Page 1 of 4

Brigadier General H.R. McMaster, USA Hoover Institution Research Fellow

Talk at ODNI on January 19, 2010.

At the beginning in Iraq, there were unexamined dynamics. The US simply didn’t recognize that many conditions had been set that made civil war likely. A kleptocracy had suppressed tribal order. Illicit networks, built to bypass UN sanctions, were very strong. The US entered with no understanding of or preparation for these dynamics. In Afghanistan, how do we avoid repeating this pattern?

1. Deal with the first-order questions. Following Clausewitz, identify the kind of war we are in. It was a simple matter to unseat the Taliban. The real issue is dealing with the collapse of the state, which creates opportunities for bad actors. Construct order.

2. Make intelligence relate to policy and operational objectives. If we need an economy, then focus intelligence analysis on the decisions that follow, such as what kind of banking is needed. If corruption needs to be suppressed, then discover how that works through networks and plan to stop it. If need ‘rule of law’, find out how enemy is intimidating population and work around it; for population protection, look into patterns of association. In Iraq, we didn’t identify how Iran had penetrated and captured ministries, and that was a key factor in establishing lawful government, a policy goal. Through 2006 the insurgency had driven toward civil war and had undercut the will and capacity for lawful government, and intelligence had not characterized this situation.

3. Understand that conflict will evolve. Assessment has to continue and recognize change, be capable of revision. In order to detect important changes, intelligence has to be able to view the situation from outside, though also at the same time be in close collaboration with military/political actors who are executing the policy.

4. Understand adversary organizations. See Sinno’s Organizations at War in Afghanistan and Beyond. He doesn’t buy the concept of “leaderless Jihad.” It is only an organization with some central control that can conduct war. A loose network can remain resilient, but it only becomes effective when it forms an organization, and once it does, it connects to legitimate institutions in finance, media, etc., and can be interrupted and influenced. The reason there are no bombs in Dubai is that the funding flows through there. He also likes Drugs and Contemporary Warfare, Army War College.

5. Place military intelligence in context. There is endless data that could be collected, but you need to avoid overspecializing and piling it up. Concentrate on synthesis that matters. Also, get the information where is resides naturally and don’t overlap the effort. For example, your local partners will know a lot about counterintelligence, so depend on them. Also, make the process enduring, don’t let troop rotations interrupt relationships. Also put more emphasis on the “battleground of perceptions”, get what you need for counter-messaging. For example, detention facilities become training facilities for adversaries, when they might become something else, such as way to collect much better intelligence during interrogations. Interrogators are asking all the wrong questions and aren’t taking the huge opportunity to learn about political factors.

6. In COIN, the local dimensions are very important. There is too much aggregation in how we treat matters, and this is exacerbated by bullet point briefings that relentlessly oversimplify. Decision makers have to actually start reading full reports, or at least pair up with briefers who make them understand the complications. In the way we communicate, we create an illusion of control and understanding. In most situations a true expert is available and should be brought in. For example, there was a West Point officer who did a PhD dissertation on how a tribe in Baghdad related to the British occupation. His advise was invaluable, and there are a lot of civilians, academics, etc. who need to be consulted and listened to. Micro tribal relations, local grievances, class, ethnics, all matter and require special treatments.

7. There is a regional, transnational dimension. It is hard to keep all these levels straight, but none of them trump the others.

8. Even your best analysts won’t know as much as indigenous sources. Just ask people, and they will often tell you what you really need to know. (But do compare and check.) Also, you have to know what to ask, and be familiar with the history, which matters to them and affects their thinking and allegiances. In intelligence, we tend to train people in a process, but that is not very important! It would be much better to read a stack of books on your subject matter. We assign people who have no idea what is going on. And especially for those people who do interact with the locals, they have to be educated, and they have to be required to report fully what they are learning from their interactions.

9. Intelligence staff need to think like operators and policy makers. Make friends, know the concerns. An example of where intelligence worked was the Counter IED Operations Center, also in some fusion cells. The relationships are tight, and the analysts fully understand a political strategy.

There are a lot of lessons, but if they are learned, they tend to stop in the operating force and are not institutionalized in the generating force. Part of the problem there is that we load up the generating force with contractors and aren’t able to flow back people with recent operational experience.

Question: Why were we so unprepared? (Also, why did we let ourselves forget what we once knew?) Answer: (no comment regarding political leadership) We got used to doing “the fun part” of war, and also took on the hubris of “full spectrum dominance.” This sort of net-centric/rapid/joint force/revolution in military affairs thinking made a kind of sense in the other services that deal with well-defined targets, but we in Army allowed ourselves to assume that this technological approach applied to land warfare. It was uncritical, and let us concentrate on cases like the Gulf war, where we had a limited objective (no regime change) and a very inept enemy. But we should have been realistic, and we did eventually adapt. And now we have new capstone concept, and a COIN manual. With McKiernan, we kept with a “persistent raiding” strategy too long.

He likes the NIC 2025 report, and the chapter 5 on Iraq. That project said in 2003 what we should have been thinking about, a coalescing insurgency.

There are problems in getting levels of intelligence to make sense together, J2, NGIC, and walking down to tactical.

We need to use outside support. There are civilians, partners who have anonymity and can move freely.

A big item is building the legitimacy of the security forces, and driving out corruption in them. This can’t be episodic. Also keep all three parts (US forces, police, military) in unity of effort.

In technical assistance to bureaucracy, need to stop imposing international best practices. Accept what is traditional, don’t build a new bureaucracy.

Select attacks that will reinforce political strategy and accommodation of groups.

Accept Flynn’s criticism. There was under-resourcing of intelligence. Bad understanding of requirements and changes in strategy. Inability to ask right questions.

There is sometimes a sense of “social science run amok.” People collect a lot of data and presume that “it” will tell you an answer. Can’t get much from a database and IT networks, but contractors keep pushing and we keep buying. But what really need is experts from anywhere, context, and also need to ask the soldiers! Make Phebe Marr a general! Also pay attention to: Charles Tripp, R. Kadeiri, Sarah Chayes, Barnett R. Rubin and Ahmed Rashid, Michael Howard, Frontline piece on Children of the Taliban, Fariad Ali Han on borders. Educate analysts (and self-educate) on the place, don’t waste time training them on the process.

As posted to Phi Beta Iota, the Public Intelligence Blog, 20 January 2010