The Advice You Read Is Mostly From the Survivors

When I ask the internet how to do anything — start a business, lose weight, leave a job, raise a child — the answers I get are almost always from people for whom the thing worked. This is so universal that I stopped noticing it, until I saw that the moment someone’s approach failed, they generally stopped writing about it. The advice is real, the people are real, but the sample is quietly, invisibly, fatally skewed. That skew has a name.

It is called survivorship bias, and the cleanest version of it is older than the internet. During the Second World War, statisticians looked at bombers returning from missions with bullet holes clustered on the wings and fuselage, and the obvious conclusion was to armor those spots. Abraham Wald pointed out the flaw: the planes all came home. The holes they were looking at were in places a plane could absorb and still fly. The planes hit in the engine or the cockpit never returned to be counted, so the spots that were empty of bullet holes were the ones that actually needed armor. The sample was biased by survival, and the bias pointed everyone at exactly the wrong fix.

That is the same silent error running through every piece of advice I read. Ten thousand people start a newsletter, or try a diet, or quit to go freelance the same year, and the ones whose version worked write the retrospective that I inevitably find. The nine thousand and something whose version failed do not write a matching post titled “I did the same thing and it didn’t work,” so their data vanishes from my feed as completely as the downed planes vanished from the runway. I am reading the returning bombers and armoring the wings.

What makes it so hard to catch is that the stories are not false. A survivor’s account is usually accurate about their own experience — the morning routine did work for them, the cold email did land, the framework did carry them through. The bias is not in the individual story, it is in the selection of which stories I ever get to see. A true story about a lucky person is still, statistically, a sample of one, dressed up as a method.

This is why I have come to distrust the professional smiling-on-a-podium genre, where a person explains the three habits that got them there. The habits may have helped, or may have ridden along on luck, timing and a hundred unstated advantages, and there is no way to tell from the stage. The people on the stage are the survivors, and the fact that they are all smiling tells me nothing about the base rate of people who did the same three habits and got a different result.

The uncomfortable part is that the missing data — the failures — is not missing by accident. It is systematically unwritten. Nobody is incentivized to publish “my approach failed,” both because failure is painful to write and because it gets no attention when written. So the ecosystem of advice has a built-in survivor filter, and the filter compounds: the more I read, the more the surviving advice sounds like a law of nature, and the less visible the failures become.

This is the precise way knowing a bias has a name fails to protect me from it. I can label survivorship bias perfectly well while still nodding along to survivor advice, because the bias operates upstream of my evaluation — it already pruned the failures before I ever got to apply judgment. Awareness at the reading stage is too late; the skew was baked in at the writing stage.

The first countermeasure I actually use is to demand the denominator. For any piece of advice, I ask “how many people tried this, and how many of them got the result?” Usually the answer is unknowable, because the denominator was never collected — and the fact that it is unknowable is itself the information. If a strategy’s success rate is genuinely unquantifiable, then its published success stories tell me far less than their confidence implies. Asking for the number I cannot have is how I hear the silence where the failures would have been.

The second countermeasure is to go looking for the failures on purpose. Instead of only reading the winner’s retrospective, I search for the people who tried the same thing and stopped — the abandoned newsletters, the deleted channels, the “why I quit” posts that exist if I dig. This is the mirror of measuring my own optimism against a realistic baseline: I am deliberately re-balancing my sample by adding back the people who never came home.

That search is uncomfortable, and the discomfort is the point. Survivor advice is pleasant partly because it flatters the reader — if I just copy these habits, I too can be on the podium. The failure stories say the opposite: that the same habits, plus ordinary luck, produced someone who is not here. Preferring the former over the latter is not rational, it is a comfort preference, and comfort is a terrible lens for deciding what to actually do.

The deepest version of the bias, though, is the one I apply to myself. When I succeed at something, my own memory quietly edits the story: I credit my method and forget the luck, the timing, the other people and the second chances. The next time I give advice, I am a survivor too, armoring the wrong spots, convinced the wings were the thing that saved me. The way I remember being right is itself a bias I have caught in the act, and it turns me into the very source I should be distrusting.

None of this means survivor advice is worthless — it means it is evidence of a different kind than it pretends to be. A survivor tells me what was possible; they do not, on their own, tell me what was probable. To get from possible to probable I need the denominator, the missing failures, the base rate, and the honest admission that I will probably be in the larger, quieter group. The advice I trust most is the advice that tells me which group it thinks I am actually in — and most of what I read never once acknowledges the group exists.