Krugman wrote an article that appeared yesterday in the NYT defending the Tobin tax, a tax on financial transactions, as a way to "deter financial speculation."
Financial speculation by itself isn't a problem, is a solution, and as an economist he should know that. Krugman's mistake is worse however: his article transparently shows that he doesn't know anything about the recent literature on the subject (or doesn't bother to show his knowledge on it), but speaks as if he knows it (or as if he's showing his knowledge).
The Tobin tax has been discredited on many different grounds by economists that have spent much more time than him trying to understand it. There's plenty of experimental, theoretical and empirical evidence against taxes that work like a Tobin tax. There's some less clear evidence in favor of it too, which however tends to show that the effects are only positive in special cases (probably not found in real markets) or that the benefits are not worth the costs.
Besides all the scientific evidence, Brazil just adopted a Tobin tax and stock market volatility didn't fall after its introduction on October 20, as seen in this Bovespa graph:
may have deserved a Nobel Prize in Economics for his contributions to international trade theory, but he doesn't help the economics profession when he writes on things he doesn't have a clue about.