Kristol plucks out this rather sensible point by Rahm Emanuel:

It would be reckless to make a decision on U.S. troop level if, in fact, you haven't done a thorough analysis of whether, in fact, there's an Afghan partner ready to fill that space that the U.S. troops would create and become a true partner in governing the Afghan country.

To which Kristol responds:

If it’s reckless to commit to 108,000 troops without a reliable Afghan partner, it’s reckless to ask 68,000 troops to fight without a reliable Afghan partner.

Well, yes. But that was true for the final years of the Bush presidency as well. Bush kept going because he could hardly withdraw, given his rhetoric about democracy, freedom, etc. But Bush had no strategy, except to keep throwing young Americans into a regional quicksand. I think it's pretty obvious that Obama's campaign critique of Iraq made it necessary for him to back Afghanistan. But he backed it as a justified war, not as a successful one. Yes, it seems he under-estimated the severity of the collapse there in his first few months.

But it's silly to argue, as Kristol does, that he could somehow have rescued the election process. He sent more troops for the election, but the depth of the corruption is not something more troops could have prevented.

So we are where we are. And it is perfectly reasonable for a young president to take stock and time in figuring where to go next. Kristol urges - surprise! - more war:

[W]hat’s reckless is further delaying the troop deployment orders. If the president issued the order now, he could always delay or revoke it later, if the political situation seemed truly insupportable.

And we know what Kristol would say then, don't we? This dog's breakfast of an eight-year war/occupation is not Obama's fault. It is, if anyone's, the neoconservatives'. And yet they now declare those tasked with cleaning up the mess as reckless. The chutzpah continues.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.