Hi, I'm Dave Mulder. This is my website, where I write about user experience and product design.

Originally Published: March 12, 2010

Survivorship bias

Imagine a room filled with 1000 stock market experts. They are asked to predict how Widget Industries (WIDG) will do over the next six months. Half the experts predict that WIDG will be up, half predict that WIDG will be down.

Over those six months, WIDG loses money. Half the experts were wrong, half were right. So you ask the people who were right, since obviously they know something the other half did not, what will happen over the following six months. Half of the remaining experts predict that WIDG will be up; the other half predict that WIDG will be down.

WIDG loses money again. Of the original 1000 experts, 2500 were correct after making two predictions. If we continue playing this game out to eight predictions, only four experts will remain. These lucky folks are afflicted with surviorship bias.

No, it is not necessarily a bad thing. Survivorship bias simply means that, in a field largely driven by chance, you survived. Though you may think you relied on your superior intelligence, or wit, or timing, you are actually the beneficiary of a rolled die (though hard work could improve your odds).

Blogging is a decent example. While millions of people blog, only a handful of bloggers have achieved fair success (enough to make blogging their career). Are we to assume that they owe their prominence to simply being a better writer or gifted person? Absolutely not. Survivorship bias and network effects are at play here.

For the most part, the successful folks in chance-driven fields are modest individuals; they recognize the role that luck has played in their lives. But they still write books, and articles that get retweeted again and again.

But, we do not need to treat their ideas like gold. They are, quite simply, ideas; and they often reach the public spotlight for no reason other than that their author got lucky a few years ago while building her audience.