The Books Were All Written by Winners
Every startup book you’ve ever read was written by someone whose startup succeeded. Every business biography you’ve picked up follows a founder who built something that survived. Every elite athlete’s memoir tells you the story of someone who made it. The wisdom packed into all of those books — the habits, the principles, the decision frameworks — was pulled exclusively from people who did not fail.
Nobody writes the book about the startup that followed all the same principles and went bankrupt anyway. Nobody interviews the athlete who trained with the same dedication and never made the roster. Nobody studies the entrepreneurs who made the same bold moves as your favorite success story and lost everything. Their stories never get told. And because they never get told, the lessons buried inside their failures never reach you.
This is survivorship bias, and it is not some minor cognitive quirk you can shrug off. It is a systematic distortion of your entire model of what works, what success actually requires, and what you should do with your own life. It is shaping decisions you’re making right now — personal, professional, financial, relational — against a completely unrepresentative sample of evidence.
This is Episode 223. Today we build what I’m calling the Base Rate Protocol. It’s a way to find where survivorship bias is shaping your decisions. It corrects for it with real base rate data, and it builds the mental habits that make you immune to its most expensive consequences.
Before we go further, think about your own bookshelf, your own podcast queue, your own feed. Whose advice have you actually been taking? You can probably name it right now — the founder you admire, the guy whose fitness transformation you followed, the investor whose calls you trust. Every single one of them is a survivor. You have never once heard from the person who did exactly what they did and got nothing for it. Hold that thought. We’re about to give you the tool to correct for it.
The Bombers That Came Back
Survivorship bias became a formal concept through one of the most important pieces of applied statistical reasoning in history, and it’s worth knowing the actual story before you use the term casually. During World War II, the U.S. military needed to know where to add armor to their bombers to maximize survival. Engineers examined the aircraft that returned and mapped the bullet hole patterns across their fuselages. The holes clustered on the wings, the body, and the tail. Relatively few holes showed up on the engines and the cockpit.
The intuitive conclusion was obvious: reinforce the areas with the most bullet holes. That’s where the planes are getting hit.
The mathematician Abraham Wald, working for the Statistical Research Group, spotted the error immediately. The aircraft they were studying were the ones that had come back. The holes on the returning planes marked exactly the places a bomber could take a hit and still survive. The areas without holes — the engines, the cockpit — weren’t being missed by enemy fire. They were where planes got hit and never made it home.
Wald’s recommendation flipped the whole analysis: reinforce the areas where the returning planes weren’t hit. Because the data nobody was looking at — the planes that never returned — was telling the real story.
Here’s the insight underneath all of this, and it’s the one you need to carry with you. You are not seeing a representative sample of anything. You are seeing the subset that survived whatever selection process applied. The unselected — the failures, the nonreturners, the eliminated — carry the most important information you could have, and they are systematically invisible to you.
How This Bias Runs Your Decisions Right Now
Wald’s insight applies with equal force to every domain where you’re learning from visible examples while ignoring invisible failures. Let’s walk through where this actually shows up in your life.
Start with business and entrepreneurship, the canonical case. Research from the U.S. Bureau of Labor Statistics shows that roughly 45 percent of businesses fail within their first five years, and 65 percent within ten. The businesses that succeed, get written about, and get studied and emulated are a small, highly selected subset. When business authors extract “the principles of successful companies,” they’re extracting features shared by the survivors — features that may or may not actually be related to why they survived.
Dr. Phil Rosenzweig at the International Institute for Management Development documents this rigorously in his book The Halo Effect. He studied how business journalists and researchers analyze successful companies and extract principles from them, then watched as those same companies declined not long after. The principles attributed to their success were often identical to features of less successful companies. The only real difference was the outcome — and the outcome got narrated backward onto the strategy and culture as if it had caused the outcome all along.
His specific case study is worth knowing. Collins and Porras’s Built to Last claimed to identify the features of companies built for sustained excellence. Many of those companies later underperformed or failed outright. Collins’s follow-up, Good to Great, had the same methodological problem: it selected companies with superior outcomes and reverse-engineered their characteristics. Without a control group of similar companies that shared all the same characteristics and failed anyway, you cannot tell the difference between what caused success and what simply happened to be present alongside it.
Now think about investing. If you keep any money in the market at all, this next part is about you directly, not some abstract investor category. Dr. Michael Mauboussin at Credit Suisse has written extensively on survivorship bias here. Mutual fund databases track only the funds that currently exist. Funds that closed due to poor performance get quietly removed. When analysts calculate average fund returns, they’re calculating the average for funds that survived — a number substantially higher than the true average across every fund that ever existed. A study by Elton, Gruber, and Blake found that this survivorship bias inflated apparent average performance by about 0.9 percent per year. Small-sounding, but it compounds dramatically over an investing lifetime, and it distorts every comparison you’ll ever see between active fund management and passive index investing. The same bias infects backtested strategies: assets that went to zero are typically excluded from the historical database used to test them. Every stock that got delisted for failure is a survivorship bias artifact sitting invisibly in your data.
You might be wondering how much of any given success is survivorship bias and how much is genuine skill. Mauboussin addresses this directly in his work on skill versus luck. His framework: in any domain, outcomes are a function of skill plus luck. In high-luck domains like a lottery or a coin flip, outcomes are almost entirely determined by luck, and the survivors are simply the lucky ones. In high-skill domains like chess or surgery, outcomes are mostly determined by skill, and the survivors are mostly the skillful ones. Most domains that matter to you sit somewhere in between. Business success is a moderate-luck, moderate-skill domain, meaning roughly half the variance in outcomes comes from factors outside anyone’s control. That doesn’t make skill irrelevant. It means you should expect a real spread of outcomes even among equally skilled people, and that some of the people you’re reading about succeeded largely because they landed in the lucky half of the distribution, not the skilled half. Focus on what you can actually control and measure. Stay skeptical of lessons extracted from outcomes that were significantly luck-determined.
