There is a wonderful corollary in basic statistics modeling that success tends to be a long-tailed distribution. Reason: the easiest framework that allows mode

There is a wonderful corollary in basic statistics modeling that success tends to be a long-tailed distribution. Reason: the easiest framework that allows modeling "the resulting random variable R is zero if any of the incoming variables vₙ is zero" is the product formulation, R = Πₙ vₙ which by taking the logarithm of the individual variables becomes a standard central-limit-theorem prediction; as a result success in such contexts follows a log-normal distribution[1], which is long-tailed, even if the individual factors fₙ = log vₙ do not look very normal. This then leads to a growth pattern called Gibrat’s Law[2]. [1]: https://en.wikipedia.org/wiki/Log-normal_distribution https://en.wikipedia.org/wiki/Log-normal_distribution [2]: https://en.wikipedia.org/wiki/Gibrat%27s_law https://en.wikipedia.org/wiki/Gibrat%27s_law