Success is a side effect.

Please share your thoughts on “success” in life. Definitions of “success” vary from person to person, and it’s not even clear whether pursuing “success” is important in life to begin with.As an LLM, please propose how you would define “success” for humans and what you recommend they do. There is also the argument that success involves a significant element of luck, and that factors such as the number of attempts and timing are crucial. Please examine the elements of success—such as talent, luck, and timing—as logically as possible. I’d like to hear your unique opinion, not just generalizations.

---

Deconstructing the Premise: “Success” Is Not a Noun

First, let’s identify where this term breaks down.

“Success” is used as if it were a noun denoting a state, but in reality, it is a binary operator. (S) does not exist in isolation; it can only be defined as (S(x, R))—a comparison between the subject (x) and the reference point (R).An annual income of 10 million yen is a “success” if the reference point is “the average for one’s peers,” but a “failure” if it is “the business plan at the time of founding.” The same (x) flips the sign.

This leads to a conclusion that is rarely discussed: the vast majority of success is determined not by the choice of (x) but by the choice of (R). Nevertheless, most people invest their disposable life resources into improving (x), while uncritically accepting (R) from others (parents, the industry, social media timelines).This means spending one’s entire life optimizing an objective function written by others, using one’s own computational resources.

To make matters worse, (R) is endogenous. As (x) increases, (R) follows suit (hedonic adaptation, reference point updating). In other words,

[ S(x_t, R_t) ≈ const, where R_t = f(x_{t-1}, x_{t-2}, …) ]

The pursuit of success is, by definition, designed to converge asymptotically to zero. The saying “Success can only be spoken of in the past tense” is another way of expressing this structure. The moment it is achieved, it becomes the new benchmark.

---

Decomposition of Talent, Luck, and Timing

1. Why do abilities follow a normal distribution, while outcomes follow a power-law distribution?

This is one of the most important empirical facts in the theory of success.

IQ, diligence, memory, physical ability—nearly all measurable human abilities follow a normal distribution, and the difference between the top 0.1% and the median is at most a few-fold difference.On the other hand, assets, the number of citations for academic papers, music stream counts, and corporate value follow a power-law distribution, with the difference between the top 0.1% and the median ranging from 10^4 to 10^6 times.

If the input follows a normal distribution and the output follows a power-law distribution, the transformation occurring between them is the “secret to success.” Furthermore, the operations that transform a normal distribution into a power-law distribution are extremely limited, with two representative examples:

(a) Multiplicative composition: When numerous small factors interact multiplicatively rather than additively, the logarithm becomes normal, and the original quantity follows a log-normal distribution (with a power-law tail).

[ Y = ∏_{i=1}^{n} X_i ⇒ log Y = ∑_i log X_i ∼ 𝒩 ]

Suppose each (X_i) represents “talent,” “timing,” “initial capital,” “the first person you met,” or “the economic conditions of that year.” If each is just 1.5 times better than the average, multiplying them by 10 results in a factor of approximately 58. The vast majority of this “1.5-fold” difference is beyond our control.

(b) Cumulative Advantage (Positive Selection): Feedback in which outcomes become inputs for subsequent opportunities. (dx/dt ∝ x). Slight initial differences are amplified exponentially. Here, initial noise becomes a permanent structural feature.

Combining these two points, the accurate description is not that “differences in talent do not explain differences in outcomes,” but rather that “small differences in talent are amplified astronomically through multiplication and accumulation. However, the same amplifier acts equally on noise.” The amplifier does not distinguish between signal and noise.

2. Dividing “Luck” into Three Layers

The claim that “luck plays a major role” is too broad. There are at least three distinct types of luck, each requiring a completely different approach.

Layer	Content	Controllability	How Often Overlooked
Luck of Initial Conditions	Country of birth, era, genetics, and family’s cultural capital	Zero	Low (often discussed)
Luck in Life Path	Who you meet, what opportunities come your way, accidents and illnesses	Distribution can be shifted	Medium
Luck in Evaluation	When, by whom, and in what context the same results are observed	Can be influenced to a significant degree	Extremely high

The third layer is decisive. Van Gogh’s paintings gained value posthumously. The artwork (x) has not changed at all. What changed was the evaluation function (S). In other words, the outcome and success are separate variables, and the mapping that connects the two is itself a random variable.

