The author has an annoying persona, but the ideas are very intriguing. He divides phenomena into two classes: mediocristan (where variability is constrained by physical properties, and standard statistics applies) and extremistan (where exteme events can be expected to wildly skew results). Markets and most financial statistics he places in extremistan (although the division seems highly arbitrary). In light of recent economic events, his thinking seems persuasive. But it's hard to see it as a guide to action.