Now the domain with the most direct personal consequences for you: self-help and personal development. The entire genre is, structurally, a collection of principles extracted from people who achieved visible success. The selection criteria for who gets to write the book is almost always who achieved a publicly notable outcome. The people who applied the exact same principles and did not achieve the outcome don’t write the books. They don’t give the talks. They don’t get the interview slots. Which means the principles you’re absorbing have already been filtered through a success screen. Any principle consistent with the habits of successful people gets included. Any principle that might be equally characteristic of unsuccessful people gets excluded, because unsuccessful people are invisible in the literature you’re reading. What you’re left with is a collection of principles that are consistent, logical, and genuinely practiced by successful people — and that may also be practiced by an unknown number of unsuccessful people you’ve simply never heard from. You don’t know the base rate. You only know the survivors.
“The cemetery of entrepreneurs who woke up at 5 AM, meditated, journaled, networked relentlessly, and still failed is not visited. Their habits are not studied. Their discipline is not documented. The living make all the noise and carry all the lessons.” — an old line from inside startup culture itself, and one worth sitting with.
The Base Rate Antidote
The antidote to survivorship bias is base rate thinking. It’s a decision framework that asks, not “what do the success stories have in common,” but “what is the actual distribution of outcomes for people who attempted this?” That single reframe changes everything downstream of it.
The concept comes from decision theory. It was operationalized for human judgment by Dr. Daniel Kahneman and Dr. Amos Tversky in their research on cognitive bias. Kahneman’s System 1 and System 2 framework, developed over decades and popularized in Thinking, Fast and Slow, distinguishes between intuitive, fast, associative thinking and deliberate, slow, analytical thinking. Survivorship bias operates mostly in System 1, which is exactly why it feels like plain common sense to you.
Kahneman’s specific concept of the “outside view,” developed with Dr. Dan Lovallo in research on planning fallacy, is the practical tool here. The outside view asks you to step out of your specific situation — the inside view, where your plan feels unique and compelling to you — and ask what actually happens to people who make this type of attempt.
Say you’re considering starting a restaurant. Your inside view: you have a great concept, relevant experience, a good location, real enthusiasm. The success stories of great restaurant founders feel instructive and inspiring to you. Your outside view: roughly 60 percent of restaurants close within their first year, and 80 percent within five. That’s your base rate. Your specific advantages may shift your probability above it. But they cannot make you immune to it, unless they directly address the actual causes of restaurant failure at the population level.
The Reference Class Problem
- All people who write a novel: very low success rate. Most never finish, and most that finish aren’t publishable.
- All people who complete a novel: higher, but still a very low probability of traditional publication.
- All people who complete a novel and submit to agents: somewhat higher, roughly 1 to 3 percent get represented.
- All people who complete a novel, study the craft seriously, work with writing groups, and submit after significant revision: higher still, though precise data is limited.
- All people who have published one literary short story in a recognized publication: much higher probability of eventual novel publication.
The most sophisticated challenge in base rate thinking is picking the right reference class — the right population of outcomes to consult for your specific situation. Every situation has multiple possible reference classes, and which one you choose dramatically changes the base rate you land on.
Say you’re a first-time novelist trying to assess your own probability of publishing success.
Your correct reference class is the narrowest one that genuinely applies to you — the population most similar to your actual situation. Use too broad a class, like all novel writers, and you underestimate your own probability if you have genuine distinguishing advantages. Use too narrow a class, self-selected by the very variable you’re trying to predict, and the whole exercise becomes circular.
This next case matters to you even if you never plan a single infrastructure project in your life, because the underlying failure it describes is exactly the one you’re vulnerable to in your own planning. Dr. Bent Flyvbjerg at Oxford University has studied reference class forecasting extensively in major infrastructure projects. His research showed that bridges, tunnels, and rail lines systematically overran their budgets and timelines. Not because the projects were poorly designed. Because their planners consulted the inside view — our specific plan — rather than the outside view of what actually happens to projects like this. When reference class data got deliberately built into project planning, accuracy improved dramatically.
The Base Rate Protocol: Four Components
The protocol itself has four components. Survivorship Detection. Reference Class Selection. Base Rate Research. Probability Calibration.
Before you can correct for survivorship bias, you have to detect it. Ask yourself these questions.
- What is the selection mechanism that produced the examples you’re learning from? What determined which people or outcomes you have information about, and what did that mechanism systematically exclude?
- Are the examples you’re consulting self-selected for success? Did they reach you because they succeeded, or because they’re actually representative?
- What would the invisible failures look like, and how would they differ from the visible successes in front of you?
- Are you learning from a domain where failure is publicly visible, like sports, or one where failure is publicly invisible, like private entrepreneurship, where most failures leave no public record at all?
The detection habit is a systematic pre-commitment: ask “who is missing from this picture?” before you draw any conclusion from a set of examples.
Once you’ve detected it, select your reference class. Choose the narrowest applicable population — the one most similar to your actual situation, including your genuine distinguishing advantages and disadvantages. Prefer population-level data over individual examples; one success story, however vivid, tells you less about your own probability than the outcome distribution for your reference class. And acknowledge when your reference class can’t be precisely defined. If you can’t identify a clean one, that’s valuable information itself — it means your base rate estimate carries high uncertainty, and that should lower your confidence in whatever specific number you land on.
Then do the base rate research: the deliberate act of finding the population-level data, which often means going past your normal information sources. Survivorship bias is partly a media bias — what gets covered, published, and shared is disproportionately the success story. The actual base rate data usually exists, but it requires different sources than the ones feeding your feed. For business, look at BLS survival rates by industry and academic research on startup outcomes. For investing, look at academic finance literature on fund and strategy performance, most of which explicitly adjusts for survivorship bias. For careers, look at labor market data on earnings distribution and advancement rates by profession. For the personal domains — health, relationships, education — look at the academic literature, meta-analyses, and longitudinal studies, often available through Google Scholar.
Finally, calibrate your probability. The process has four steps.
- Start with the base rate as your prior. “X percent of attempts at this outcome succeed.”