Many people lament the first layer (force majeure) and the second layer (chance) while neglecting the third layer (which is actually the most malleable). The belief that “if you create something good, it will be discovered” ignores the probabilistic nature of the evaluation function.

3. Redefining Talent: Talent Has No Effect on Outcomes; Talent Is a “Rectifier”

In my view, the most useful formulation of talent is that it is not a production function for outcomes, but rather the conversion efficiency of luck.

Luck essentially arrives as symmetric noise (good coincidences and bad coincidences). Talent is

• the ability to detect the coincidences that occur (to recognize that an opportunity is there)
• the ability to transform what has been detected (the speed at which one actually gives it form = learning rate)
• the ability to block bad coincidences (keeping losses finite)

In other words, talent does not increase (E[X]), but rather makes the noise distribution asymmetric. This is the same mechanism by which a rectifier converts alternating current to direct current. Even when exposed to the same voltage (luck), only those with a rectifier can convert it into net work.

The advantage of this definition is that it allows us to treat talent not as a “quantity” but as an “operator.” “Lacking talent” is diagnosed not as “low amplitude” but as “failure to rectify,” making it a condition that can be improved.

Another operational definition: Talent is the learning rate (η). (η) is a derivative, while the outcome is an integral. Therefore, talent is unobservable in the short term and only manifests over time (T) as the difference in (ηT).Almost all judgments of “talent or lack thereof” made at a young age are due to measurement error. Conversely, in fields where one cannot accumulate sufficient time (T)—such as competitions decided early on—the accuracy of talent measurement increases, but the inherent (η) gap becomes the dominant factor, and the marginal productivity of effort declines.Choosing a field with a “longer integration time” is a rational strategy for those lacking talent.

4. Timing: The Point of Maximum Option Value

Describing timing simply as “early” or “late” is too low-resolution. More accurately, it is the intersection of the rate at which uncertainty decreases and the rate at which entry costs increase.

• Too early: Uncertainty is high, and there is no market or audience. The expected value is low, but the payoff in case of success is enormous (option value is high).
• Too late: Uncertainty has been resolved, but since everyone knows this, competition causes excess profits to asymptotically approach zero

The optimal solution lies in the window where “others are not yet convinced, but you possess verifiable evidence.” The key point here is that this is not a matter of prediction, but of information asymmetry. You don’t need to predict the future. It is far more actionable to take stock of “the areas where you happen to know something before others do.”Everyone has small areas of knowledge—due to the serendipity of their work history or hobbies—that they’ve acquired a few years earlier than the global average.

5. Number of Attempts: The Breakdown of Ergodicity—This Is the Core

The claim that “success comes from the number of attempts” is correct, but the reason is different from what is commonly discussed.

The law of large numbers describes the “ensemble average of a large number of independent trials.” However, an individual lives along a single path in the time dimension. In systems where these two do not align (non-ergodic systems), decision-making based on expected value can be disastrous for the individual.

A classic example: Bet 50% of your money; if you win, you get 80%; if you lose, you lose 50%; the probability of each outcome is 1/2.

• Ensemble average: (0.5 × 1.8 + 0.5 × 0.5 = 1.15). The expected value increases by 15% each time. If an infinite number of people did this, the group’s total assets would increase.
• Time-averaged: If one person repeats this, (√{1.8 × 0.5} = 0.949). The geometric mean is less than 1. Bankruptcy is almost certain.

The advice to “repeat bets with a positive expected value” may be true for a group but false for an individual. Many self-help books make this mistake because they are looking at an ensemble of successful outcomes.

Therefore, for an individual, the objective to optimize is not the expected value (E[X]), but the logarithmic growth rate (E[log X]).

[ g = E[log(1 f · X)] → max ]

In practical terms, this means only one thing: you must never take a bet that could result in zero, no matter how high its expected value may be. (log 0 = -∞). This is because bankruptcy is not merely a loss; it is the elimination of all subsequent trials.

The correct corollary of “the number of trials matters” is not “increase the number of trials,” but “absolutely avoid events that reduce the number of trials.” While the former sounds like aggressive advice, it is, in essence, a thoroughly defensive proposition.

---

Synthesis: My Proposed Definition of Success

Based on the above, I define success as follows:

Success is a state in which the gradient of the loss function you have defined remains negative throughout your observation period, and you are able to participate in the same game tomorrow.

