What I believe in

The exercise of cataloguing one's beliefs is both humbling and clarifying¹. Unlike the confident proclamations we see in political discourse, genuine introspection reveals a landscape of varying certainties, provisional commitments, and acknowledged ignorance. This post attempts such a cartography not because my particular beliefs matter much, but because the process of explicit belief articulation serves multiple epistemic functions².

First, it forces precision where we might otherwise rely on vague intuitions. Second, it exposes inconsistencies between stated and revealed preferences. Third, it creates accountability through public commitment. Finally, it models intellectual honesty in an environment that often rewards performative certainty over genuine uncertainty.

What follows is my attempt to sort various social and cultural beliefs by confidence level. I've focused on areas where reasonable people disagree, particularly those touching on contemporary culture war dynamics. The goal isn't to convince but to demonstrate a framework for thinking about controversial topics with appropriate epistemic humility.

High confidence beliefs (80-95% certainty)

Institutional trust requires institutional trustworthiness

The current crisis of institutional trust in Western societies³ represents a rational response to institutional failures rather than mere populist irrationality. When experts consistently fail to acknowledge uncertainty, when institutions prioritize reputation management over truth-seeking, and when elite consensus forms around empirically questionable propositions, declining trust becomes epistemically justified.

This doesn't mean all institutions are equally untrustworthy, nor that expertise should be dismissed wholesale. Rather, it suggests we need better mechanisms for institutional accountability and more honest communication about uncertainty. The solution to declining trust isn't better propaganda but better institutions.

The COVID-19 pandemic provided numerous examples of this dynamic. Public health officials made confident claims about transmission mechanisms, treatment protocols, and policy effectiveness that later proved incorrect or oversimplified. Rather than acknowledging uncertainty, many doubled down on initial positions. The predictable result was accelerated trust decline.

Free speech norms should err toward permissiveness

Contemporary debates about speech regulation often conflate legal rights with social norms. While the First Amendment constrains government action, private platforms and social institutions operate under different frameworks. Nevertheless, I believe both legal and social speech norms should err toward permissiveness⁴.

This isn't because all speech has equal value, but because the alternative of empowering authorities to determine acceptable discourse has consistently produced worse outcomes. The current moment's speech restrictions, whether implemented by platforms, universities, or professional associations, disproportionately target dissent from orthodox positions rather than genuinely harmful content.

The standard objection involves harm prevention. Speech can indeed cause harm through incitement, harassment, or misinformation. But harm-based restrictions require careful calibration. Expansive definitions of harm inevitably capture legitimate discourse, while narrow definitions may miss genuinely problematic speech. The bias should favor false negatives (permitting some harmful speech) over false positives (restricting beneficial speech).

Nuclear energy is underutilized due to irrational fear

The global energy crisis and climate change demands require massive clean energy deployment. Nuclear power provides reliable, carbon-free baseload electricity at scale, yet remains underutilized in most countries due to public fear rather than genuine risk assessment⁵.

Modern nuclear reactors are extraordinarily safe. Deaths per unit of energy produced are lower for nuclear than for any fossil fuel and comparable to renewables. Even including Chernobyl and Fukushima, nuclear energy has caused fewer deaths than coal causes annually through air pollution. The waste storage problem, while requiring careful management, is technically solved and manageable at current scales.

Opposition stems largely from availability bias and dread risk psychology. Nuclear accidents receive extensive media coverage despite their rarity, while the ongoing health impacts of fossil fuel pollution remain largely invisible. Regulatory frameworks designed in response to public fear rather than evidence have made nuclear construction unnecessarily expensive and slow.

Meritocracy is both impossible and necessary

Pure meritocracy systems that perfectly reward individual merit cannot exist. Measurement problems ensure that any merit assessment captures only proxies. Environmental factors like family background, educational quality, and luck significantly influence outcomes. Social connections and implicit biases distort evaluation processes.

Yet meritocratic ideals remain valuable. Societies that abandon merit-based selection in favor of identity-based allocation or nepotistic systems generally perform worse across multiple dimensions. The goal should be improving rather than abandoning meritocratic processes.

This requires acknowledging tradeoffs. Perfect meritocracy is impossible, but approximate meritocracy outperforms alternatives. Efforts to correct for bias and environmental disadvantage are worthwhile, but should preserve the essential link between competence and responsibility. The current trend toward wholesale rejection of merit-based systems threatens institutional effectiveness without meaningfully advancing equality.

Prediction markets should guide more decisions

Markets aggregate information more effectively than committees, experts, or polling for many prediction tasks⁶. Participants with superior information or analytical ability profit by moving prices toward truth, while those with poor judgment lose money and influence. This creates incentives for accuracy that most other forecasting methods lack.