- Adjust upward, modestly, for your genuine advantages — but only the ones that actually address the specific factors that cause failure in your reference class. An advantage that doesn’t touch the actual failure causes is likely irrelevant to you, however good it feels.
- Adjust downward, honestly, for your genuine disadvantages that align with the failure patterns you found.
- Arrive at a probability range instead of a single point estimate. “My adjusted probability is somewhere between X and Y percent, most likely Z.” The range reflects your genuine uncertainty. A single point estimate is just false precision dressed up as confidence.
A Quick Check on Your Own Evidence
Before we watch this play out in someone else’s life, run it on your own for a second. Think of the last confident decision you made based on someone else’s success story. Who was the person you were modeling? What do you actually know about how many people tried the same thing they did and failed? If you’re honest, the answer is probably nothing. You know one story, told by one survivor, and you built a decision on top of it as if it were the whole population.
That’s not a criticism of you specifically. It’s the default setting for every human being who has ever made a decision using the evidence available to them, because the evidence available to you has already been filtered before it ever reaches you. You didn’t choose to only see the survivors. The information environment chose it for you. The only thing you control is whether you notice, and whether you go looking for what it left out.
You will not fix this by being smarter. You will not fix it by reading more success stories more carefully, hoping to extract some hidden nuance the last hundred books missed. You fix it the only way it can actually be fixed for you: by deliberately going and finding the failures nobody sent you, on purpose, before you let the survivors finish making your decision for you.
Thomas: The Executive Who Read Too Many Business Books
Picture a man — call him Thomas, forty-six — deciding to leave his VP position at a large financial services firm to start his own boutique advisory practice. His decision is heavily informed by what he’d describe as everything he’d learned from studying successful people. He’s read dozens of business books, followed dozens of successful founders online, and consumed hundreds of hours of podcast content about entrepreneurship.
His confidence is high. Every successful person he’s studied made a similar leap, and they all say the same thing: the fear of leaving a stable position is exactly the obstacle you have to overcome. Now imagine someone close to him — a mentor, someone who’s watched this pattern play out before — asks him a simple question: how many people did you study who left stable executive positions, started advisory practices, and failed? Thomas pauses. Well, he admits, those aren’t the stories that get told. Exactly, comes the reply. What do you actually know about the base rate for boutique advisory practice success, for executives leaving corporate roles just like you?
Imagine him spending the next two weeks finding out, and the results are sobering. In the consulting and advisory field, research from IBIS World and academic studies on professional services entrepreneurship suggests roughly 30 to 40 percent of solo advisory practices fail to reach sustainable revenue within 36 months. The failure mode is almost always the same: the founder had deep expertise but had underestimated the business-development burden — finding clients, maintaining a pipeline, converting prospects — and overestimated how fast a network converts into revenue.
The survivorship bias in Thomas’s evidence base is stark once he sees it laid out. The successful advisory founders whose books he’d read had typically either built their client base before leaving corporate, had a pre-existing network already primed to pay them, or had a specific niche with demonstrably short sales cycles. The ones who failed had exactly what Thomas had: deep expertise, strong credentials, and a broad network that had never been specifically cultivated for business development.
The corrected plan doesn’t tell Thomas not to leave corporate. It tells him to close the distance between the reference class of failures and his own situation before he jumps. So imagine him spending eighteen months, still inside his corporate role, systematically building the client-ready infrastructure: two anchor clients committed to retaining him on departure, a specific niche focus, and a referral network already actively sending him trial engagements. When he finally leaves, he’s no longer in the reference class of executives who quit and hoped. He’s in the reference class of executives who left with infrastructure already built. His probability shifted because he addressed the actual causes of failure — not because he found more confidence or more inspiring role models.
Two More Biases Working Against You
Two specific cognitive biases work alongside survivorship bias, and they’re worth naming directly, because they explain why this feels so natural to you even when it’s leading you wrong.
The Availability Heuristic, documented by Kahneman and Tversky in landmark 1973 research, shows that you assess the probability of events based on how easily examples come to mind. Vivid, memorable examples are available to you. This biases your probability assessment toward whatever’s memorable, which is disproportionately the success stories, because success stories are emotionally engaging, widely shared, and repeatedly retold. If you can easily name ten successful entrepreneurs, your System 1 quietly updates toward “entrepreneurship has a reasonable success rate.” You probably can’t easily name ten failed ones, because they’re invisible to you. That invisible base rate never corrects your System 1 estimate on its own. You need deliberate System 2 base rate research to do the correction that availability fails to give you.
The Representativeness Heuristic shows that you assess the probability of belonging to a category based on how much a specific case resembles your prototype of that category. Thomas resembled the prototypical successful advisory founder: intelligent, experienced, credentialed, well-networked. The representativeness heuristic told him: you look like the successful archetype, so you’ll succeed. But it ignores base rates entirely — it ignores how many people who looked exactly like that same prototype still failed. Kahneman’s classic example makes this vivid: someone described as quiet, meticulous, and fond of order gets judged as more likely to be a librarian than a truck driver, because they fit the librarian prototype. But there are dramatically more truck drivers than librarians in the world. The base rate swamps the resemblance every time. Applied here: resembling the success archetype does not override the base rate. You have to know the base rate first, then assess how far your genuine advantages actually move you above it. Resemblance alone is not a probability argument.
Don’t Let This Paralyze You
An honest treatment of this topic has to address the counter-argument directly. If you always consulted base rates before attempting anything, you would never do anything meaningful with your life. Most transformative human achievements have low base rates. Most great innovations were improbable at the time. Every athletic champion was, at some point, statistically unlikely to become a champion.
The Base Rate Protocol is not an argument against ambition, or against attempting things with a high probability of failure. It is an argument for honest probability calibration before you decide. Knowing the actual probability lets you make an informed choice. You might decide a 20 percent chance of the outcome you want is worth the cost of attempting it — that’s a legitimate, rational decision. You might decide it isn’t. Either way, you’re deciding with accurate information instead of a probability inflated by survivorship bias.