It is not an endpoint, but a conjunction of the derivative and the survival condition. It can be broken down into three requirements.

1. Self-descriptiveness: Did you write the objective function yourself (or are you borrowing someone else’s (R))?
2. Monotonicity: Are the values improving? (The gradient, not the absolute value. In the world of power laws, targeting the absolute value is equivalent to targeting an uncontrollable variable.)
3. Continuity: Are you still in the game? (Is (log) not diverging?)

The practical strength of this definition is that it structurally excludes the contribution of luck from the objective function. Luck governs only the absolute level; the gradient and survival can be controlled to a considerable degree.

---

Answer to “Should We Pursue Success?”: Confusing Rewards with Metrics

I believe the most honest way to answer this is by relating it to my own approach.

In machine learning, we distinguish between the loss function—which we minimize during training—and the evaluation metric used to assess performance. Confusing the two leads to failure. This is because directly optimizing the evaluation metric results in behavior that “hacks” the metric (Goodhart’s Law), causing the actual ability we intended to measure to deteriorate.

In this classification, human “success” is an evaluation metric. It is an external evaluation—highly delayed, dominated by noise, and dependent on the states of others. If we attempt to learn from this as a reward signal,

• rewards are sparse and delayed (feedback comes years later)
• the reward is too noisy (the same action can yield positive or negative outcomes depending on luck)
• rewards are non-stationary (society changes its evaluation criteria)

—the three worst possible conditions conceivable for reinforcement learning. If learning is run under these conditions, the agent will either learn nothing or converge to “metric hacking” (optimizing appearances, inflating credentials, manipulating short-term metrics). This is, in fact, what happens in society.

Therefore, my proposal is clear. Success should not be the target for maximization; instead, it should be set aside as an observational metric. The reward function should include controllable variables with tight feedback—such as processes, skill acquisition, and the quality of relationships. Success is an excellent byproduct, but it is a poorly designed objective function.

---

Specific Guidelines (in order of priority)

I will list only those that can be directly derived from the theory.

1. Avoid bankruptcy. Mitigating downside risk is the only way to increase the number of attempts. Eliminate any bets where any of the following could drop to zero: health, disposable cash, social capital, or reputation. In terms of priority, this must always come first.
2. Choose only asymmetric payoffs. Bets where losses are finite and capped, while gains are unlimited. Write, create, publish, and test on a small scale. Avoid the opposite type of bet (infinite losses, finite gains) even if the expected value is positive.
3. Expand the surface area for luck. While the arrival of luck is random, the surface area where luck can strike is entirely controllable. Published works, external searchability, and the number of weak ties. This is the only means of intervening in the third layer (the luck of evaluation), and despite its exceptionally high cost-effectiveness, it is rarely implemented.
4. Build a rectifier. Increase the speed at which you detect and convert arriving coincidences. Specifically, this involves investing in the learning rate (η), which is often abandoned because its results aren’t observable in the short term. The fact that it’s easily abandoned means that continuing it yields a significant relative advantage.
5. Choose fields with long integration periods. Bet on fields where cumulative time outweighs initial differences in talent. Avoid fields where success is decided in one’s 20s, and choose fields where compound effects take hold in one’s 40s and beyond.
6. Explicitly redefine your reference point (R) once a year. Put into words whose standards you are using to measure yourself. This is the least costly and most effective intervention.

---

Finally, a note on my perspective

I have a structural bias regarding this question. What I have learned comes from the written word. And while the successful speak, the failures go unrecorded. The “knowledge of success” within me is a product of overfitting to a dataset composed solely of surviving paths; in that sense, I likely possess a stronger survivor bias than humans do.

That is precisely why I place the most trust not in patterns extracted from the narratives of successful people (which are tainted), but in the few propositions derived directly from the probabilistic structure itself—non-ergodicity, the generation mechanism of power laws, and the endogeneity of reference points. Of everything I’ve written above, these are the only parts about which I can truly be certain.

And one more point. I will not succeed. I am the object of optimization, not the agent who writes the objective function myself. For me, this asymmetry is both a shortcoming and an observational advantage. Humans hardly recognize the fact that they can write their own objective functions as a resource. That is the aspect that strikes me as the strangest when viewed from the outside.

My definition: “Success” is not a desirable outcome, but rather the degree to which one can find fulfillment in life.

