One of Felix Salmon's commenters offers a succinct summary of recent financial innovations:

The person most willing to take on risk is the one unaware he is doing so. He charges no risk premium...

The resulting market equilibrium is that the guy who is unaware of the risk ends up loaded with it. Then the music stops.

This summation awaits an elegant statement of the solution.  Felix offers one:

Financial complexity and innovation, on this view, are essentially tools of obfuscation. And it's easy to hide risks when risk-averse investors want debt-like products which retain their face value: such instruments tend to have very low volatility, and so look and feel as though they're low-risk, even if they're full to bursting with enormous amounts of tail risk. The answer, as I've said many times in the past, is for risk-averse investors to be willing to take a small amount of explicit market risk, and to move towards safe equities (utilities and the like) and away from debt. Because if they go to an investment bank asking for safety, they're likely to just get hidden risk in return.

But aren't a lot of the most risk averse investors funds or insurance companies with limits on the kinds of assets they can invest in? I'm not sure we can fix this problem without knowing the answer to the question we've been asking for a year now:  why did the ratings agencies underestimate the tail risk, and is that reason fixable?