Silent Evidence: The Dead Don't Write Memoirs

Here is one of the most consequential ideas in Taleb's book, and one of the most useful for understanding why you are wrong about more things than you know:

The most important data on almost any question is unavailable because the losers, the failures, and the extinct are not there to report their experience.

The dead don't write memoirs. Taleb credits this insight to Cicero, who tells the story of Diagoras the atheist.

Diagoras was shown into a temple and presented with paintings. The paintings depicted sailors who had prayed and then survived shipwrecks. "See?" said the believers. "Prayer works. The sailors were saved."

Diagoras looked at the evidence and asked one question: "But where are the portraits of those who prayed and drowned?"

The answer was silence. The drowned did not paint portraits. Their story was not recorded. The temple contained evidence only of survivors, which is not evidence at all — it is selection bias.

This principle, which Taleb calls silent evidence, distorts nearly every domain of human belief.

The Survivors in Every Domain

Writers and Entrepreneurs

The shelves of bookstores groan with biographies of successful entrepreneurs and authors. J.K. Rowling's story of 12 publisher rejections before success. Steve Jobs at Apple. Elon Musk at Tesla and SpaceX.

The library of unsuccessful novelists and failed founders doesn't exist. There are no memoirs from the thousands of writers who were rejected and gave up. No podcast from the entrepreneur whose company burned through capital and disappeared.

Reading only the winners, we wildly overestimate the success rate of the strategies they used.

When Rowling describes her journey, she provides a narrative of how her determination and belief in herself led to eventual success. That narrative is true. But what we don't see is the thousands of equally determined writers who believed just as strongly and whose books were never published.

We cannot infer anything about the strategy from observing only the winners.

Cosmetics and Medicine

Every advertisement for a cosmetic product features someone for whom it worked. The person's skin improved, their hair grew, their confidence increased. Thousands of testimonials, all confirming the product's efficacy.

The millions for whom the product did not work are silent. They did not buy again. They did not post testimonials. They are not in the visible universe of evidence.

The same applies to medical "success stories" and health interventions. We see people who recovered. We see people for whom the treatment worked. The patients who deteriorated or got worse are less visible, especially if they did not survive to testify.

Evolutionary Reasoning

Here is a particularly dangerous form of silent evidence: evolutionary argument.

It is common to hear that "evolution optimizes," that organisms develop the best solution for their environment, that our bodies are well-designed for the world we inhabit.

This argument is only possible because the failed solutions don't exist. Evolution does not optimize anything. It generates enormous numbers of mutations, almost all of them harmful or neutral. The useful ones persist because their carriers reproduce. Most mutations are silent — they failed, and their lineages ended.

Calling the useful mutations "selected" obscures the mountain of silent evidence: the billions of failed mutations we never observe because their lineages ended.

Evolution doesn't optimize. It tries everything at scale and keeps what doesn't die immediately.

The impression of directionality and optimization is a product of only seeing what survived.

Venture Capital

Every venture capital firm publishes its successful exits. Sequoia's role in Apple and Google. Kleiner Perkins's role in Amazon and Netscape. The narrative is one of brilliant talent-spotting.

What they don't publish: the failed investments, which vastly outnumber the successes. A typical top-tier VC firm may invest in a hundred companies; a handful will return almost all of the fund, most will return zero, and some will lose money.

The public narrative is "VCs are brilliant talent-spotters." The private reality is "VCs are running a portfolio strategy that depends on extreme outliers covering massive failure rates."

The business works only because the silent failures fund the visible wins.

The Historical Record

History is written by the survivors.

We form our picture of Roman life from surviving manuscripts and buildings. The 99% of written material that was lost is completely absent from our historical record. The vast majority of their daily lives — the mundane, the ordinary, the forgotten — left no trace.

When we study a civilization, we are studying the artifacts that were preserved, not the civilization itself. We are systematically biased toward studying the durable and deliberately preserved, which is not representative of what actually was.

Building Codes and Unobserved Failures

Here is a chilling example: A city in an earthquake-prone region has buildings that have stood for fifty years without major incident. An architect concludes the building code is sound.