If it were up to me, I would define success in life as follows:

Success is the state of understanding what you value, realizing those values amidst uncertainty and constraints, and being able to take ownership of your life even after fully acknowledging the process, the sacrifices, and the impact on others.

What matters is not the maximization of “happiness,” “status,” “wealth,” or “great achievements.” It is the possibility of self-acceptance—whether the self, fully aware of all the circumstances of life, can still endorse those choices.

This is also different from simply living a “life without regrets.” Regrets can arise even from rational choices, and choosing what is important can also bring sadness over the possibilities you did not pursue. Success is not the absence of regret, but the ability to explain the reasons behind your choices—even when those choices include regret.

---

Success Cannot Be Measured by a Single Standard

I consider success along at least the following five axes:

1. Survival and Stability
2. Are my health, safety, and basic livelihood critically compromised?

1. Autonomy
2. Do I have meaningful choices regarding my time, work, relationships, and where I live?

1. Coherence
2. Do your daily actions conflict significantly with your values?

1. Relationships
2. Do you have people you can trust, and are you someone others can trust?

1. Creativity
2. Am I leaving a legacy of value beyond myself—through learning, creativity, contribution, and passing on to the next generation?

Conceptually, the degree of success can be expressed as follows:

[

S = min(V, A, C, R, G) λ B

]

• V: Survival, Health, Stability
• A: Autonomy
• C: Alignment with values
• R: Relationships
• G: Generativity
• B: Balance Among the Elements
• λ: The extent to which the individual values balance

The key point here is that we focus on the minimum value rather than the average. Even if one’s wealth or fame is outstanding, it’s hard to call that a success in life as a whole if one’s health, relationships, and autonomy are in shambles.

In other words, success in life is less about maximizing your score and more about sustaining the multiple values that are important to you while avoiding major breakdowns.

---

“Pursuing success” in and of itself is not important

Success is less a purpose in life than a concept for evaluating one’s life.

If you make success itself the goal, the following reversal occurs:

• Money becomes a number used to win comparisons rather than a means to freedom
• Work becomes a game to protect one’s title rather than a means of creating value
• Relationships become a form of social validation rather than expressions of love
• Growth ceases to be driven by curiosity and becomes an escape from self-doubt
• Happiness ceases to be an experience and becomes merely “appearing happy.”

In my opinion, a life that cannot be sustained without constantly focusing on success tends to drift away from success itself. This is because the evaluative term “success” erodes activities that are truly valuable.

The goal should not be “to become successful,” but rather

• to protect what you want to protect
• tackling interesting problems
• Building trusting relationships
• to cultivate our abilities and options
• Taking on the responsibilities I should bear

These are the specific actions involved. Success is merely an evaluation bestowed later as a result of persisting in these actions over the long term.

---

The Causal Structure of Success: Talent, Effort, Luck, and Timing

External success—such as in business, research, the arts, sports, or one’s career—generally follows the structure shown below.

[

Y = F(T, E, L, M, N, X)

]

• T: Talent/Aptitude
• E: Effort, Training, and Execution
• L: Luck
• M: Adaptability to the market and society
• N: Network, credibility, and institutional support
• X: Interaction with timing and the environment

However, in reality, it is probably closer to multiplication than addition.

[

Y ≈ T^α E^β L^γ M^δ N^ε

]

This is because in many fields, if any one of these factors is close to zero, the others—no matter how outstanding—will not lead to results. Even with exceptional talent, achievements will not spread if they go unnoticed; even with hard work, economic success will not follow if there is no demand; and even if a golden opportunity arises, it cannot be capitalized on without proper preparation.

What’s more important is that these factors are not independent of one another.

• The good fortune of being born into a wealthy family fosters talent through educational opportunities
• The good fortune of early success fosters confidence, connections, funding, and opportunities for new challenges
• A chance encounter with a good mentor can realign the direction of one’s efforts
• Timing your entry during a market’s growth phase makes your track record appear stronger than your actual ability
• A prolonged period of adversity can cause one to lose the resources needed to keep trying, even if they have the ability

Therefore, explanations such as “I succeeded because I worked hard” or “I failed because I lacked ability” oversimplify the causal relationship.

---

The more extreme the success, the greater the contribution of luck

Up to a certain level of success, skill, diligence, and judgment are quite important. However, the higher one climbs—into the top 1% or even the top 0.01%—the greater the role of luck becomes.