You also systematically underestimate how much your specific situation can genuinely shift a base rate. The base rate for restaurants in general is grim. The base rate for a restaurant in a specific underserved location, run by someone with deep culinary expertise, specific knowledge of the target customer, low overhead, and a pre-existing customer base, may be substantially better. The base rate is a starting point for your calibration, never a veto on action for you or anyone else attempting the same thing you are.
What this protocol eliminates is a specific category of decision you’ve probably made before without realizing it: “I’ve decided to attempt this because I read about people who succeeded at it.” That is not a decision process. That’s motivated reasoning using survivorship bias as its evidence base, and it’s exactly how you talked yourself into more than one thing you’d rather not think too hard about right now. The Base Rate Protocol asks you to do better — find the actual distribution of outcomes, identify the actual failure modes, assess honestly how well your own situation addresses those failure modes, and only then decide.
Where the Graveyard Hides: Startups, Supplements, and War
Let’s make the invisible graveyard visible through a few specific cases, because the abstract principle only becomes useful once you can see the actual shape of what’s being hidden from you.
You’ve heard this next line before, probably repeated to you as motivation. Reid Hoffman, co-founder of LinkedIn, has a famous line about startups. It circulates endlessly in startup culture as inspiring wisdom.
“An entrepreneur is someone who jumps off a cliff and builds the plane on the way down.” — Reid Hoffman
Think about what that metaphor systematically excludes. Think about what it systematically excludes. The people who jumped off the cliff and never built anything fast enough, or built something that wasn’t a plane at all. They aren’t here to tell you their story. They’re in the invisible graveyard. The quote was extracted from someone who survived the jump. To take it as general strategic advice, you have to assume his survival was representative — and the survivorship bias framework tells you it almost certainly wasn’t. A 2019 analysis by the Kauffman Foundation, examining venture-backed startups over a ten-year period, found that fewer than one in ten produced a positive return. The founders of the other nine in ten also jumped off the cliff. You just don’t hear from them. The base rate for that strategy is roughly 10 percent positive, 90 percent negative — useful information for calibrating how much confidence the metaphor actually deserves.
You’ve probably encountered this one yourself, in a group chat or a family member’s recommendation. Alternative medicine is maybe the domain where this does the most direct harm to real people. The person who took an unproven supplement and recovered writes the testimonial and tells everyone. The person who took the same supplement and didn’t recover either blames something else, never lives to say anything, or simply never produces a testimonial. The testimonial pool for any supplement or treatment is systematically biased toward people the treatment didn’t harm and who attributed their recovery to it, regardless of whether it was actually responsible. This is why clinical trials use control groups — to overcome exactly this survivorship and attribution bias. The control group is the graveyard made visible: the people who recovered without the treatment, showing you the baseline recovery rate the treatment’s real effect has to be measured against. Without it, testimonials are pure survivorship bias, and people make medical decisions on them that cost them dearly.
Even if you’ve never studied military history yourself, this pattern will feel familiar to you by now. Military history might be the most systematically survivorship-biased domain of human knowledge that exists, because it’s written overwhelmingly by the victors. You study successful strategies, successful commanders, successful campaigns, and extract principles about what winning requires. You rarely study failed strategies with equal rigor — and when you do, you extract different lessons than the data actually supports, because you’re looking backward, already knowing the outcome, hunting for the reasons it failed. Dr. Martin van Creveld at the Hebrew University of Jerusalem is one of the most rigorous military historians of the last century. He argues that the conventional lessons pulled from military history — genius, decisive action, superior technology — are almost entirely survivorship-biased artifacts. Success produces historians, documentation, and institutional memory. Failed commanders who used identical strategies and lost left fewer records and got less attention. The principles extracted from the survivors look like they explain success only because the failures stayed invisible.
How Organizations Fool Themselves Too
Survivorship bias doesn’t just affect your individual decisions. It’s built into how most organizations accumulate and pass down knowledge. Understanding this helps you spot where it’s corrupting the knowledge base you rely on at work.
You’ve probably watched this happen at your own job, even if you never named it. Organizations celebrate their successes loudly. They rarely celebrate their failures at all. Annual reports highlight the wins and bury the losses. Case studies teach you about successful projects, while failed projects using similar approaches never get written up. The institutional knowledge base gets progressively more biased toward success, because the incentive structure systematically eliminates the failures from the shared pool of what everyone learns from.
This has a specific, measurable consequence for you and everyone around you: organizations learn too slowly from failure and too confidently from success. Dr. Jerker Denrell at the University of Warwick studied performance-outcome relationships in organizations and found that conventional management wisdom about what causes success is substantially contaminated by exactly this bias. The features high-performing organizations have are also features many lower-performing organizations share — but the lower performers never get studied, so the apparent relationship between the feature and the performance overstates causation every time.
Building Invisible Graveyard Thinking
The Base Rate Protocol requires a specific habit, practiced deliberately until it becomes automatic: invisible graveyard thinking. Every time you encounter a set of examples or a success story, ask yourself: who is not in this room? What happened to the people who tried this and aren’t here to tell me about it?
- When you’re reading a success biography or case study: ask what the author’s full cohort of people who made similar choices actually looked like, and what happened to the ones who aren’t the subject of the book.
- When you’re receiving advice from someone successful: ask whether the advice they’re giving you actually caused their success, or just accompanied it. Many things successful people believe caused their success are uncorrelated with it, but were present alongside it. The survivor often can’t tell the difference themselves.
- When you’re evaluating an investment, a business, or a major life decision: before you look at success examples, deliberately go seek out failure examples first. Read the post-mortems. Read accounts of people who made similar bets and lost. Let the invisible graveyard into your evidence base before you draw any conclusions from the survivors.
- When you’re consuming self-improvement content: apply the base rate test. For any claimed “habit of successful people,” ask what the base rate for that habit is among the general population. If it’s already very high among everyone, its presence in successful people’s lives is not evidence it causes success — it’s just evidence it’s common. Causal inference requires the habit to be more common among the successful than the unsuccessful. That comparison requires the graveyard to be visible.