Prediction markets consistently outperform expert surveys, traditional polls, and institutional forecasts across domains from elections to economic indicators to geopolitical events. They provide not just point predictions but measures of uncertainty through price distributions. They update continuously as new information emerges rather than relying on periodic expert consultations.

Policy applications could include infrastructure project cost estimation, regulatory impact assessment, and strategic planning. Corporate applications might cover product launch success, merger outcomes, and technology adoption timelines. The main barriers are regulatory restrictions and cultural resistance to market mechanisms in non-commercial contexts.

Medium confidence beliefs (60-80% certainty)

Gender differences are real but overstated

The contemporary gender debate suffers from false binaries. One side claims biological sex determines all relevant differences, while the other insists gender represents pure social construction. Both positions oversimplify complex empirical questions.

Substantial evidence supports average psychological differences between men and women⁷. These include differences in personality traits, cognitive abilities, and preferences. Cross-cultural studies suggest biological rather than purely cultural origins for many such differences. Evolutionary psychology provides plausible mechanisms.

However, these differences are typically small in magnitude with enormous within-group variation. Most individuals don't conform to gender stereotypes. Many observable differences likely reflect social factors rather than biological ones. Policy should generally ignore group averages and focus on individual capabilities.

The practical implications concern institutional design. Sex-segregated institutions sometimes make sense, particularly in contexts involving physical competition or privacy concerns. But most educational and professional contexts should remain sex-integrated while accommodating individual differences.

Cultural assimilation benefits both immigrants and hosts

Immigration debates often polarize between unrestricted multiculturalism and nativist exclusionism. A more nuanced position recognizes that cultural assimilation while preserving beneficial diversity generally produces better outcomes.

This isn't because immigrant cultures lack value. Many immigrant communities possess advantageous traits that benefit host societies. Rather, it's because functional societies require shared norms and common languages for effective coordination.

Research suggests that immigrants who assimilate linguistically and culturally experience better economic outcomes and social integration. Their children also perform better academically and socially. Host societies benefit from immigrant contributions while maintaining social cohesion.

The key distinction involves which cultural elements should be preserved versus adapted. Language acquisition and civic norm adoption seem clearly beneficial. Religious practices and cultural traditions that don't conflict with liberal democratic values can be maintained. Practices involving coercion or rights violations require modification or abandonment.

Educational interventions have modest effects

The education reform movement promises to eliminate achievement gaps through pedagogical innovations, increased funding, or structural changes. While well-intentioned, these efforts typically produce disappointing results⁸.

Large-scale studies consistently find modest effects for most educational interventions. This includes class size reduction, teacher training programs, curriculum reforms, and technology integration. Even intensive interventions like charter schools or voucher programs show mixed results.

This doesn't mean education is unimportant. Rather, it suggests that non-school factors like family background, peer groups, and individual characteristics matter more than educational reformers typically acknowledge. Attempts to engineer dramatic improvements through school-based interventions alone are unlikely to succeed.

Better policies would focus on identifying and scaling interventions with proven effectiveness, while maintaining realistic expectations about achievable improvements. The current tendency to promise transformational change creates cycles of reform and disappointment that ultimately undermine educational institutions.

AI poses peal alignment risks

The rapid advancement of AI capabilities creates genuine risks that deserve serious attention, even if specific doom scenarios seem unlikely. As AI systems become more powerful and autonomous, ensuring they pursue human-compatible goals becomes increasingly important and difficult.

Current AI systems already exhibit concerning behaviors: unexpected emergent capabilities, difficulty in specification of objectives, and optimization for metrics that diverge from intended outcomes. These problems will likely worsen as systems become more capable and are deployed in higher-stakes environments.

The alignment problem is fundamentally about creating systems that robustly pursue intended objectives even in novel situations. This requires advances in interpretability, robustness, and value learning that remain unsolved research problems. Market incentives alone may not prioritize safety sufficiently given competitive pressures.

However, apocalyptic scenarios requiring perfect solutions or immediate moratoriums seem overconfident. Gradual capability development allows iterative safety improvements. Many proposed risks depend on questionable assumptions about AI development trajectories. Regulatory and technical solutions can coevolve with capability advances.

Universal Basic Income Could Improve Social Welfare

Technological automation and economic inequality have renewed interest in Universal Basic Income proposals. UBI offers theoretical advantages over current welfare systems: reduced bureaucracy, eliminated poverty traps, and enhanced individual autonomy.

Pilot programs provide mixed evidence⁹. Some studies show positive effects on health, education, and entrepreneurship. Others find minimal impact on work incentives or consumption patterns. Most pilots are too small and short-term to assess long-term equilibrium effects.