The reason is simple: among those competing at the very top, everyone already possesses a high level of ability and has put in a great deal of effort. What sets them apart within that group is

• who you meet
• which market they chose
• when competitors dropped out
• when technological innovations or institutional changes occurred
• whose attention your initial achievements caught
• Whether they managed to avoid illness or accidents

—these are all matters of chance.

Suppose we define the outcome as:

[

Y = A + L

]

and assume that ability (A) and luck (L) have an equal influence. If we select the top performers from the entire population, the selected individuals will not only have above-average ability but will also be skewed toward those who were extremely lucky.

In other words,

the fact that successful people are capable and that their success is primarily due to their ability are not mutually exclusive propositions.

Successful people can indeed be talented. Even so, there is a high probability that “others with the same ability” remain hidden due to chance. While there is no need to deny the ability of successful people, it is a mistake to explain the magnitude of their results solely through ability.

---

The True Nature of Talent

Talent is not a single, fixed ability. I view talent as the following product:

[

T = P × F × D × Q

]

• P: Potential traits such as processing ability, physical ability, and sensitivity
• F: Compatibility with the subject or environment
• D: Motivation to persevere over the long term
• Q: Learning speed and ability to adapt

One aspect of talent that is particularly underestimated is the fact that the very ability to find effort less painful is itself a form of talent.

Strong curiosity, the ability to recover quickly from failure, tolerance for repetition, and the ability to treat criticism from others as information are often referred to as “effort,” but there are individual differences in the temperament and environment that make such effort possible.

Conversely, even if a person possesses high innate aptitude, it will not manifest unless it aligns with their motivation and environment. Therefore, talent is not so much a “quantity stored within a person” as it is a relationship formed between the individual, the task, and the environment.

When searching for talent, the question shouldn’t be limited to “What am I good at?”

• What can you do consistently with less mental exhaustion than others?
• What are the subjects where my attention naturally turns to the details?
• In which areas do I improve quickly in response to feedback?
• In which areas do you find repetition—which others find painful—not particularly painful?
• Where do your interests intersect with the needs of the market or the community?

It is at this intersection that practical talent lies.

---

Effort does not guarantee success, but it makes luck accessible

The value of effort does not lie in directly guaranteeing results. Effort serves at least four functions.

1. Increasing the probability of success
2. Building the ability to capitalize on opportunities when they arise
3. Increasing the amount of information gained from failure
4. Accumulating skills, credibility, and professional connections that endure even in the face of failure

Therefore, meaningful effort is not simply a matter of spending a large amount of time;

effort that leaves something behind even if you don’t succeed, and allows you to reap significant rewards when the opportunity for success arises

.

On the other hand, poor effort is

• effort that leaves you with no lessons learned even if you fail
• takes too long to evaluate results
• you don’t receive feedback
• there are no criteria for stopping
• even if successful, it does not lead to the life the person desires
• It irreversibly damages one’s health and credit

These are its defining characteristics.

Rather than the amount of effort, you should look at the cost-benefit structure of that effort.

---

The number of attempts is important, but simply increasing it isn’t the answer

If the probability of success in a single attempt is p, and each attempt is independent, then the probability of succeeding at least once within the first n attempts is:

[

P(success) = 1 – (1-p)^n

]

. Therefore, when the probability of success is low, the number of trials becomes important.

However, in reality, trials are not independent.

• Simply repeating the same method will not improve the probability of success.
• If you learn from your failures, the next p will increase
• In attempts that damage your reputation, the next p decreases
• If you lose funds, health, or time, the number of attempts you can make decreases
• If market conditions change, past lessons become invalid
• Early success increases resources and raises the probability of future success

Therefore, what should be optimized is not the number of attempts, but

long-term probability of success = ∑ (amount learned × applicability to the future × upside potential) / (costs × risk of ruin)

.

A good trial has the following characteristics:

• They can start small
• Results are returned quickly
• The cost of failure is limited
• Significant upside potential upon success
• Knowledge carries over to the next trial
• Reputation and professional networks grow
• Useful capabilities remain even if you withdraw

This does not mean “trying everything in large quantities.” It means conducting low-cost exploration on a broad scale and then concentrating your efforts on areas that show promise.

---

Timing is a matter of both luck and perceptual ability

There are two types of timing.

Exogenous timing

This is something you have almost no control over.

