[eanzenberg]

Because it's very useful for symmetric distributions. If the distribution is highly non-symmetric, then the mean != maximum likelihood, which is probabilistic.