George Will sings the praises of my great-great-great (maybe one or two more greats...I'll leave it to my father the genealogist to figure out) uncle, Gen. William Tecumseh Sherman.

One of Sherman's less well known quotes opens Will's column: "I fear the world will jump to the wrong conclusion that because I am in Atlanta the work is done. Far from it. We must kill three hundred thousand I have told you of so often, and the further they run the harder for us to get them."

To summarize Will's point:

Donald Rumsfeld says his preference is for al Qaeda fighters to surrender rather than fight to the death: "It ends it faster. It's less expensive." Gen. Richard Myers, chairman of the Joint Chiefs, says: "This is not a war of extermination." Such statements are perhaps obligatory and even sincere.

However, is surrender really less expensive in the long run? It is a reasonable surmise that a reformed terrorist is a very rare terrorist, and that the rate of recidivism will be high among terrorists who are forced to surrender but continue to believe they are doing God's will when they commit mass murder of infidels. So, as far as is consistent with the rules of war and the husbanding of the lives of U.S. military personnel, U.S. strategy should maximize fatalities among the enemy, rather than expedite the quickest possible cessation of hostilities.

We've already seen the truth in this as Taliban leader Mullah Mohammed Omar has "disappeared" amid the mass "surrendering" that occurred when Kandahar fell. We don't have nearly enough al Qaeda prisoners. The Marines on the ground in Afghanistan have to be on alert because of fears that there are many of these fanatical Arab fighters who've surrendered, only to regroup in the hills.

A dead terrorist is preferable to an imprisoned one. A dead one isn't likely to escape, kill guards or any of the other mischief that prisoners can so often get into.