Survivorship: The Mental Model of Always Asking for the Graveyard

What is survivorship as a mental model?

Survivorship is the habit of asking what the evidence leaves out before you learn from it. Its rule: always ask for the graveyard. If you only study the planes that came back, the companies that grew or the people who got rich, you are looking at a filtered sample, and the filter may be the whole story.

This page is about the method. The error it prevents, survivorship bias, has its own page. Here the focus is what to do: how to find the missing cases, and how to change your decision once you do.

The model rests on three questions:

• What was the filter? Something decided which cases you can see: survival, success, publication, popularity.
• What does the graveyard look like? Who tried the same thing and dropped out?
• Do winners and losers differ on the trait I care about? If both groups did it, it doesn’t explain winning.

Where it comes from

The best-known example comes from World War II. Abraham Wald, a mathematician and statistician, worked with the Statistical Research Group at Columbia University, which analyzed military problems for the United States. One question was where to add armor to bombers, since armor is heavy and can’t go everywhere.

The data came from aircraft that returned from missions, with damage spread across the wings and fuselage. The obvious answer was to armor where the holes were. Wald’s insight was the reverse: these were the planes that survived being hit there. Hits in the areas with few holes, such as the engines, were the ones that tended to bring planes down, which is why those planes weren’t in the sample. Armor belonged where the returning planes were clean.

The popular retelling compresses Wald’s work, which was a set of technical memos on estimating vulnerability from survivors’ damage. The logic is faithful, though: the missing data was the most important data.

Long before Wald, Cicero told a story about Diagoras, who was shown painted tablets of sailors who had prayed and survived a shipwreck, as proof that prayer works. He asked where the tablets were of those who prayed and drowned. Nassim Nicholas Taleb retells it in The Black Swan (2007), calling the failures history doesn’t record “silent evidence”.

How to use survivorship

1. Name the sample and the filter. Write down what you are learning from and how it reached you: “founders interviewed on podcasts”, “funds still open today”, “returning aircraft”.
2. Estimate the graveyard. How many started? How many dropped out? Use base rates: what share of new restaurants, startups or funds usually survive?
3. Go find some of the dead. Look at shut-down companies, closed funds, abandoned projects, people who tried and quit. Even a handful changes the picture.
4. Compare the trait across both groups. If winners and losers both worked 80-hour weeks or both dropped out of college, that trait doesn’t explain success.
5. Use inversion. Ask what killed the failures, then check whether your plan avoids those causes. The graveyard teaches more about risk than the winners teach about method.
6. Adjust your odds, not just your story. Use probabilistic thinking: if the base rate of success is low, your plan needs a margin for being in the majority.

A quick version for everyday use: whenever you hear “everyone who did X succeeded”, reply in your head with “how many did X?” If nobody knows, the claim is a story, not evidence. The same goes for reviews, testimonials and your own memories of what worked. The cases that went badly tend to leave the record quietly, through closed accounts, deleted posts and things people prefer not to mention.

3 real-world examples

Business

Say you study ten famous founders who dropped out of college and conclude that dropping out helps. Asking for the graveyard means asking how many dropouts started companies that failed, and how many graduates built large companies. Once you include them, the pattern usually weakens or disappears. What the winners share with the losers can’t explain the difference. Look instead for what the winners did that the failures didn’t.

Personal life

Imagine a friend says old furniture was built better, pointing to a 100-year-old oak table. The poorly made furniture from that era mostly broke and was thrown away long ago. What survives is the sturdy share. The model doesn’t say old furniture was worse; it says the surviving pieces are not a fair sample of what people used to buy.

Investing

Fund performance tables often show only funds that still exist. Funds that did badly tend to be closed or merged, and they drop out of the averages, which makes the category look better than investors’ real experience was. When you compare a strategy’s track record, ask whether closed funds are included. If they aren’t, the reported average is likely flattering.

Common mistakes

• Copying winners’ habits. Habits that the failures shared explain nothing.
• Assuming the graveyard is small. In most competitive fields, failures far outnumber survivors.
• Stopping at “it’s survivorship bias”. Naming the bias isn’t the point. Find the missing data and change the decision.
• Overcorrecting into cynicism. Survivors can still teach something, especially about what they avoided. The model asks for comparison, not dismissal.
• Forgetting your own filters. Your feed, your network and your memory are filtered samples too.
• Trusting the sample you were handed. Lists, rankings and case studies were selected by someone. Ask how before you use them.

Survivorship vs survivorship bias

The Mentarium take

Most advice you meet is survivor testimony. Books, podcasts and posts come from people who made it, and they sincerely credit what they did. Their account may be true and still not useful to you, because the people who did the same things and failed aren’t on the stage. “Always ask for the graveyard” is the cheapest correction we know.

Survivorship bias is on the Bias Radar for business decisions, and it is easy to see why: strategy discussions run on case studies of winners. In the Stress-test step of The Mentarium Decision Loop (Frame → Map → Stress-test → Decide → Review), add one line to your notes: “Who tried this and failed, and what did they have in common?” If you can’t answer, you have found your next piece of research.

Note the contrast with the Lindy effect. Lindy uses survival as evidence on purpose: for ideas and technologies, lasting is the signal. Survivorship warns against using survival as a recipe. Both are true at once. Bet on what has lasted, and don’t assume you know why it lasted until you have seen what didn’t.