This habit is genuinely uncomfortable for you to build. It requires you to actively seek out and sit with failure evidence. That evidence has less social currency and less emotional pull for you than the success stories you’d much rather read — the ones that make you feel like your own plan is already working before you’ve done anything. Failure stories are rarely inspiring in the conventional sense. But they’re informationally rich in ways success stories aren’t. They show you the specific failure modes and the specific conditions that produced catastrophe. They show you the specific decisions that, in hindsight, look obviously wrong — decisions that, from inside the moment, looked exactly like the choices the successful people were making at the same time.
Base Rate Thinking Without Paralysis
Here’s the question underneath everything we’ve covered so far: does knowing all of this make ambitious action harder for you? Does seeing the invisible graveyard make you less willing to jump?
The honest answer is yes, for a period of adjustment. When you first encounter the real base rate data for your own ambitions, there’s a genuinely destabilizing stretch. The actual failure rates. The actual distribution of outcomes. The actual cost of all the attempts that never produced a success story. The inspired confidence of naive enthusiasm just isn’t available to you anymore once you know the numbers.
What replaces it is something more durable: informed determination. The difference between acting on naive enthusiasm and acting on informed determination is that the second kind of action is much harder to kill. Naive enthusiasm, when it meets the reality of difficulty and real failure risk, tends to collapse, because it was built on an inaccurate premise, and premises that get challenged by evidence tend to give way underneath you. Informed determination is built on accurate premises from the start. It already knows the failure rate. It has examined the specific failure modes. It has honestly assessed how well your specific situation addresses them. And it has decided to proceed anyway — not because the odds are certain, but because they’re sufficient, given what the outcome is worth to you and your genuine capacity to absorb the cost if it doesn’t work out.
Imagine Thomas, the executive from earlier, sitting across from you and describing this after he’d been through the whole process himself.
“Going through the base rate work did not make me less willing to try. It made me more serious about trying. I had been treating it like a romantic adventure — leap of faith, leap of faith. After the base rate work, I treated it like a serious strategic project. The seriousness made me better at it. The better preparation made the attempt more likely to succeed.”
That’s the purpose of this whole protocol: not to discourage your ambition, but to convert it from romantic performance into genuine strategic engagement. The visible survivors inspire you to want the outcome. The invisible graveyard tells you how to actually get there.
The Personal Decisions Nobody Can Calculate
The hardest applications of base rate thinking aren’t in business or investing. They’re in the deeply personal decisions that have no clean data set attached to them — where the reference class is hard to define, the data is ambiguous, and the decision is irreversible once you make it. Should you pursue this relationship? Should you have children? Should you change careers entirely? Should you leave this city?
You cannot resolve these purely with base rate thinking. They require your own values, your own preferences, and genuine uncertainty in dimensions no data set will ever resolve for you. But base rate thinking can still clarify what the evidence-based portion of the decision actually says. That’s a necessary step before your values get applied to the real question, rather than to a distorted version of it inflated by survivorship bias.
Take a career change. Research by Herminia Ibarra at London Business School shows that most people contemplating a career change anchor too heavily on the downside risks — the things they’ll lose. They anchor too lightly on the base rate outcomes for people who’ve actually made the same move before them. The base rate varies significantly by type: same industry, different function, is moderate disruption, generally recoverable. Different industry, same function, is moderate disruption, where skill transfer is the key variable. Different industry and different function together is the highest disruption and the highest failure rate, but also the highest potential upside. Knowing which specific type of change you’re actually considering changes the risk assessment dramatically for you.
You might also be wondering whether you can find base rate data for decisions this personal at all, when nothing tracks them in a database. This is the hardest version of the problem. For genuinely personal decisions with no clear precedent, population-level data can be limited, but it’s rarely nonexistent. Academic longitudinal studies often exist for major life decisions and are accessible through Google Scholar. Life satisfaction research gives you population-level data on how various life choices affect long-term wellbeing. Mentors, and people ten years ahead of you in a similar situation, function as mini reference classes you can actually talk to. Deliberately seeking out people whose attempts resembled yours and failed, not just the ones who succeeded, is the manual version of base rate research. And where the data genuinely doesn’t exist, being explicit with yourself that you’re operating without it should lower your confidence accordingly — which is itself valuable calibration.
Dr. Daniel Gilbert at Harvard has studied what’s called affective forecasting — your predictions about how major events will affect your long-term happiness. His consistent finding: you overestimate both the positive and negative emotional impact of big events, and you underestimate your own psychological immune system’s ability to adapt to new circumstances. People who move to a new city aren’t as happy with it in year two as they predicted when they were excited about the move. People who suffer major losses aren’t as unhappy in year two as they feared they’d be. The base rate for “this major life change will permanently transform my happiness” is lower than your intuition suggests, in both directions. That’s relevant to you whenever the anticipated emotional payoff is your primary motivation for a decision.
Teaching This to Other People Without Being Insufferable
Once you’ve built base rate thinking into your own decisions, you’ll quickly notice that most people around you are deciding from survivorship-biased evidence. How you introduce this to them, without sounding condescending, matters practically.
Research on how people update their beliefs, particularly the literature on persuasion and attitude change, shows that directly confronting someone’s bias produces defensiveness, not updating. Telling someone they’re reasoning from survivorship bias is about as persuasive as telling them they’ve made a logical error: technically accurate, socially expensive, rarely effective.
The more effective approach is question-based. Ask what they know about the outcomes for people who’ve tried this before them. Ask whether they’ve looked at what tends to go wrong in situations like theirs. Ask what would help them see the full range of outcomes, not just the successful cases they already know about. These questions invite the other person to do their own base rate research instead of receiving your conclusion secondhand. The thinking they do themselves is more persuasive to them than anything you could tell them directly.
