I don't go by the ratings. I buy wine that tastes good. Statistically, anybody's ability to predict what will be a good wine a decade from now is limited.

A good rule of thumb is to assume that everything matters.

The reason is they failed to learned the primary lesson we should have learned from when Long Term Capital Management went belly up ten years ago. That is, investments that seem uncorrelated can be correlated simply because we're interested in it.

Maybe you'll take the cash out. So a credit card company or a bank that goes into the business of saying we're going to be the broker, we're going to sell you a mortgage that you're going to be able to pay off, we're going to help you reduce your credit card debt, we're going to help you save for retirement, we're going to put you into mutual funds that have low fees rather than high fees.

If rather than setting the minimum balance as the lowest possible amount, so we keep people in debt for as long as possible, we raise the minimum payment and encourage people to pay off their credit cards, we're going to make less money, but we're going to have costumers that are more solvent.

So the world is much more correlated than we give credit to. And so we see more of what Nassim Taleb calls "black swan events" - rare events happen more often than they should because the world is more correlated.

When should we nudge and when should we shove, I think, it's a political judgment. Obviously in some situations we need shoves, we need laws. Fraud is against the law, murder is against the law, drunk-driving is against the law. We don't need just nudges.