What other models can they use, though? Every other distribution assumes even more about the problem domain, and would thus correlate even worse with black swans. Sounds kind of like a no free lunch theorem for finance.
Anyways, do you know if anyone is trying to come up with something better?
There are other types of distributions which fit black swan type data better, such as power-law distributions, Gumbel distribution, etc. I don't know if research in extreme value theory is still a hot topic, but it tries to address these problems. Here is a paper on risk management and extreme value theory: http://tinyurl.com/aa7t6m
Here is a nice paper on risk measures, and points out some of the problems with VaR: http://tinyurl.com/ar3336
My apologies if you're unable to grok the contents of these papers...they're both a little technical but I can't think of anything off the top of my head that's more accessible.
Anyways, do you know if anyone is trying to come up with something better?