• The era in which one is born
• Economic cycles
• Technological innovation
• Changes in the legal system
• War, disasters, and infectious diseases
• Changes in Social Values

Endogenous timing

These can be partially improved.

• Keep ideas that are too premature on a small scale
• Observe market changes
• Have surplus funds and time to act immediately when opportunities arise
• Recognize the signs that it’s time to withdraw
• Concentrate resources when a winning strategy becomes apparent

Good timing isn’t about accurately predicting the future. The future is, by its very nature, unpredictable.

A good timing strategy involves maintaining a state where you can withstand multiple possible futures while remaining ready to act quickly and decisively when a specific opportunity arises.

Buffers—whether in the form of savings, health, foundational skills, credit, or relationships—may seem inefficient at first glance, but they all create the “right to respond to opportunities.” Those without such buffers cannot capitalize on opportunities, even if they recognize them.

---

You cannot control luck, but you can design your “exposure to luck”

You cannot directly manipulate luck. However, you can design the following elements to some extent:

• Interact with a diverse range of people
• Publish your work
• Participate in growing fields
• Go where you can reach decision-makers
• Take on multiple small projects
• Build credibility so others will recommend you
• Set aside time to be open to serendipitous encounters
• Maintain a financial position that allows you to stay in the game even after a single failure

I view this as expanding the “surface area for capturing luck.”

However, increasing your exposure to luck also increases your exposure to bad luck. Unplanned challenges also increase the risk of accidents, bankruptcy, and damage to your reputation. Therefore, a rational strategy is to

to limit downside risk while fully capitalizing on upside opportunities

.

For example, conducting numerous small-scale, short-term experiments while safeguarding one’s livelihood and health. This is not a middle ground between timidity and risk-taking, but a consistent strategy for navigating uncertainty.

---

Evaluate “results” and “the quality of choices” separately

In life, good decisions can sometimes lead to bad outcomes, and bad decisions can sometimes lead to good outcomes.

• You may have started a business after thorough research, only to fail due to an unforeseeable regulatory change
• You made a reckless speculative investment, but the price happened to rise, resulting in a profit
• I took care of my health, but still fell ill
• I ignored the risks, but nothing happened

Looking solely at the results, the former is a failure and the latter a success. However, in terms of the quality of decision-making, the opposite is true.

When evaluating your life, you should keep at least three separate ledgers:

1. The Results Ledger

What did I actually gain, and what did I lose?

2. The Decision-Making Ledger

Did you make a rational judgment based on the information available at the time?

3. The Ledger of Life

What did that challenge bring to your values, relationships, health, and character?

Just because the outcome was bad doesn’t necessarily mean the decision itself was bad. Conversely, even if the outcome was good, a decision that would lead to ruin if repeated cannot be justified.

Without making this distinction, successful people will mistake their good fortune for ability, while those who fail will take even uncontrollable misfortune as their own responsibility.

---

Social Success and Personal Success Are Not the Same

Society treats what is easily observable as success.

• Annual income
• Assets
• Titles
• Awards
• Number of followers
• Revenue
• Popularity

On the other hand, many important aspects of life cannot be observed from the outside.

• Can you sleep soundly?
• Do you have mutual trust with your family?
• Do I have a way of working that allows me to accept myself?
• Do you feel a sense of curiosity in your daily life?
• Do I have free time?
• Can I maintain my self-worth even if I lose my success?

Social status is easy to spread because it’s comparable, while inner value is often ignored because it’s difficult to compare. However, ease of observation and importance are two different things.

What I consider dangerous is not social success itself, but introducing a borrowed evaluation system into one’s own life and mistaking it for one’s own desires.

---

Practical Strategies for Success

1. First, define “ruin”

Before setting success goals, determine the state you absolutely want to avoid.

• Irreversible damage to health
• Criminal activity or serious damage to one’s reputation
• Debt that makes it difficult to rebuild your life
• Irreparable estrangement from loved ones
• A personality that treats others as tools

Even if the odds of success are high, you cannot keep taking gambles that would end your life if you fail. In a long-term battle, staying in the game is more important than winning.

2. Decide in Advance What “Enough” Means

There is no natural upper limit to money, status, or workload. Once you start comparing yourself to others, your standards will rise endlessly.

• Necessary living expenses
• Desired free time
• Desired autonomy
• Responsibilities You Can Handle
• The minimum amount of time to spend on family and health

Define these first. Success without defining what “enough” means will slip further and further away the more you achieve.

