked the group to choose between removing a poem Nick Bostrom wrote criticizing the Future of Humanity Institute (FHI) or removing Holden Kamoski's "astroturfing" (fake reviews) of GiveWell.
• The group consensus leans toward removing the astroturfing to prevent a reputational hit for Holden, despite arguments that his apology might have set a positive example for accountability.