Decentral Intelligence Agency

I have yet to see a good argument for creating a new director of intelligence. It's true that the intelligence agencies failed to share information. But an epi-central director of intelligence doesn't solve that problem and may make it worse. The implicit model of the 9/11 Commission is command and control - move all the information from the roots of the tree to the top of tree and then one all-encompassing-mind will evaluate it and make the right decision. Does that model sound familiar? Sure it does, that's the model of economic planning that is currently lying on the ash-heap of history. It's the model that Mises and Hayek subjected to withering criticism in the socialist calculation debate of the 1930s.

In brief, consider the following two defects of the economic Czar model. First, even if the information were to make it all the way to the top it would be difficult, well nigh-impossible, for a single mind to grasp it all and make it useful. This is especially true when there are no prices and hence no way of aggregating the information into a common unit (the so-called terror market was one way of alleviating this problem). Second, information is lost as it moves up the hierarchy - it has to be because not all information is easily communicable, bandwidth isn't infinite, and the people at the top demand information loss because as you move up the tree the amount of information becomes overwhelming.

An intelligence-Czar faces exactly the same problems. So what can be done? The intelligence agencies need tools that can spread information rapidly and widely and that are open to anyone with information whether they are at the bottom or the top of the hierarchy...Sound familiar? Yes, blogs and wikis are the right idea. And no I am not being flip. A central information repository that everyone can access may be part of the solution but centralizing information is not the same as centralizing decision making authority (remember "groupthink?" - the solution to our intelligence problems must face the problems of 9/11 and the problems of Iraq which are not the same.) Other ideas are to reward information sharing instead of hoarding - we should probably classify less information not more - and to rotate staff across bureaus in order to encourage collaboration and informal information sharing. Others more expert in this area will have more specific suggestions but my primary suggestion is that the models to follow are those of markets, webs, and networks.

Addendum: See also Tyler's related and important discussion.