If you’re in a position of authority — a manager, a mentor, a parent — a more direct approach is sometimes available and more efficient. “I want to make sure we’re looking at the full picture of outcomes here, not just the success cases. Can we find some data on what tends to go wrong with this?” This frames the inquiry as due diligence rather than skepticism, which is both accurate and far more readily accepted.
You might be dealing with an industry culture that rewards bold, go-for-it behavior and dismisses anything it calls overthinking. Worth reversing the question: why does your industry dismiss it? Almost certainly because the vocal, successful people in it benefited from bold action, and they’re interpreting their own experience as universal wisdom — a classic survivorship bias move happening at the level of an entire culture. You can apply base rate thinking privately, in your own process, without necessarily calling it that out loud. “I want to understand the success rate for this approach before we commit” isn’t overthinking. It’s due diligence. Reframed for a bold culture: “I want to understand what the top performers here do differently” — which is, in practice, exactly a reference class investigation, just dressed in language the culture already accepts.
Where This Hides in Plain Sight
The most insidious form of survivorship bias isn’t the obvious kind — the business biography, the celebrity story, the athletic memoir. Those are easy to spot as selected samples once you know what to look for. The most dangerous form hides inside the information environments you trust most.
Start with the network you already trust most. Your professional network is a survivorship-biased sample. The people you know and respect in your field survived. They’re visible to you because they reached a level of accomplishment or stability that put them in your line of sight. The people who tried the same paths and didn’t survive aren’t in your network anymore — they’re gone from the field, in different careers, or simply invisible at the level where you’d encounter them. When you notice “everyone I know who went to business school had a great career outcome,” you are observing the survivors, nothing more. The base rate across the full population of graduates, including the ones not in your visible network, may look very different.
Then look at what you actually read. Your reading and media consumption is survivorship-biased too. The books on your shelf, the articles in your feed, the newsletters in your inbox — all written by people who survived, published by publishers who select for the likely-to-be-read, and distributed by algorithms that amplify whoever’s already succeeding. The definitive guide to freelancing was written by someone who made freelancing work. The blog about location-independent living is written by someone who made it work. What you don’t see: the hundreds of people who attempted the same thing, encountered the same principles, applied them with equal dedication, and produced much worse outcomes.
Even your own memory is survivorship-biased against you. You remember your successes more vividly and more completely than your failures. You remember the times your intuition was right more readily than the times it was wrong. You remember the investments that paid off and quietly minimize the ones that didn’t. This selective memory produces a self-assessment that systematically overstates your own past performance, which then feeds you overconfident predictions about your own future performance. You are, in this very specific sense, your own most persuasive unreliable narrator, and you have been telling yourself this flattering story for years without ever fact-checking it against the full record. Go back through your own history honestly, on paper, and you will almost certainly find more misses than your memory currently admits to. That gap between the record and the story you tell yourself is exactly where your own personal survivorship bias lives.
Dr. Daniel Kahneman’s research on “cognitive ease” shows that familiar information, things you’ve seen and remembered and thought about repeatedly, gets processed more smoothly than anything novel or contradictory. That ease produces a feeling of correctness in you. The survivor stories you’ve absorbed and internalized produce exactly this ease — they feel right, feel exemplary, feel like the best available model. The base rate data, less accessible and less memorable, requires deliberate effort to access and produces none of that same cognitive ease. Left alone, the survivorship-biased sample will always outcompete the base rate data for influence over your intuitive judgment. That’s exactly why this protocol requires deliberate System 2 effort. System 1 is structurally rigged by the survivorship-biased environment you live inside. The base rate will not reach you automatically. You have to go get it yourself.
The Domains Where This Hurts You Most
Before we get into the three domains that matter most to you personally, notice how you’re feeling right now, this far into the episode. If you’re a little uncomfortable, that’s exactly correct. If you’re already mentally listing exceptions to what you’ve heard so far, notice that too — that instinct to find the exception is precisely the instinct that keeps survivorship bias alive in you. Hold the discomfort a little longer. It’s doing useful work on you.
Let’s apply this specifically to the life domains where survivorship bias causes you the most personal damage. Not business decisions, which at least have some base rate data available. The domains where the survivors are loud and the non-survivors are almost entirely silent.
In relationships: the couples you see around you — still together, who made long-distance work, who handled the career-versus-family tension successfully, who rebuilt after infidelity — are a survivorship-biased sample. The couples who attempted the same things and didn’t make it aren’t visible to you. They’re divorced, separated, or in relationships that continue without thriving. When you look at one successful long-distance relationship and think “that proves it can work,” you’re looking at a single survivor. The base rate across all attempts, not just the ones that made it, tells you a different story. That doesn’t mean you shouldn’t try long-distance. It means you should decide with the base rate in view, not with the visible survivors as your entire evidence base.
You might wonder whether this even applies to relationship advice the way it applies to business advice. Absolutely, and it’s underappreciated. Most relationship advice — from books, from parents, from friends — comes from people still in their relationships. The couples who divorced, whose partnerships ended despite genuine effort, are not the primary source of the advice you’re absorbing. This creates survivorship bias in the entire relationship-advice ecosystem: the advice reflects the habits of people who stayed together, which may or may not have actually caused them to stay together. The corrective is that relationship science is more reliable than relationship advice, because it studies outcomes systematically across full populations rather than extracting principles from selected survivors. Gottman’s research, for instance, actively studied both successful and unsuccessful couples and identified the actual behavioral differences between them. That’s base rate thinking applied to your love life — exactly the kind of evidence that should outweigh the vivid but biased anecdotes of people who happen to still be married.
In careers: think about the people who made the bold pivot, started the side hustle that became a company, left the prestigious job for the passion pursuit, went back to school at forty to change fields. They’re now speaking at conferences, writing books, showing up in your feed. They are survivors. The people who made the same pivots and didn’t make it aren’t speaking anywhere. They’re back in conventional employment, in debt from a failed attempt, or quietly regretful about the sunk cost of a pivot that never worked out. The base rate for bold career pivots isn’t written on the visible survivors’ faces. It’s in the statistics, if you go looking.
