When Ignatius endorses an idea, it is time to reconsider

The problems with the CDI are the same problems that bedeviled past efforts to build Afghan tribal militias. Here's how Filkins summarizes the program in his article:

In the Pashtun-dominated areas of the south and east, the anti-Taliban militias are being led by elders from local tribes. The Pashtun militias represent a reassertion of the country's age-old tribal system, which binds villages and regions under the leadership of groups of elders.

[...] In July, a long-running dispute between local Taliban fighters and elders from the Shinwari tribe flared up. When a local Taliban warlord named Khona brought a more senior commander from Pakistan to help in the confrontation, the elders in the Shinwari tribe rallied villagers from up and down the valley where they live, killed the commander and chased Khona away.

In other words, the militias are short-term success stories, because their local interests currently line up with those of the central government. The Shinwari tribe doesn't like the Taliban; Hamid Karzai doesn't like the Taliban; Gen. Stanley McChrystal doesn't like the Taliban. Everyone wins! (Except the Taliban, of course.)

But what happens next? Remember, the militias represent local interests; they have no particular loyalty to the Kabul government or CENTCOM. Filkins writes, vaguely, that the Americans "will tie [the militias] directly to the central government." A more likely outcome is that the U.S. will empower local militias at the expense of the central government.

That's what happened to the Soviet Union, which adopted a similar strategy. The militias held off the mujahideen, but they also undermined the government in Kabul. The leaders of those militias -- Rashid Dostum is perhaps the best-known example -- became the warlords who fought incessantly throughout the 1980s and 1990s.

It's particularly dangerous to base the militias on tribal identity. Gilles Dorronsoro discussed this in his Afghanistan policy brief (pdf) earlier this year; he noted that tribes are prone to double-crossing, and that the U.S. lacks a keen understanding of tribal politics. (Boone makes a similar point in his Guardian article today.)

The tribal approach also assumes tribes are coherent entities which can be bought off -- a fallacy, as Christian Bleuer wrote in a post earlier this month:

... all the proposals for Pashtun tribal militias that will hopefully take on the Taliban all assume one fallacy: that Pashtun tribal identities, when and where they exist, are coherent social and/or political entities with a hierarchy of authority. They are not. There is no leader of, say, the Examplezai who can be approached, recruited and paid to deliver all of the Examplezais. THERE IS NO CHIEF. THERE IS NO LEADER. THERE ARE ONLY NOTABLE ELITES WITH FLUCTUATING LEVELS OF INFLUENCE.

Finally: By empowering local militias, the U.S. meddles with local power structures and creates winners and losers. What happens to the losers? They get angry; they retaliate; a cycle of violence is born.