I just listened to Planet Money's latest podcast, The Case Against Patents, and had several comments on it. (And perhaps I should preface this by mentioning that I'm a regular reader of TechDirt, and if you search for patents there, you'll find a lot of problems with the system. And probably links to the works of the economists interviewed for the show.)
My biggest problem with the podcast has to do with the language used. In particular, referring to working from someone else's patent as "stealing". It isn't stealing; nothing is being taken. And if I invent something, and can't use that idea better than someone reading my patent, I deserve to lose the possible opportunity cost losses from that person using that idea. That opportunity loss is the only thing "taken", and under no circumstances should that be referred to as stealing.
Further to that point, developing a patent is roughtly one percent of the cost of developing a product. If I develop that patent first, I've got a leg up on doing that other 99%; I've started first.
And frankly, if the market opportunity itself isn't enough incentive, in its own right, to turn that idea into a product, then the idea isn't worth a patent anyway (or shouldn't be).
And a note on drugs: putting the government more directly in charge of choosing directions would certainly have some downsides, but here's one upside. There would be development on antibiotics again; right now, there's almost nothing, because it doesn't pay well enough. They'd rather develop drugs for chronic conditions, because it forces people to keep buying the drug.
Also, the governments review all the studies that drive approval forward already, so the government knows how it's done.
There's another benefit as well. When studies show bad side effects, they are sometimes re-done to try to avoid those showing up (which is one of the reasons we get things like Fenphen and Ephedra going on the market, even though they'll sometimes kill people. Not the only reason, but it does happen). When those studies are re-done, the government usually doesn't see the bad studies until after the fact (if ever). If the government is directly running the studies, they'll see all of them.