But what was the largest earthquake in that fifty years? Probably smaller than a 200-year event — the kind of earthquake that only comes once every couple of centuries.

The buildings are survivors of a specific range of stresses. They are not evidence that they would survive the full distribution of stresses the region can produce.

When the big quake arrives, the silent evidence — the unbuilt version of the code that would have prepared for it — becomes audible in the form of collapse.

Most safety confidence is built on survivorship at small stresses.

The Practical Distortion: Success Literature

This is where silent evidence bites hardest in everyday life:

You read a book about morning routines of successful CEOs. Five a.m. wake-ups, cold showers, meditation, journaling. The implicit claim: these routines cause success.

But you never see the book about morning routines of unsuccessful people, because unsuccessful people don't write books. Among the millions of people who wake at five a.m. and take cold showers, the CEO rate is approximately the same as the general population.

The habits are not the mechanism. They are visible characteristics of a small subset of a distribution that was heavily filtered before you observed it.

Yet success literature sells millions of copies based on this bias.

The same applies to biographies of successful investors, successful entrepreneurs, successful academics. Each one shows a path through a branching tree. The other branches — the paths that led to failure — are not in the book.

What would actually be useful is base rates: of all the people who pursued this path, what fraction succeeded? What is the distribution of outcomes? What were the failure modes?

But that book would not sell, because it would not tell you what you want to hear.

The Business Failure You Never Heard About

Here is a question I consider: How many businesses are started every year? Millions, globally.

How many of those fail? Most. The ones that fail quietly disappear. The founders move on. No one writes about them (unless it's a memoir about "how I bounced back," which is still a success story, just a delayed one).

When we think about business, we think about the visible companies. We see the ones that survived and sometimes became large.

We do not see the graveyard of the ones that failed.

This creates an entirely distorted view of entrepreneurship. The failure rate looks lower than it actually is. The strategy that the successful ones used looks more reliable than it actually is.

Because we are observing only the survivors.

Your Own Network

Here is a version of silent evidence that affects you personally:

All the successful people you know work hard. All the CEOs you know have strong work ethics. All the wealthy people you know are disciplined.

So you conclude: "Hard work and discipline cause success."

But you never see the hard-working, disciplined people who didn't make it. They are not in your network. They failed and faded away. They are silent.

Among all the hard-working, disciplined people in the world, the success rate is far lower than you'd guess from observing only the successful people you know.

This is the fundamental error of network-based inference.

How to Correct For Silent Evidence

Since silent evidence is built into how information reaches us, how do you protect yourself?

Taleb's method is straightforward: Always ask: "What would I see if the survivors are all I'm allowed to observe?" Then mentally expand the population to include the invisible ones.

For any success story: How many people attempted this? What fraction succeeded? What happened to the rest?

For any safety record: What conditions has this system survived? What conditions might it not survive? What is the evidence that it would survive those?

For any habit or strategy of successful people: What is the base rate of success for people who do this? How much of the success is explained by selection into the sample vs. explained by the habit?

These questions are not rhetorical. You should try to find actual numbers.

If you can't find the numbers, treat the claim with deep skepticism. The invisibility of the silent evidence is exactly what makes the visible evidence misleading.

The Anthropic Principle

There is a deep version of this problem that Taleb mentions: the anthropic principle.

We exist only in a universe capable of supporting us. The universes that couldn't support life — that are silent, that we never observe — are completely absent from our experience.

This makes it very hard to reason about the precise parameters of the universe, because we cannot compare it to other universes that were close but slightly different. We can only observe this one.

The universe looks "fine-tuned" for life because we are observing it from inside. Of course we are observing a universe capable of supporting us — we could not exist to observe a universe that wasn't.

This is not evidence of design. It is the problem of silent evidence applied to cosmology.

Practical Rule for Your Life

Before accepting a strategy or a piece of advice based on success stories, always ask:

What is the base rate?

Not "did it work for this person?" but "of all the people who tried this approach, what fraction succeeded?"

If you cannot find a base rate, do not trust the advice. The absence of a base rate is a sign that the visible evidence is being selected by a process you don't understand.

You are observing survivors of an invisible filter. The filter, not the strategy, might be what explains the success.