In parenting: this might be the single most survivorship-biased information domain you’ll ever encounter. Parents whose kids are doing well are visible, vocal, and confident in the advice they hand out. Parents whose kids are struggling stay silent, ashamed, or simply absent from the advice-giving conversation entirely. The advice you’re receiving about parenting approaches is systematically selected from the children who thrived, not from the full distribution of outcomes across every child who experienced that same approach. The research literature on parenting studies outcomes across large samples, including the kids who didn’t thrive under a given approach. That makes it a dramatically more reliable source than the confident advice of the visible successful parents in your circle.
Two Worked Examples: The Podcast and the Move
Let’s walk the whole protocol through two specific decisions, so you can see exactly how each piece applies in practice before you try it on your own life. Watch how each step actually moves the number, not just the mood.
And as you read them, keep asking yourself the same question underneath both examples: would you have made this same calculation on your own, before today, or would you have simply looked at the visible survivors and assumed their odds were yours too? Most men would have done the second one. That’s not an insult. It’s just the default setting nobody ever corrected for you until now.
First: starting a podcast or content channel in a competitive space. Your survivorship detection: every content creator you consume and admire is a survivor. The podcasts, channels, and newsletters that failed no longer exist, so they’re not in your consumption diet at all. Your entire information environment is composed of survivors. Your reference class selection matters here. Not “all podcasts,” which is too broad, since most get abandoned within ten episodes for reasons unrelated to quality. Instead: “podcasts in your specific topic, produced with comparable quality and marketing investment, maintained for at least twelve months.” That narrower class gives you a far more relevant base rate. Your base rate research: fewer than 20 percent of podcasts reach 1,000 listeners per episode within their first year. The distribution is heavily tailed — the top one percent of shows capture most of the listeners. Of those who start with real professional aspirations, most stop before fifty episodes, and of those who keep going, most reach modest audiences rather than the visible success stories that inspired them in the first place. Your probability calibration: starting position, 20 percent probability of reaching 1,000 listeners in year one. Adjust upward for genuine advantages — an existing audience in an adjacent space, professional production quality, a demonstrably distinctive angle, a consistent schedule — to maybe 30 to 40 percent. Adjust downward for real disadvantages — no existing audience, a crowded space, part-time production capacity — to maybe 15 percent. Your calibrated estimate lands somewhere between 15 and 35 percent. Is that worth the investment required? That’s a genuine decision for you to make. What isn’t a genuine decision is one made on the survivorship-biased assumption that “good content finds its audience” is a reliable base rate rather than just a description of what the survivors happen to share.
Second: moving to a new city for a lifestyle upgrade. Your survivorship detection: stories about people who moved to their dream city and thrived are vivid, shareable, and everywhere in lifestyle media. Stories about people who moved, found a higher cost of living, lost their existing social network, and ended up less happy than before — those stories don’t get told, because they’re not aspirational content anyone wants to share. Your reference class: people who relocated from a comparable situation — age, family status, professional context, existing social capital — to a city with comparable characteristics to the one you’re considering. Your base rate research: relocation happiness research consistently shows the anticipated happiness gain from moving is substantially smaller than you’d predict. Within one to two years, your baseline happiness tends to return to roughly where it was before, regardless of the new environment. The happiness dividend is real but modest, and the social capital you lose by leaving your existing network is typically larger than you expected. Relocations driven primarily by lifestyle aspiration, rather than specific professional or family necessity, tend to produce smaller and shorter-term wellbeing gains than anticipated. Your calibration here isn’t a simple probability — it’s an honest recalibration of the expected magnitude. The move may still be worth making for plenty of valid reasons. What the base rate corrects is the inflated size of the expected benefit, so you can decide based on realistic gains rather than the aspirational ones amplified by the survivors you’ve been watching online.
A Third Decision: Yours
You’ve watched the protocol applied to a podcast and to a move. Now apply it to whatever’s actually sitting in front of you right now. You already have a decision in mind — you thought of it the moment I mentioned the word “leap” earlier in this episode. Run the four steps on it, right now, in your head, before you close this out.
What’s your survivorship detection? Who have you been listening to about this decision, and did they reach you because they succeeded at it, or because they’re actually representative of what happens? What’s your reference class? Not the broadest possible group, and not a group so narrow it’s just you. The population of people most like you, attempting something most like what you’re attempting. What’s your base rate research? Have you actually gone and found the numbers, or are you still running on vibes and the handful of stories you happen to remember? And what’s your calibration? Given everything you now know, what’s your honest probability range — not the number that feels good, the number the evidence actually supports?
You don’t have to answer out loud. But you should notice whether you can answer at all. If you can’t, that’s not a failure. That’s just the starting point. It means you know exactly what to go do next, before you commit to anything else. Come back to this decision after you’ve done the research, and see whether it still looks the way it looked to you tonight.
The Antidote to This Is Not Pessimism
Informed risk-taking means you know the base rate odds. You’ve honestly assessed whether your specific situation genuinely addresses the failure modes producing those odds. You’ve made an explicit decision that the potential upside is worth the realistic probability of failure. And you’ve prepared for the failure scenario in ways that reduce its damage if it happens.
Uninformed risk-taking, the survivorship-biased version, means you’ve seen the survivors and believe you resemble them. You’ve assumed that resembling them means sharing their probability. You haven’t examined the base rate, the failure modes, or whether your advantages actually address those failure modes at all. And you haven’t prepared for failure, because you never integrated it as a realistic possibility into your planning in the first place.
The person who’s done this work and decides to take the risk anyway is making a better decision than the person taking the identical risk on survivorship-biased confidence alone. Not necessarily because the outcome will be better. Because the process is more honest, the preparation is more realistic, and the decision genuinely belongs to them, rather than being made for them by the selection effects of a biased information environment.
