The experiment design was flawed in the following way. What if, in real life, there are people who are much more likely to take big chances with their whole career trajectory? Also, what if there are people who are much more likely to perceive which products will change the world, and when. The combination of these two qualities is the key.

Even if these qualities are distributed on a normal curve, the result of possessing the right combination of qualities will have an extremely large effect in a winner take all world.

Sure, there will be some randomness involved. Start with two kids who both have the same optimal qualities, and only one will end up a billionaire. But my guess is that the second kid will also end up ahead of the average. If the simulation did not end up with this result, then my guess is that something was left out of the simulation.