It's not that earth-shattering - people have been saying this for years - but I haven't seen it put this well, for mass consumption, in a long time. Phil Bobbitt writing in Newsweek:

It is often asked, "How can we win a war against terror? Who would surrender? How can we make war against an emotion (terror) or a guerrilla technique (terrorism), neither of which are enemy states?" These questions assume that victory in war is simply a matter of defeating the enemy. In fact, that may be the criterion for winning in football or chess, but not warfare. Victory in war is a matter of achieving the war aim. The war aim in a war against terror is not territory, or access to resources, or conversion to our political way of life. It is the protection of civilians within the rule of law.

I like this formulation - a lot. I like it because it responds squarely to terrorism itself, rather than terrorists as would-be players on the global stage. I like it because it does not limit its reference to US citizens but to civilians everywhere. But I should add it doesn't quite answer the question posed: what are benchmarks for 'winning'?

The rule of law part is easy. Don't torture; don't arbitrarily detain; don't let invasiveness of security procedures outweigh security gains. But to make that calculation you need metrics of acceptable security gains. You need some basis for declaring that civilians - ours and theirs - are sufficiently protected. I'm not talking color codes and threat levels. I'm talking about a way of judging how effectively the government has protected against additional terrorist attacks without violating the law.

Now I know that the US government has made so many mistakes legally that it's almost a hypothetical question. But counter-factually, suppose the Obama administration upheld every constitutional principle, starting now. How could we judge its success at forestalling new attacks? How could we set a threshold for declaring "victory" that would neither be invalidated by the next successful or quasi-successful attack, however small? Counter-factually, what level of civilian protection in Afghanistan or Swat would constitute a "victory" over terror there, given that it could never be 100% as long as some criminals are bent on murdering innocents? If terrorism were kept within a certain risk threshold and citizens had no worse to fear from their own government, wouldn't that be a victory? And if so what is that threshhold exactly?

Paul Campos had some thoughts in the WSJ last week. If you compare  Americans' risk of death by terrorism to death by traffic accident or gunshot wound, he'd argue we're already well within the "acceptable risk" margin and likely to stay there no matter whether a few terrorists succeed or fail here and there at doing something more than setting their undies on fire. (This begs the question of whether the risk margin for traffic accidents is actually too high from a human security perspective, but either way, terrorism is hardly the biggest threat to life in our country.)

But Newsweek's editors seem to have taken a different message from Bobbitt's argument - that it's impossible to define victory. Instead of taking seriously the idea of how to measure victory on Bobbitt's terms, their latest issue features a long, admittedly interesting but ultimately distracting conversation about how ambiguous the concept of "victory" is today. That whole discussion misses Bobbitt's point, I think. Victory on conventional terms is no longer possible in asymmetric wars. Instead of belaboring that, let's redefine our terms and create some valid metrics to go with them.