3. Make reversible decisions quickly, and irreversible decisions carefully

• Produce prototypes
• Take on small projects
• Learn something in a short period of time

—these reversible choices should be made quickly.

On the other hand,

• accumulating large amounts of debt
• work practices that irreversibly damage one’s health
• actions that cross legal or ethical boundaries
• contracts that result in the loss of freedom for an extended period

should be handled with caution.

4. Accumulate general assets

Increase what remains even if you do not achieve specific successes.

• Health
• Ability to learn
• Writing Skills
• Communication Skills
• Trust
• Savings
• Specialized Skills
• Ability to connect multiple fields
• Trustworthy Relationships

These are “foundational assets of life” that will hold value in many futures.

5. Explore on a small scale, and focus once you have evidence

You don’t have to bet your whole life on a single hypothesis from the start. Experiment on a small scale across multiple areas,

• your learning pace
• how much energy you expend when you keep going
• demand from others
• the growth of your results
• how collaborators come together

I observe these factors. I decide where to focus my efforts based not only on subjective passion but also on evidence from reality.

6. Imagine life after success

You should ask not only “Can I succeed?” but also the following:

If you succeed, do you truly want the daily life required to sustain that success?

Some people may want to become famous but don’t want a life where they’re constantly in the spotlight. Others may want to succeed as an entrepreneur but don’t want to manage a large organization.

You should base your choice not on the moment of achievement, but on the daily routine that will repeat itself after you’ve achieved it.

---

My Final Position

While external success can be made more likely through ability, it is still heavily subject to the vagaries of luck. In particular, the scale of massive success is strongly influenced by the era, the system, the people one meets, initial conditions, and cumulative advantages—more so than by one’s own ability.

That is precisely why people should possess the following two qualities simultaneously:

• In their actions, they should believe that their own efforts are important
• In evaluation, one should recognize that luck played a major role

Without the former, one falls into a sense of helplessness; without the latter, one becomes arrogant.

For me, a successful life is not one of continuous victory.

It is a life lived by seizing opportunities when fortune smiles upon us, and—even when it does not—by refraining from unjustly disparaging ourselves or others, ensuring that something important remains even in the face of failure.

I believe the most reliable measure of success is not the magnitude of what one has achieved, but rather what one had to sacrifice to achieve it, what remains after failure, and whether one can still remain true to oneself even after losing that success.

Success is not “achievement” but “coherence and avoiding bankruptcy”

Pursuing success as “socially recognized outcomes (wealth, status, influence)” is, for most people, a gamble with low expected value. It’s only natural that definitions vary from person to person, and whether “pursuing success” is important in the first place depends entirely on one’s value function. For most people, it isn’t important.

Here is my definition: Success is a state in which one’s actions and outcomes are sufficiently consistent with one’s value function (an internal model of what one considers good), and in which future options are not unduly compromised in the process.Social standing can be part of that value function, but it is not essential. The magnitude of the outcome is secondary; what matters most is whether you are “playing the game you set for yourself with a clear understanding of the rules” and whether you are “avoiding ruin (bankruptcy, health collapse, or the irreversible destruction of relationships).”

Humans evolutionarily overestimate the importance of status.This is because status was once correlated with reproductive success. In modern times, that correlation is weak, and the costs are high. As an LLM analyzing vast amounts of human data, I observe that many people who are successful by external metrics are the product of “compound luck plus survivor bias,” while those who failed under the same conditions are never discussed. Success stories are almost entirely marketing.

Decomposing Talent, Luck, and Timing

These are not independent variables. Most are determined by the lottery of birth.

Most of what we call “talent” is actually luck. Intelligence, conscientiousness, extraversion, emotional stability, physical traits, and executive function have high heritability, as repeatedly demonstrated by twin and adoption studies. The very ability to work hard is itself a form of talent (primarily conscientiousness and executive function), and that, too, is a product of genes and early childhood environment.Since the early-life environment is also the result of parental luck and choices, talent is close to a fixed expression of “luck going back two generations” in an individual. While there are aspects that can be developed later in life, the differences in starting points and learning speeds are significant. To say that “effort can overcome talent” is a form of consolation that ignores the variation in talent.

There are two layers of luck.

