ANOTHER WORD OR TWOJeff Medcalf
has added his own take
to the discussion on abolishing the CIA
that was initiated by Dave
. An excerpt:"The entire structure of our intelligence agencies — military and civilian, agency- (CIA, FBI) or department- (State, Defense) based — is structured to prevent an enemy from acquiring the capability, and acting on the intent, of using their military to attack an unprepared United States. At that function, our intelligence agencies are supremely good, probably unmatched except by the British and possibly the Israelis.
But our intelligence agencies are unable, due to the very structure that makes them good at preventing a Pearl Harbor repeat (think 9/11 with bombers instead of terrorists), from institutionally understanding non-state actors the way they can understand states. And since that is structural, nothing short of structural reform will fix it: Dave is absolutely correct there. But I do agree with Mark, also: we don't want this to be a purely military function.
What I would suggest as an organizational model is a broadly-distributed network with minimal bottlenecks and control nodes. There should be small agencies geared to particular methods of intelligence gathering (electronic intercept, covert spying, reading the newspapers of the world, etc) or particular types of information (military construction, equipment design, agricultural output, talking points in negotiations, etc). These agencies should feed the information and the source of the information into a single agency whose job it is to evaluate the intelligence's credibility based on past experience with that source or method rather than on how "believable" the intelligence is, and to sanitize the information to include the evaluation of reliability, but remove any information that would identify the source or method used. This evaluated information could then be used by analysis cells attached to every policy decision maker, as well as feeding into certain field operations (most notably, the military). Organizations with particular needs (battlefield and theater intelligence for the military, political intelligence for an embassy) would retain the ability to gather intelligence themselves, and use it directly, while also feeding it into the evaluation agency for the rest of the government to use. "
I would add the suggestion that an analytical unit dedicated to the alternate methodology
of scenario forecasting
, which could be housed in the NIC, would be a useful addition. In particular, for problems in which there rests a high level of uncertainty
. Another one, would be to build more thorough metacognitive practices
into the analytical process, as suggested in this substantive, yet very readable, CIA monograph:The Psychology of Intelligence Analysis ADDENDUM:Art Hutchinson
of Mapping Strategy
stopped by to offer a clarification in the comments section. As Art is a nationally known consultant on strategic thinking, I thought it would be useful to highlight his remarks here:"Scenarios and forecasts are fundamentally different animals, though they share the quality of being subject to rather easy manipulation by those responsible for architecting the process. The one (scenarios) imagines possible futures, some of which may be well outside the realm of what one might call a "forecast" for the purposes of getting groups to think outside a conventional frame. The other (forecasts) are IMHO useful in the current geopolitical context only to the extent that they incorporate dynamic, highly distributed opinion-forming mechanisms (of which highly distributed sources are a necessary but insufficient component)... which brings the whole thing back 'round to prediction markets - those clever devices that Tradesports is using to precisely forecast things like presidential elections better than polls or pundits, but which died an ugly death in the public square with the tarring and feathering of Adm. Poindexter and the Policy Analysis Market 3 years ago.
Net/net: Yes, the CIA could benefit from both interactive scenario-based thinking processes and prediction markets. They will not benefit if they seek simply to divine a single "most probable" future scenario. Once birthed, those behemoths tend to live on for decades in big institutions and reduce rather than facilitate the clear synthesis of unexpected data."
Thanks Art !