#7 - Julia Galef on making humanity more rational, and why Twitter isn’t all bad

Julia Galef's Current Projects and Methodology

• Galef is conducting a part-time, independent research project contracted by the Open Philanthropy Project to investigate expert disagreements on high-impact global questions.
  • The project aims to host dialogues between experts with differing models and influential non-experts (from tech, finance, government, media) to identify the "cruxes" of their disagreements.
  • A "crux" is defined as a specific underlying belief or assumption such that changing it would causally alter one's stance on the primary topic.
  • Galef intentionally avoids framing these conversations as efforts to "change minds," preferring a goal of jointly modeling the landscape of differing views to reduce resistance.
  • The project includes an element of "community building" to establish a norm of epistemic curiosity among influential actors, rather than strictly producing peer-reviewed research.
• Galef hosts the Rationally Speaking podcast (since 2010) and is writing a book on improving human reasoning and judgment.
  • She argues that books serve as a superior vehicle for spreading a thesis compared to blog posts because they provide a concrete "handle" for public discourse, despite the existence of padding.
  • The podcast averages approximately 35,000 listeners per episode, with occasional peaks near 100,000.
  • Galef notes that her audience appears highly engaged; ads for GiveWell run on the podcast generate a disproportionate number of donations compared to other platforms.

Methodological Insights and Communication

• Galef identifies the "double crux" process (developed by the Center for Applied Rationality) as a related but more formal technique than her current "undebate" style.
  • Her approach is less rigid to facilitate convivial dinner settings, prioritizing the creation of a cultural norm around rigorous inquiry.
  • She has found that focusing on understanding the specific reasons behind a belief often yields better results than focusing on the goal of shifting the belief itself.
• Galef highlights specific lessons regarding communication and public perception derived from her experience on Twitter and in writing.
  • Simply adding a sentence to clarify a misinterpretation is often ineffective; audiences with strong priors often reinterpret information to fit existing beliefs regardless of explicit corrections.
  • She notes that Twitter's 140-character limit (at the time of the interview) forces conciseness but also leads to messy strings of tweets that obscure nuance.
  • Communication on short-form platforms often fails because it lacks the "cost of trolling," which is higher when consuming long-form content.

Career History and Transition

• Galef spent one year in a PhD program in economics before dropping out due to ideological and personal misalignment.
  • Ideologically, she was concerned about the lack of rigor in social science research, citing observations of "p-hacking" and the manipulation of survey wording to achieve desired results.
  • Personally, she identified as a generalist who preferred discussing a wide variety of topics over the narrow specialization required in academia.
• Prior to her current career, Galef briefly pursued a path in urban design and architecture (circa 2008–2009) as a freelance journalist.
  • She left this path due to similar concerns regarding the lack of empirical rigor and the difficulty of conducting experiments in complex systems like city planning.
• She attributes her career success largely to networking with smart, interesting people, which provided access to opportunities not available through traditional job boards.
  • She acknowledges that her ability to take career risks was supported by a financial safety net, including stipends from her parents during her early post-PhD years.

Institutional Application and Future Outlook

• Galef believes that improving human judgment in high-level institutions (e.g., Congress, World Bank) is currently hindered by misaligned incentives.
  • Politicians often face short-term political costs for accuracy (unpopularity) while long-term rewards for accuracy are uncertain or nonexistent.
  • She identifies the U.S. Intelligence Community as a notable exception where incentives for accuracy are stronger, citing the work of IARPA (funded by Jason Matheny) and Phil Tetlock's Superforecasting.
  • The Intelligence Community is noted for having fewer pressures to be "entertaining" or "mainstream" compared to public pundits.
• Galef observes positive trends in the scientific community, specifically the "replication crisis" leading to increased rigor.
  • There is a growing trend of journals and institutions requiring data sharing and pre-registration, though this creates tension with the traditional "publish or perish" academic incentive structure.
• She suggests that the most valuable future research lies in long-term, real-world studies on de-biasing interventions, which are currently underfunded compared to the study of biases themselves.
  • Potential academic paths for aspiring researchers include behavioral economics, cognitive science, or psychology.
• Galef envisions a future "loose-knit community" of 100 influential people across tech, VC, government, and media who have revised their views through structured dialogue on the 12 most important future issues.

Critique of Rationality and Effective Altruism (EA) Communities

• Galef advises rational and EA communities to be wary of over-relying on explicit reasoning frameworks (like utilitarianism) when they produce counterintuitive or ethically "suspicious" results.
  • She cites Paul Christiano's concept of "integrity for consequentialists," suggesting that discomfort with a framework's output should trigger a revision of the model rather than blind adherence.
• She raises the possibility that the EA community's current rapid growth and broad outreach strategy may be a strategic error.
  • She contrasts the EA community's scaling needs with the scientific community's need for high epistemic standards and selectivity, noting it is unclear which pole is optimal for EA's current stage.
• Galef notes that while she has not observed widespread dishonesty within her specific circles, the community appears to have higher integrity standards than the general population.

Forward-Looking Statements and Conferences

• Galef states she is prepared to pivot her focus entirely to a specific global catastrophic risk if she updates her beliefs to conclude such a risk is imminent and likely.
• She identifies two primary conferences for networking in this space:
  • The Northeast Conference on Science and Skepticism (held annually in New York), which focuses on scientific literacy and evidence.
  • Nexus (at nexus.org), where she holds live podcast tapings and which attracts a mix of rationalists and EAs.