Underbelly: I posted about this before but I shouldn't let pass this cool summary from The Economist on the question of when government industrial policy--aka "picking winners"--actually works:
First, the more it is in step with a national or local economy’s comparative advantage, the more likely industrial policy is to succeed. Drives to spur high-tech entrepreneurship in areas of heavy manufacturing, for instance, face a struggle. According to Mr Lin of the World Bank, following comparative advantage has produced clear successes for some developing countries. Chile, for instance, moved from basic industries such as mining, forestry, fishing and agriculture to aluminium smelting, salmon farming and winemaking thanks to a number of government initiatives.
Second, policy is least prone to failure when it follows rather than tries to lead the market. Curiously, Sheffield Forgemasters might have been an example of the former: Westinghouse, an American company, had suggested to the Yorkshire firm that it should try to break Japan’s monopoly on ultra-large nuclear steel forgings.
Third, industrial policy works best when a government is dealing with areas where it has natural interest and competence, such as military technology or energy supply. The worst problems unfold when politicians intervene in purely private domains with short-term goals, bailing out old firms to save jobs or spending lavishly on white elephants. The present round of industrial policy will no doubt produce some modest successes—and a crop of whopping failures.