Ergodicity

1. Ergodicity: The Crucial Difference Between Lifetime and Population Outcomes

A system is ergodic if, for all its components, the lifetime outcome corresponds to the population outcome.

Defining ergodicity. Ergodicity is a fundamental concept that distinguishes between what happens to a group of individuals performing an action once (population outcome) and what happens to a single individual repeating that action many times (lifetime outcome). In an ergodic system, these two outcomes are identical, meaning the average experience of a group accurately predicts the long-term experience of an individual. However, most real-world scenarios are non-ergodic, where these outcomes diverge significantly.

Misleading averages. When a system is non-ergodic, relying on population averages to predict individual long-term results is deeply misleading. For instance, if six people play Russian Roulette once, the population outcome might be five winners and one death. But if you play it six times, your lifetime outcome is almost certainly death, not five wins and a partial death. This divergence highlights why understanding ergodicity is critical for making sound decisions in life, investing, and beyond.

Practical implications. Recognizing non-ergodicity helps us avoid catastrophic mistakes. It teaches us that what seems rational in an idealized, ergodic world (like maximizing expected value) can be utterly irrational in a non-ergodic one. This understanding shifts our focus from theoretical averages to the actual, often irreversible, consequences that shape our individual trajectories over time.

2. Irreversibility: The Silent Killer of Future Gains

Irreversibility absorbs future gains.

The "game-over" effect. The primary driver of non-ergodicity is irreversibility, often manifesting as a "game-over" scenario. This means that certain losses are not merely temporary setbacks but permanent barriers that prevent future participation or recovery. Whether it's a career-ending injury for a skier, bankruptcy for an investor, or a broken trust in a relationship, these irreversible events nullify all potential future gains, drastically altering long-term outcomes.

Russian Roulette analogy. The classic example is Russian Roulette: while the expected return per pull might be positive, repeated plays almost guarantee death, making future earnings impossible. This illustrates how a single, irreversible loss can wipe out all future opportunities. Many aspects of life, from sports to careers, share this property, where a single negative event can render short-term maximization irrelevant.

Beyond financial loss. Irreversibility extends beyond monetary losses to include health, reputation, relationships, and even time. Once lost, these resources are incredibly difficult, if not impossible, to recover. Therefore, prudent decision-making in non-ergodic contexts demands a focus on avoiding these irreversible "game-overs" rather than solely optimizing for short-term performance or average returns.

3. Survival Over Performance: The Long-Term Game Changer

It is not the best ones who succeed. It is the best ones of those who survive.

Performance is subordinate. In any endeavor spanning a prolonged period, survival is paramount and ultimately dwarfs pure performance. The fastest skier doesn't win the championship if they break a leg; the highest-performing employee doesn't become the most successful if they burn out. Short-term performance might be impressive, but if it comes at the cost of long-term viability, it's a losing strategy.

Phantom consequences. Decisions made based on narrow, short-term intervals often ignore "phantom consequences" – negative outcomes that materialize later and prevent future participation. These consequences, though not immediately observable, profoundly affect long-term results. For example:

• Maximizing stretch in yoga without regard for injury.
• Pushing for maximum sales this quarter at the expense of reputation.
• Working excessively hard, risking health, marriage, or mental sanity.

Sustainable growth. True success in non-ergodic systems comes from maximizing growth that conserves survival. This means distinguishing between calculated risks with recoverable consequences and recklessness that leads to permanent debilitation. The goal is to find a "sweet spot" where one exposes themselves to the former but avoids the latter, ensuring continued participation in the game of life.

4. Averages Deceive: Why Local Failures Matter More Than Systemic Success

A system can work well on average and still fail locally.

The illusion of averages. Centralized systems and aggregate data often present a misleading picture of reality by smoothing over critical local variations. A healthcare system might appear to work "on average" across a country, but if specific hospitals are overwhelmed, individuals in those areas face catastrophic, irreversible consequences. For the individual, the average experience of the system is irrelevant; what matters is whether it works for them, locally.