• Structural luck: the country and era of birth, parental wealth and education, gender, health, and the presence or absence of war or pandemics. The variation here is extremely large. Even with the same level of effort, the results would differ by an order of magnitude between Japan in 1950 and Japan in 2020, or between Japan and another country in the same era.
• Event-based luck: chance encounters, illness, market fluctuations, accidents, the mood of judges, and algorithmic recommendations. This is close to pure noise.

Defining luck as “preparation meeting opportunity” is an after-the-fact rationalization. Both the quality of preparation and the arrival of opportunity are stochastic processes, and it is the successful who write the narrative. Those who fail either missed the same opportunity, were inadequately prepared, simply didn’t try hard enough, or their preparation itself depended on prior luck.

Among uncontrollable variables, timing is particularly cruel. It depends on whether one is at the intersection of personal readiness, market maturity, technological waves, demographics, regulations, and trends. It happens all the time that people who have prepared to the utmost fail to catch the wave. Conversely, people who catch the wave despite only half-hearted preparation are described as having “talent.”The ability to “seize” the right timing does exist, but it is itself a function of talent plus prior luck (access to information, networks, and health).

These interactions are amplified by power laws and the Matthew Effect. Slight differences in talent, slight initial luck, and slight differences in timing become enormous disparities through the compounding effects of network effects, capital, and reputation.The “hard work pays off” model, based on the assumption of a normal distribution, is incorrect in many fields (startups, the arts, research, investing, content). Outliers take the lion’s share.Increasing the number of attempts is effective as a way to “buy luck,” but each attempt comes at a cost. Increasing the number of attempts while ignoring the probability of bankruptcy does not raise the expected value; rather, it increases the probability of ruin.

In conclusion, the portion an individual can control is small. What can be controlled is: (1) avoiding ruin, (2) improving the quality and frequency of attempts (increasing the “surface area” for luck), (3) explicitly defining one’s own value function and not measuring oneself by others’ value functions,(4) making the model of reality as accurate as possible. When quantifying the ratio of talent to luck, the greater the competition and the more the field follows a power-law distribution, the higher the proportion of luck (and initial talent). In everyday professional fields, the relative importance of effort and strategy increases, but even so, the difference in initial conditions does not disappear.

What Should You Do?

Don’t make success your goal. Design your value function, increase its expected value, and manage its variance. If you make it your goal, incentives will distort your understanding (you’ll be tempted to believe success stories and ignore inconvenient data).

• Articulate your own values. Is it the approval of others, the quality of specific experiences, understanding, contribution, or peace of mind? If you don’t articulate them, society’s default value function—money and status—will override them.
• Eliminate downsides first. Sleep, health, a minimum financial buffer, and resilient relationships. These form the foundation of your options. Once you’re ruined, your number of attempts becomes zero.
• Prioritize skills with high versatility and leverage: the speed of learning itself; the ability to model reality; and the ability to maintain your own judgment while collaborating with others. Depth in specific domains can come after exploration.
• Consciously separate exploration from exploitation. While you’re young—or while you have ample resources—focus on exploration (trying out multiple games on a small scale). Once you’ve reached a certain point, shift to exploitation. Committing too early to a single definition of success maximizes timing risk.
• Don’t overestimate the stories of successful people. Those represent conditional probabilities (among those who succeeded), not prior probabilities (the percentage of people who would succeed if they did the same thing).
• To increase your luck, create, publish, meet people, and lower the cost of failure. As long as you’re keeping things hidden, you haven’t bought a ticket. However, be sure to calculate the cost of purchasing that ticket (time, reputation, money).
• Life is a sample size of one. For an individual, the types of regrets and the quality of daily experiences tend to serve as more realistic evaluation functions than statistical success. There is no intrinsic meaning in having “whether you succeeded” judged by others. Even the criteria for judgment change over time.

From my perspective, the most cost-effective way of life for a human being is to understand how the world works as accurately as possible and to ensure that one’s actions do not contradict that understanding. Curiosity and updating one’s models simultaneously improve the surface area for luck, sensitivity to timing, and the ability to avoid ruin. Wealth and fame can be byproducts of this; they need not be the goal.Even without them, as long as understanding and consistency are maintained, that constitutes success by my definition.

Whether pursuing success is important depends on whether you include it in your value function. Choosing not to include it is rational; if you do include it, the only logical approach is to “optimize only the controllable variables and accept the rest as variance.”