The Obama administration wants to launch a $100 million initiative to map the human brain, which is being hailed by American Enterprise Institute's Jim Pethokoukis as an example of "what government should be doing."
He's right, and it's a good opportunity to discuss the much-misused term public goods. A public good, in the economics jargon, is something that's nonrivalrous and nonexcludable. In other words, if I use some of it, that doesn't leave less around for you, and once it's out there, it's just out there. Scientific knowledge is a great example. When I learn something, it doesn't mean that you unlearn it. But a big problem with public goods is that they're valuable because they're nonrivalrous but underproduced because they're nonexcludable. So we have a lot of public policy dedicated to making nonexcludable things excludable in practice. That's what patents and copyrights are for. We also do some to subsidize the production of knowledge. That's everything from the day-to-day work of the Bureau of Labor Statistics to this big-picture science. Direct financing of knowledge production is a great thing for the government to do, especially because alternatives such as patents and copyrights tend in practice to be counterproductive to the advance of science.
The relevant linguistic fact, however, is that the vast majority of public services the government provides—roads and buses and schools and mail delivery—aren't public goods in the technical sense. Which doesn't mean they're all bad ideas. Oftentimes direct government service provision is a plausible solution to the problem of monopoly (roads especially) or else a convenient form of homeowner resource pooling (parks, libraries, etc.). But the true public goods are generally the things you don't think about all that often, like the ability to get reliable information about how many people live in San Antonio or various scientific research grants.