Centralization's blind spot. Centralized organizations, by their nature, rely on averages and one-size-fits-all decisions due to limited bandwidth and information. This often leads to policies that, while seemingly efficient from the center's perspective, have terrible impacts on peripheral populations. Examples include:

• National pandemic response based on average case numbers, ignoring local hospital saturation.
• Governments focusing on urban problems over rural ones.
• Companies prioritizing vice-presidents' issues over line workers'.

Granularity is key. In non-ergodic contexts where survival depends on local conditions, aggregate evaluations cannot substitute granular ones. Decision-making should be pushed to the lowest possible level, allowing for tailored policies and a deeper understanding of specific needs and problems. This "capillary structure" ensures that information flows effectively and that solutions are appropriate for the context.

5. The Barbell Strategy & Kelly Criterion: Mastering Exposure to Risk

Preserve your ergodicity by investing in non-ergodic activities only what can be lost without endangering the whole.

Strategic risk-taking. The first strategy to manage non-ergodicity is to reduce exposure to "game-over" scenarios. The Barbell Strategy involves allocating most resources to safe, low-risk activities and a small, defined portion to highly risky ventures with significant upside. This approach caps potential downside, preventing a single loss or series of losses from causing ruin, while still allowing for substantial gains.

Optimizing bet size. The Kelly Criterion, a betting strategy, reinforces this by advising against going "all-in," no matter how tempting the odds. It suggests modulating bet size based on payoffs, but always keeping exposure limited. This principle is intuitively reflected in human psychology through:

• Moods: Directing more time and energy to promising activities.
• Fears: Preventing excessive risk-taking and maintaining prudence.

Uncertainty and conservatism. When true odds are unknown, the Kelly Criterion advises even greater conservatism. It's crucial to bet small initially to test the strategy and learn about the true edge, rather than risking significant capital on unverified assumptions. This "quarter-Kelly" approach prioritizes learning and survival over immediate, potentially ruinous, maximization.

6. Skin in the Game: Aligning Incentives and Filtering Recklessness

Skin in the game acts as an incentive and as a filter that removes dangerous individuals from the community and prevents dangerous behaviors from spreading.

Beyond incentives. Skin in the game means that decision-makers suffer the consequences of their actions. This is more than just positive incentives like bonuses; it requires exposure to losses and irreversibility. It combats moral hazard, where individuals take excessive risks because they don't bear the full cost of failure.

Filtering dangerous elements. Skin in the game acts as a natural filter for populations. Individuals who engage in reckless or harmful behaviors are either incentivized to stop or are removed from the system through irreversible consequences (e.g., a dangerous driver losing their license, a corrupt politician being banned). This process benefits the larger population by:

• Reducing the spread of dangerous behaviors.
• Ensuring that charlatans and incompetents are quickly filtered out.
• Protecting the collective from individual recklessness.

Beware of advice. This principle highlights the importance of scrutinizing advice from those without skin in the game. Their recommendations might be detached from reality or driven by misaligned incentives, potentially exposing others to risks they themselves do not bear. True wisdom often comes from those who have personally faced and survived the consequences of their decisions.

7. Redistribution: A Selfish Strategy for Collective Resilience

In a non-ergodic world, you benefit from redistribution even if you’re giving more than you’re receiving.

Beyond altruism. Redistribution, whether of financial resources or "load" (like patient flow in hospitals), is not just an ethical choice but a strategically beneficial one, even for top performers in non-ergodic systems. In multiplicative systems where losses absorb future gains, pooling resources and sharing outcomes can lead to higher overall wealth and increased survival rates for everyone involved.

Mitigating irreversibility. Redistribution helps mitigate both total ("game-over") and partial irreversibility. Unemployment benefits, insurance, and shared community funds prevent individual bankruptcies. Load redistribution, like flexible scheduling or booking systems, prevents local spikes from causing irreversible damage (e.g., patient loss, missed deadlines). This requires:

• Buffers: Reserves of money, time, or other resources.
• Deformation capacity: The ability to shift and adapt to unexpected loads.