The economic concerns are serious. Large-scale UBI requires substantial government revenue, likely necessitating significant tax increases. Inflation risks could erode purchasing power. Political economy dynamics might lead to benefit erosion over time.

Nevertheless, technological displacement may eventually require some form of unconditional income support. Current welfare systems create perverse incentives while failing to address technological unemployment. UBI represents one potential solution, though implementation timing and design details remain crucial.

Effective Altruism Provides Useful Framework Despite Limitations

The Effective Altruism movement applies quantitative analysis to charitable giving and career choices. Its emphasis on measurable impact, cost-effectiveness, and neglected cause areas has influenced philanthropic discourse and individual decision-making.

EA's analytical approach yields valuable insights. Global health interventions provide much higher utility per dollar than typical developed-world charities. Animal welfare concerns receive insufficient attention relative to the scale of suffering. Existential risk mitigation deserves more resources given potential consequences.

However, the framework has notable limitations. Quantification requirements may bias analysis toward easily measured outcomes. Utilitarian assumptions conflict with deontological and virtue ethics approaches. Elite capture and groupthink risks exist within EA communities.

Recent events have exposed additional problems. The FTX collapse revealed governance failures and misaligned incentives. Some EA-affiliated organizations have made questionable strategic decisions. These failures don't invalidate the entire approach but suggest the need for greater institutional humility.

Remote Work Benefits Are Sustainable Long-Term

The COVID-19 pandemic accelerated remote work adoption, creating a natural experiment in distributed employment. Initial results suggest productivity maintenance or improvement for many knowledge work roles, along with significant benefits for work-life balance and geographic flexibility.

Remote work reduces commuting costs and time, allows access to broader talent pools, and enables living in lower-cost areas. Many employees report higher job satisfaction and better family relationships. Productivity measures show minimal decline for most roles, with some studies finding improvements.

However, long-term effects remain uncertain. Collaboration, innovation, and mentorship might suffer from reduced in-person interaction. Company culture development becomes more challenging. Career advancement patterns might change in ways that disadvantage remote workers.

The optimal arrangement likely varies by role, industry, and individual preferences. Hybrid models combining remote and in-office work might capture benefits while mitigating drawbacks. But the pandemic has permanently shifted expectations and demonstrated the feasibility of distributed work arrangements.

Space Exploration Deserves Significant Investment

Space exploration and development offer potential benefits that justify substantial resource allocation, despite uncertain timelines and speculative nature of many proposed applications. The combination of scientific knowledge, technological development, and long-term survival considerations makes space investment worthwhile.

Scientific benefits include understanding of planetary formation, stellar evolution, and potential extraterrestrial life. Technological spillovers from space programs have historically generated innovations in computing, materials science, and communications. Space-based solar power and asteroid mining could eventually provide enormous economic value.

Most importantly, space capabilities provide insurance against existential risks. Asteroid impacts, supervolcanic eruptions, or other catastrophes could threaten human civilization. Establishing self-sufficient space settlements creates survival redundancy that might prove crucial over long time horizons.

However, current costs remain extremely high relative to near-term benefits. Many proposed space applications remain technically challenging or economically unviable. Terrestrial problems might deserve higher priority given limited resources and more certain impact timelines.

The question involves portfolio allocation across different time horizons and risk levels. Some space investment seems justified as long-term insurance and scientific advancement, even if specific projects fail or prove uneconomical.

Conclusion

This exercise in belief cartography reveals the vast territory of my own uncertainty. Even in areas of supposed expertise, confident positions are rare. Most beliefs cluster in the middle ranges of probability, subject to revision based on new evidence or better arguments.

This uncertainty isn't a weakness but a feature. Overconfidence remains one of the most persistent cognitive biases¹⁰, particularly in complex social domains. Acknowledging ignorance enables better decision-making and more productive discourse.

The cultural moment seems particularly hostile to such epistemic humility. Political and social pressures reward confident assertions over tentative explorations. Social media amplifies extreme positions while marginalizing nuanced views. Academic incentives favor novel claims over careful replication.

Yet intellectual honesty demands resistance to these pressures. Truth-seeking requires acknowledging the limits of our knowledge while continuing to reason carefully within those constraints. The alternative retreat into tribal certainties offers psychological comfort at the expense of epistemic progress.

The beliefs articulated here will undoubtedly evolve. Some positions will strengthen as evidence accumulates. Others will weaken or reverse entirely. A few might prove embarrassingly wrong. But the process of explicit belief formation and public commitment serves its purpose regardless of any particular belief's truth value.

In an age of epistemic crisis, modeling intellectual honesty may be more valuable than advocating for specific positions. The hope is that others will engage in similar exercises, creating a culture that values truth-seeking over truth-claiming and genuine understanding over performative certainty.

---