Great article in the WSJ a few days back. This article talks about recent research (in particular by Harrison Hong of Princeton) that looks at why bubbles form.
1. Bubbles occur when investors disagree about the significance of some event. The internet being the obvious one. It's hard to bet on prices going down (limits to arbitrage etc), so the optimists dominate and prices shoot up.
2. Bubbles are then identified by intense trading.
3. Bubbles continue even when "smart" investors know that prices are too high - as individually they cannot attack the bubble. Only when they act simultaneously can their actions have an effect.
The conclusion is that it might be in the best interest of the Fed to contain bubbles rather than let them run their course.
Robert Shiller in his book "Irrational Exuberance" talks about bubbles also. He focuses on the feedback loop between the media and market participants. Bubbles make news and are great for exchanges (lots of trading) so plenty of people have an incentive to keep them going. Bubbles also falsely give the impression to investors that they have stock picking skill when in reality they are just riding the bubble with everyone else. The skill is not in buying the stock, but knowing when to sell.
On a related note - an interesting article on the freakononmics blog here talks about how we tend to think we are above average (drivers, stock pickers etc) which relates us back to a recent post on stock trading and speeding....