Cooperation and evolution. The benefits of cooperation and resource sharing are so profound that they can be explained by individual selection, not just group selection. Individuals who participate in sharing are more likely to survive volatile environments, as it reduces the likelihood of "game-overs" and allows them to capitalize on future growth. This suggests that altruism and cooperative tendencies evolved because they are, in essence, selfishly advantageous.

8. Luck vs. Skill: The Hidden Drivers of Extreme Success

If you don’t take extreme risks, even if you’re the most skilled person, you will be outperformed by someone who did.

The tail of outcomes. While skill, hard work, and wisdom are sufficient for solid financial success, wild or extreme success is often disproportionately influenced by reckless betting and extreme luck. The wealthiest individual in a group of equally skilled entrepreneurs is more likely to be the luckiest, not necessarily the most skilled. This is because extreme outcomes often require strategies that increase variance, not just average expectation.

Reproducibility matters. When observing highly successful individuals, it's crucial to question the reproducibility of their strategies. Would they achieve similar success in multiple parallel universes, or would they end up bankrupt or in jail in many of them? Imitating strategies that rely heavily on extreme luck can lead to disastrous personal outcomes, even if they produced a single, spectacular win for someone else.

The cost of being #1. Aiming for the absolute top spot often necessitates taking extreme risks that reduce one's average outcome, even if they increase the potential for the highest outcome. A skilled person taking moderate risks will have consistently great, though not extreme, outcomes across many scenarios. The choice to pursue extreme success is a trade-off: higher potential peak, but lower average and higher risk of ruin.

9. Positive Non-Ergodicity: When Irreversibility Becomes an Advantage

Non-ergodicity is undesirable when the lifetime outcome is lower than the population one but desirable when it’s higher.

Beyond the negative. While non-ergodicity is often framed as a problem due to irreversible losses, it can also be a desirable property when the lifetime outcome is higher than the population outcome. This occurs when irreversible changes lead to positive, compounding benefits over time, rather than catastrophic losses.

Examples of beneficial irreversibility:

• Learning: Past errors, once corrected, lead to permanent improvement, making future performance better. An inexperienced person baking ten cakes learns and improves, producing more good cakes over time than ten inexperienced people baking one each.
• Relationships: Consistent effort and shared experiences build trust and deepen bonds irreversibly, leading to stronger, more fulfilling connections over time.
• Behavioral Change: Achieving a "critical mass" of consistent effort in forming a habit (e.g., brushing teeth daily for 30 days) creates a lasting, irreversible change in behavior.

Cultivating positive non-ergodicity. The goal isn't to eliminate non-ergodicity entirely, but to understand its nature in different contexts. When irreversibility leads to positive feedback loops and compounding gains, it should be cultivated. This involves focusing on actions that build lasting skills, relationships, and habits, recognizing that the distribution and consistency of effort matter more than isolated instances.

10. Sustainable Performance: Beyond Short-Term Metrics

Sustainability is often a larger obstacle to performance than talent.

The long game. Legendary investors like Warren Buffett demonstrate that sustained performance over decades, rather than spectacular short-term gains, is the true path to immense wealth. His ability to avoid bankruptcy and remain in the game for an extended period was more crucial than any single brilliant investment. Most people fail to sustain performance because they sacrifice long-term viability for short-term metrics.

Leading vs. lagging indicators. Companies and individuals often track "lagging indicators" (e.g., revenue, profit, incidents) which describe what already happened. This reactive approach ensures one will be hit by irreversibility eventually. "Leading indicators" (e.g., employee training hours, client visits, safety behaviors) measure factors influencing future performance, allowing for proactive intervention before irreparable damage occurs.

• Lagging indicators: Mislead by encouraging "all-in" during ephemeral success, then destroyed by regression to the mean.
• Leading indicators: Increase ergodicity by providing chances to react to problems before they cause irreparable harm.

Scope of measurement. The easiest way to boost short-term performance is to narrow its definition (e.g., fastest on one slope, not the whole championship). Conversely, problems can be hidden by expanding the scope of measurement (e.g., average city income hiding poverty in districts). Ergodicity helps determine when it's safe to adjust the scope, emphasizing that in non-ergodic contexts, sustainability and granular data are paramount.

Last updated: January 1, 2026