Big crashes will always occur, but I would not blame the recent crash on the gaussian models. I'd rather say that (almost) everyone underappreciated the risks connected to the real estate prices.Part of the reason they underestimated those risks is that they paid attention only to the middle of the distribution, where things are approximately normal. I wouldn't say the explicit Gaussian-ness of the models was the reason for the trouble, but it's hard to imagine a Gaussian model providing any sort of reasonable risk estimate for the type of thing that we saw happen. It was so far outside the "business as usual" range that no risk estimate based on what was happening on most days would have been legitimate.
Taleb pushes for a strategy that consists of buying a lot of very safe assets and blending them with bets on "extreme events" (like buying far out-of-the-money put options). Is that a viable long-term strategy? I have my doubts, since there are no evidence suggesting that 'uncertain' strategies have greater returns that more quantified ones.
I don't know about this; it's all a question of price. If far out of the money puts are really underpriced compared to how often they "hit", then he could be right. My immediate impression is that the crappy prices you tend to get due to low liquidity in the extreme tails might make a profitable strategy tough to come by.
One could certainly look at the historical data over the past several decades and see whether such a strategy might have been profitable (which wouldn't necessarily tell you whether it will be profitable in the future, but might shed at least some light on the matter), but I don't have options data going back very far, so I'm not the man for the job...