Opponents argue that killing people is wrong, or that killing innocent people is wrong and even the best legal system sometimes makes mistakes, or that a system that sometimes makes mistakes ought to limit itself to mistakes that can be corrected—and letting a wrongfully convicted defendant out of prison is easier than bringing him back from the dead. I find all three arguments unconvincing.
If killing people is wrong, we should have no military—and although pacifists accept that position, neither I nor most opponents of the death penalty agree. Killing innocent people is a bad thing, but in a world of uncertainty it cannot be entirely eliminated. Every driver on the road faces some small risk of a heart attack, epileptic fit, or stroke that would convert his car into a lethally unguided missile. It does not follow that nobody should drive.
If the death penalty deters more effectively than other punishments, then using it lets us either deter more crimes and so reduce the number of innocent victims who are killed or raise our standard of proof while maintaining deterrence, convicting fewer people, hence fewer innocents, but punishing them more severely. Executing an innocent defendant is a bad thing—but perhaps less bad than letting two innocent victims be killed or locking up three innocent defendants for the rest of their lives.
The irreversibility of the death penalty is good rhetoric but bad argument. It is true that a mistaken imprisonment can be in part corrected, while a mistaken execution cannot. But in practice the criminal justice system very rarely discovers its mistakes—and an unjust imprisonment that is never recognized as such is just as irreversible as an unjust execution.
There is an argument against the death penalty that I find more convincing than any of these—one which is, as it happens, also an argument for it. Executing people is cheap, imprisoning them expensive. This is not true in the U.S. at present, because sentiment against capital punishment has resulted in elaborate and time consuming procedural constraints on its implementation. But in a society that is serious about capital punishment, hanging someone costs a lot less than housing, guarding and feeding him for fifty years.
In a legal system run by benevolent philosopher kings, cheap punishments would be an unambiguous benefit. In our world, it means that a large cost—the loss of a life—is imposed on someone else by people who bear a very small cost for doing so. Having A make a decision most of whose costs are born by B is a recipe for bad decisions—in this case lethally bad.
The point is nicely illustrated by a famous, but probably apocryphal, historical incident. During the Albigensian crusade, one of the leaders supposedly asked the Papal legate how they were to distinguish the heretics in the city they had just taken from the good Catholics who happened to be their neighbors.
“Kill them all. God will know his own.”
The risks of cheap punishments.
For a broader and more academic discussion, see my "Why Not Hang Them All: The Virtues of Inefficient Punishment," Journal of Political Economy, vol. 107, no. 6 1999 pp. S259-269. There is a webbed version in my Law's Order (search for "Why Not Hang Them All).
[I have ignored a fourth argument against the death penalty—that it doesn’t deter—since I believe the factual claim is probably false. That would be—for the past twenty years or so has been—a longer argument.]