The value of ‘hierarchical attenuation’

Peter MerrittFunLeave a Comment

Or Figures Never Lie – We Do

In Reflections on Intelligence, R. V. Jones briefly reviews the British decision to adopt a convoy system during World War I.

The case brings to light two useful insights:

  • Know your data’s heritage and;
  • Value the opinions of junior analysts. It also emphasizes how easy it is to work at cross purposes within the same organisation.

The Problem

The Admiralty was dead set against a convoy system. One of their main arguments was that convoys would require an impractical number of escorts. They based their conclusion in part on the estimate that “5,000 ships entered or left British ports in a typical week.” As Jones notes, even in convoys of 40, this would require hundreds of escorts. Fortunately,

…When a relatively junior officer in the Admiralty, Commander R. G. Henderson, went over the figures he found that they included the movements of all ships over 300 tons. These therefore included all coasters and cross-Channel ferries making several movements each week. When he removed these from the reckoning by counting only ships over 1,600 tons, which were carrying the overseas trade, the weekly figures fell from 5,000 to around 250.

By applying a simple filter, Henderson reframed the entire problem. What was seemingly impossible suddenly became feasible. What is doubly interesting is the origin of the original figure. Apparently,

… The Admiralty had initially deliberately counted in all the smaller ships for good reasons, so as to play down the proportionate magnitude of the U-boat successes, and to make the neutrals think that they would be reasonably safe using British ports. As the war went on, it seemed that the Admiralty had forgotten how it had biased the figures, and it had become the dupe of its own propaganda … ”

It May Be The Right Answer - But Was It The Wrong Question?

Lloyd George and Churchill overruled the Admiralty and instituted convoys. Later, writing in The U-boat Crisis, Churchill noted the value of junior officers’ opinions:

“…In the naval service the discipline of opinion was so severe that had not the channel, or safety-valve, of the Committee of Imperial Defence been in existence, these opinions [of junior officers] could never have borne fruit or even come to light. The firmly inculcated doctrine that an admiral’s opinion was more likely to be right that a captain’s, and a captain’s than a commander’s, did not hold good when questions entirely novel in character, requiring keen and bold minds unhampered by long routine, were under debate.”

Hierarchical What?

Elsewhere in his book, Jones expands on the concept of understanding “the actual conditions of battle ‘at the sharp end.’” He lauds Churchill for knowing when to go through his immediate circle of advisors, and cut through to the level of detail where the work was actually done, a practice Jones refers to as “hierarchical attenuation.”

Speaking of Churchill, Jones states “He recognized, of course, that in intelligence he could only do this on rare occasions, but always his instinct was to have as few stages as possible between himself and the front line; these stages being manned by men whom he felt he could trust.”. Churchill, of course, did not invent the idea. In Command in War, Martin van Creveld describes a very similar concept. He first observes that:

“…Climbing through the chain of command … such reports tend to become less and less specific; the more the stages through which they pass, the greater the danger that they will become so heavily profiled (and possibly sugar-coated or merely distorted by the many summaries) as to become almost meaningless.”

To guard against this danger and to keep subordinates on their toes, a senior consultant needs to have in addition a kind of directed telescope—the metaphor is an apt one—which they can direct, at will, at any part of the problem (including your own organisation) in order to bring in information that is not only less structured than that passed on by the normal channels but also tailored to meet his momentary (and specific) needs.

Clarity – Always, Clarity

Why should you, the consultant, care? To consult effectively you must get as close as possible the “true” state of affairs (or, more accurately, the multiple perspectives of the true state of affairs). An initial briefing from the client, however wonderfully precise and packaged it may be, will have already gone through many stages of filtering (including the dreaded ‘internal politics’ which affect any human group larger than one).

But to do your job well, you‘ll have to be prepared to dig deeper if you want to truly understand the key system elements in actuality. Without starting the whole specification from scratch each time (another fault to which some of us are prone), if this means not plunging immediately into all that lovely design and coding whilst confirming what you’ve been told with some good old Business or Systems Analysis, so be it; this isn’t a hobby, it’s why we get the big bucks. Imagine your assessment/design/program as a parachute – would you wear it and jump, or are there a few checks you’d like to make first…?!

So, re-filter any suspect data–as Commander Henderson did with the “5,000 ships” figure– stepping out of the executive-suite onto the proverbial factory floor, do your best (within your established contractual relationship). Chase down the “truth” stubbornly and persistently, and if necessary, remind your customer that time spent refining the right question will pay off handsomely in the end. And the real adversaries of your client will be doing the same…

Further Reading: