Silent Evidence: Definition and Why It Matters

Silent evidence is the systematic absence of data. It's what you don't see because it failed and disappeared. It's the book that never got published, the startup that went bankrupt, the candidate rejected in the first round. It's the deaths we don't count because they're too obvious to record.

I think about silent evidence constantly because it distorts almost everything we try to learn from data. We study the survivors. We ignore the graveyard.

The Ancient Example

Taleb uses a perfect story from ancient Greece: Diagoras was a skeptic who visited a temple. The priests showed him paintings of sailors who survived storms and had made offerings to thank the gods. "See?" said the priests. "The gods protected them."

Diagoras replied: "Where are the paintings of those who made offerings and drowned?"

That's silent evidence. The graveyard doesn't have a gallery. The dead don't sit for portraits. We see the survivors and the evidence of their salvation, but we don't see the evidence of failure—because failure is silent.

How It Corrupts Data

Silent evidence corrupts every field that tries to learn from history.

Business: We study successful entrepreneurs and try to copy their habits. But how many failed entrepreneurs had exactly those habits and went broke? We don't know because they're not famous. We're looking at the survivor's testimony and ignoring the graveyard.

Investing: We look at investors with great track records and try to understand what made them successful. But how many investors had the same strategy and failed spectacularly? The data is silent. The graveyard is invisible.

Medicine: We study treatments that led to recovery and assume the treatment caused it. But we don't see the patients who received the same treatment and got worse, or the patients who recovered without treatment. Silent evidence distorts causation.

Self-help: Books by successful people teach you what to do. But how many people followed that exact advice and failed? The market only shows you the survivors' books. The failures' books were never written.

The Real Problem

Silent evidence makes small sample sizes dangerous. You look at a handful of successful people and extract rules. But you're not seeing the thousands who followed the same rules and failed. Your sample is censored. The data is incomplete.

This is especially dangerous when combined with the narrative fallacy. We take the survivors, construct stories about why they won, and ignore everyone else. We build certainty on invisible graves.

Protecting Against It

I try to always ask: what's not in my dataset? What failed and disappeared? What would I see if I studied the people who tried and lost, the companies that folded, the strategies that seemed good but backfired?

Silent evidence can't be eliminated—failure is often truly silent. But you can stay aware of it. You can discount the confidence you place in survivor data. You can remember that every gallery of success stands on a much larger graveyard.

Go deeper:

For the full breakdown of silent evidence, survivorship bias, and how to think about invisible data, read Silent Evidence and the Graveyard: What You Don't See.