Every time I hear about another drone blowing up another terrorist leader in Pakistan, I wonder how far that method of warfare can improve. Drone technology and tactics have made great strides. What is the limit?

You can expect normal improvements in drone flying time, vision, weaponry, and the obvious stuff. It's a safe bet there will be more drones in the sky. And the human intelligence that is necessary to find targets will probably continue to improve. For a place like Afghanistan, are drones plus effective intelligence enough to control the country?

Imagine the Taliban regaining power in Afghanistan. The problem with being in power is that it makes you relatively easy to locate, and drones can destroy anything they can find. There is no practical way for the Taliban to hold power if our drone capabilities reach a certain level. I doubt we are at that level, but could we get there?

I can't imagine a terrorist training camp lasting long if the sky is full of drones. And the heroine fields would only last as long as we wanted them to. We could also force people coming into or out of the country to use border crossings we control. Everyone else gets attacked by drones. That takes a lot of drones, and that's expensive, but probably not as expensive as old-fashioned occupation.

I think the next big leap in drone technology will be artificial intelligence for locating targets. Humans would still have to make firing decisions, but I can imagine drones finding suspicious patterns of movement on their own and alerting humans. For example, any vehicle that stops at night on a road used by U.S. ground forces might be suspected of planting an IED. A human could decide if the suspect was up to no good.

There are probably a number of movement patterns followed by insurgents and terrorists. Maybe drones could learn to detect children in any outdoor group, based on their relative size, and assume such a group is not looking for a fight. Perhaps combatants follow routes less travelled by enemy ground forces, or travel only at night, or have more metal objects with them. The point is that drones will someday do a good job of identifying suspected bad guys automatically.

One great benefit of using drones to target Taliban and Al Qaeda leaders is that turnover creates leadership incompetence. After the tenth time drones kill the number three operations guy, only an idiot is willing to take that promotion. The smart terrorists ask for transfers to the Quality Control Division to wait things out. So while it might be true that there will always be replacements, quantity doesn't compensate for smart leadership.

Perhaps the exit strategy for U.S. conventional forces from Afghanistan is more linked to drone improvements than to anything else. We just don't know it yet.