Consider an interesting experiment: Suppose a 25-year-old has been playing Russian roulette in a casino. Then it is conceivable that in less than a year, this young man will go bankrupt.

But what if 100,000 young people are playing together? There are always a few of them, dozens of young people, who can become multi-millionaires before the age of 30, right? So, can you still trust the "successful experience" they share?

And because of the wrong attribution and survivorship bias, we tend to forget the failed cases and remember a small percentage of the successful cases, thus inferring many reasons and patterns for success, although they are often completely random.