When a fireman saves a child from a burning building, he deserves a reward. But the right reward is probably some combination of cash and adulation, as opposed to, say, a license to commit arson. There's a general principle here: It's a good idea to reward good behavior, but it's a bad idea to sanction bad behavior, even as a reward.
Ordinarily, everyone snickers when this principle is violated. My university used to reward excellence in teaching with a lighter teaching schedule. OK, maybe teaching fewer classes is not the exact moral equivalent of arson, but the irony has a similar flavor and it was not lost on anyone.
Why, then, don't we snicker at the patent system? Inventiveness is good for consumers, monopoly power is bad for consumers, and we reward inventiveness by granting 17 years of monopoly power. Why 17? It's someone's idea of a compromise between too little and too much. The result is both too little and too much. It's too little because a 17-year patent is frequently worth far less than the full social value of the invention, so inventiveness is underrewarded and we don't get enough of it. It's too much because any gratuitously granted monopoly power is too much.
The solution, clearly, is to reward successful inventors the same way we reward successful salesmen or successful baseball players: by paying for performance, not with monopoly power but with cash. Who should pay? Mark McGwire is paid, ultimately, by the fans who buy tickets, which makes sense because the fans enjoy watching him. But who should have paid Thomas Edison?
A plausible answer is: the people who enjoy Edison's inventions, which is to say the public at large, through their elected representatives. Thus Professor Michael Kremer, late of MIT and now of Harvard, proposes that all new patents should instantly be purchased by the government and placed in the public domain.
There is at least one successful precedent. When Louis Daguerre invented photography in 1837, the French government purchased the patent and placed it in the public domain. But Kremer seems to have been the first to propose automatic patent buyouts as a matter of policy.
The big question is: What price should the government pay? How can we know what a given invention is worth, and how do we ensure that well-connected inventors don't translate their political clout into inflated prices for their patents? Kremer's answer: Put each new patent up for auction. When the auction is over, flip a coin. If the coin comes up heads, the high bidder pays his bid and gets the patent. If the coin comes up tails, the government pays the high bid and gets the patent. That way, the government never pays more than some private bidder was willing to offer.
Better yet, throw a biased coin that comes up tails, say, 90 percent of the time. Then 90 percent of all patents end up in the public domain, which is not as good as 100 percent but far better than none at all. We do have to give the private bidders some hope of winning so they'll take their bidding seriously.
T here's still a problem: An inventor can rig the auction by getting one of his friends to submit a wildly inflated bid. If the coin comes up tails, the government overpays; if it comes up heads, the inventor essentially buys the invention from himself and suffers no loss. To overcome this problem, Kremer suggests that the high-bid auction be replaced by, say, a third-bid auction: The high bidder wins, but he (or the government) pays the third-highest bid. To rig an auction like that, you'd need three ringers instead of one--still not impossible but perhaps acceptably difficult. (The key insight here is due to the pirate Blackbeard: Three can keep a secret if two are dead.)
You might think inventors would grumble about settling for the third-highest bid, but keep in mind that in a third-bid auction, people tend to bid higher. The third-highest bid in a third-bid auction can easily be about the same as the highest bid in a high-bid auction.