Wednesday 25 Nov 2020 | 02:55 | SYDNEY
Wednesday 25 Nov 2020 | 02:55 | SYDNEY

So what are intelligence agencies really for?


Sam Roggeveen


10 January 2008 13:18

My post of yesterday brought this further response from Paul Monk: 

Tetlock’s study of predictive judgments demonstrated just how faulty even well-credentialed experts tend to be in their estimates. This certainly applies to intelligence analysts, and the world of secrecy within which they are cocooned exacerbates the problem in several ways. Yet intelligence analysts or agencies are asked to make estimates or predictions, often at short notice and without in-depth reflection. This raises two questions: your original question as to whether this is the primary purpose of intelligence agencies and a second question, as to whether they can deliver the goods.

My own argument, in my earlier comment, was that the primary purpose of intelligence agencies has always been to gather secrets, keep them secret and communicate them to their sovereigns (not ‘the people’, let us be clear, but a select set of representatives and civil servants of ‘the people’). A study of the history of almost any intelligence service will show that its political masters seldom pay heed to its estimates or predictions, unless these are especially alarming or conform to their already preconceived views and policy preferences. Is there some normative sense in which it should be otherwise? Perhaps, but we would then need to have considerable confidence that the agencies were good at their estimates and predictions. Alas, it is not only Tetlock’s work which suggests that such confidence would be misplaced.

What political decision-makers generally do is trust their own estimates and seek from intelligence agencies data to support them. This is the case not only with dictators (Stalin was notorious in this regard), but the leaders of democratic states. There is, also, an increasing demand for current intelligence, which detracts from whatever capacity the agencies might otherwise have for taking a longer or deeper view of things. As CIA veteran Carl Ford has observed, this results in them churning out piles of insubstantial fluff, too often trying to feed the perceived preferences of their political masters. The consequences are themselves rather predictable. The real 'intelligence' function – running scenarios of future possible challenges to one’s assumptions and preferences, checking for the possible emergence of system-destabilizing surprises, adjusting and calibrating the actual thinking behind policy commitments – simply does not take place, in a systematic manner, at any level.

Daniel Ellsberg’s memoir, Secrets, and his book of reflections thirty years before, Papers on the War, still repay close reading on this subject. His classic observation about the underlying cognitive problems that were hamstringing the US Government in the Vietnam War remains pertinent in ever so many cases, right up to the present: the US Government, starting ignorant, did not and would not learn. There was a whole set of anti-learning mechanisms in place which guaranteed unsuccessful and maladaptive behaviour, including pervasive secrecy, which enabled Government officials at many levels, but especially the President himself, to get away with errors and to believe they had choices which, rigorously speaking, they should not have had. Are intelligence agencies ‘for’ correcting these problems? Perhaps notionally so, but they simply cannot. The rules under which they operate make it all too difficult. We need to understand the constraints on intelligence agencies and either work hard at better equipping them, or rely on other means for seeking to challenge and correct strategy and policy.

Photo by Flickr user deepsignal, used under a Creative Commons licence.