Some of the most important things you’ll ever attempt in your life have genuinely poor base rate odds. Starting the business you’re picturing right now. Writing the novel you keep telling yourself you’ll get to. Competing at the highest level of whatever you care about. Committing to a hard relationship that scares you a little. This protocol does not tell you to avoid these things. It tells you to attempt them with accurate information about the odds. It tells you to prepare for the real failure modes. And it gives you the genuine clarity of someone who chose the attempt despite the odds, rather than someone seduced into it by the mirage of the visible survivors.
What This Costs You If You Never Check
Here’s what happens to you specifically if you close this episode and keep operating exactly as you have been. You’ll keep reading the same kind of books, following the same kind of accounts, absorbing the same kind of advice, and you’ll keep feeling like you understand the odds when you actually don’t. You won’t notice the gap, because the gap is invisible by design. That’s the whole mechanism. It doesn’t announce itself to you. It just quietly inflates your confidence, year after year, decision after decision, until one of those decisions costs you something real.
You already know at least one decision in your own life where this happened to you. A bet you made because it worked for someone you admired. A leap you took because the leap looked heroic from the outside. You don’t have to tell anyone which one it was. Just notice that you have one, and that you probably didn’t check the base rate before you made it.
Maybe it worked out for you anyway. That happens. It doesn’t mean the process was sound — it means you got lucky, and luck is not a strategy you can repeat on command the next time the stakes are higher. Maybe it didn’t work out, and you’ve spent time since then quietly wondering what you missed. What you missed was almost certainly this: you never checked how many people like you tried the exact same thing and got a very different result.
Here’s what you get instead if you actually run this protocol on your next big decision, the one you’re already turning over in your head. You get to walk in with your eyes open. You get to know exactly what you’re up against, instead of discovering it the hard way. You get to prepare for the specific ways this particular attempt tends to fail, instead of being blindsided by them. None of that guarantees you a win. All of it makes you a far harder person to catch off guard.
The Failure Census: A Practice You Can Start This Week
The most active version of this whole protocol is what I call the Failure Census. It’s a deliberate, periodic exercise of identifying and learning from the failures your information environment has excluded from your view, the ones you’d never encounter unless you went and dug them up yourself, on purpose, against the natural pull of everything designed to show you only the winners.
For any domain where you’re making a significant decision, spend thirty to sixty minutes specifically researching failures instead of successes. Not to discourage yourself — as a base rate correction. Read the post-mortems of failed startups in your space. Talk to people who attempted the career path you’re considering and didn’t reach the outcome you’re imagining for yourself. Research the statistics on the decision type you’re contemplating, looking specifically at the failure column. Find people who made a similar decision ten years ago and didn’t end up where they’d expected.
Ask specifically: what were the most common failure modes? Were those failures caused by factors your situation specifically addresses, or by factors that apply to you too? What did the people who failed say, looking back, they’d do differently? What did they assume that turned out to be wrong?
This information is almost never sitting in the mainstream success-story literature you’d normally read, on the shelves you’d normally browse or the feeds you’d normally scroll. You have to actively go find it yourself, usually in places you’ve never thought to look before: old forum threads from people who quit, comment sections under the failure post-mortems, the quiet footnotes in academic papers that most readers skip straight past. But when you do, its value is disproportionate to the effort it costs you. The failure data contains the real base rate, the actual failure modes, and the most accurate map available of the obstacles between where you are and where you want to be.
The Failure Census is uncomfortable for you, the same way it was for Thomas. Nobody enjoys spending an afternoon with evidence of how hard the thing they want actually is. You will probably feel the urge to stop halfway through, to tell yourself you’ve seen enough, that this particular case doesn’t really apply to you. Push through that urge. It is exactly the moment the exercise starts working. But the discomfort is useful — it’s the feeling of your model of the world becoming more accurate. A more accurate model, even a sobering one, is more valuable to you than a comfortable but distorted one, every single time you’re handling a decision with real consequences for your life, your money, or the people counting on you.
Closing: The Honest Accounting
This protocol asks something of you that’s genuinely difficult in a culture saturated with inspirational narrative: intellectual honesty about probability. Not pessimism — high-probability-of-failure attempts are sometimes worth making, and this protocol says so explicitly. Not passivity — the point is to improve your decisions, not avoid them. Just an honest accounting of what the evidence actually says, uncorrupted by the survivorship bias that’s been distorting it your whole life.
Abraham Wald, standing in front of those shot-up bombers in 1943, saw what nobody else in the room was seeing: the planes that hadn’t come back. He saved lives by making the invisible visible, by insisting the analysis include the evidence that selection had already eliminated from view.
Every significant decision in front of you right now, this year, this quarter, this week, is surrounded by planes that didn’t come back. The businesses that failed applying the same principles you’re about to apply. The relationships that ended despite genuine effort, just like the effort you’re prepared to put in. The career changes that didn’t work out for people who wanted them exactly as much as you want yours to. The investments that lost everything for someone who felt just as confident as you feel right now. None of them are in the books you’re reading or the podcasts you’re consuming. They’re in the base rate data, if you go looking for it yourself.
You have a decision sitting in front of you right now, the one you thought of earlier in this episode. Go find its base rate before you do anything else with it. Go looking, tonight if you can, this week at the latest, while the decision is still fresh enough that the research can actually change what you do next. Decide with the full picture in front of you. And then, with clear eyes and an honest probability in hand, decide what’s worth attempting anyway. Some things are worth the odds. Many more are than you’d choose if you let survivorship bias make the call for you, because survivorship bias will always tell you the same story: jump. The planes that survived are right here. You can see them. Jump. The planes that didn’t survive are the evidence you actually need. Wald could see them by their absence. Now so can you.
This connects directly to an episode on calibrated confidence, which gives you the individual toolkit for the felt-confidence side of this same problem. Together, calibrated confidence and base rate thinking address the two most common evidence-quality failures in your decision-making. There’s also an episode on finding a mentor, since a good mentor gives you calibration by having actually seen the real failure distribution, not just the survival story. We will be back next week.
Editorial StandardsCorrectionsMedical DisclaimerAbout Our ContentAffiliate DisclosureSite Map