Moulton Lava

Moultonic Musings

My Photo
Name:
Location: New England, United States

Thursday, October 31, 2013

The Tragedy of Bureaucratic Thinking

"In any bureaucracy, the people devoted to the benefit of the bureaucracy itself always get in control and those dedicated to the goals the bureaucracy is supposed to accomplish have less and less influence, and sometimes are eliminated entirely." ~Jerry Pournelle, Pournelle's Iron Law of Bureaucracy
Ever since the dawn of civilization, bureaucratic minds have sought to reduce esoteric theory, insight, and wisdom into a simple set of rules that a modestly educated clerk could administer.

Rulers beginning with Hammurabi of Mesopotamia adopted this divinely inspired goal of reducing all high-minded guidance to a compact set of rules.

Alas, history, science, and modern mathematics have collectively revealed that such a goal is not divinely inspired, but devilishly impossible.

In biblical times, it was the Tribe of Levi who played the role of the bureaucrats who sought to reduce the esoteric wisdom of Torah (the word variously means "theory," "science," or "customary system of guidance") to a finite set (numbering 613 in this case) of rules.

This Magnanimous Legacy of the Levites (Levi Natan) came to be recognized by the esoteric theologians as a subtle mistake that was remarkably hard to explain. It's a (mathematical) mistake having to do with Chaos Theory, which reveals that rule-based systems are inherently chaotic (in the modern mathematical sense of the term).

The esoteric theologians noted this issue by creating the allegory of Leviathan (Levi Natan -- the Gift of the Levites -- but subvocalizing the (mis)leading N in Natan). Leviathan is characterized as an unbridled chaos monster.  Natan means Wonderful Gift.  Atan means banal gift.  Thanks a lot.

And that's what rulesets are. Unbridled sources of chaos.

It's a profound and unsettling insight, to be sure. But it's an essential one to appreciate. For if we wish to craft a graceful system of guidance, we have to do it properly, with properly chosen functions, and not by displacing those crucial functions with a simplistic set of rules.

In a word, rule-based systems are dysfunctional. They do not achieve the goal of a high-functioning guidance system.

4 Comments:

Blogger Higs; said...

Every act, no matter how unsuccessful, successfully tells a story.

6:23 PM  
Blogger Higs; said...

This is the nature of the beast--whether it be behemoth, leviathan, that skyqueen-thang, or otherwise. The world serpent, the world wolf, Loki's daughter, Hel, or some other allegedly daemonic brute.

The whole point of stories is that not all behaviors are successful. Not everything we do works. What doesn't work becomes discourse and reverberates throughout the cosmos.

6:36 PM  
Blogger Moulton said...

Alas, the stories are often stories of loss, stories of tragedy, stories of lamentation. Real life is not quite so generous with Hollywood endings.

8:54 PM  
Blogger J Thomas said...

Evolutionary systems typically react to natural selection by devising "rules", collections of genes that result in phenotypes that on average survive better.

The "No Free Lunch" theorem says that any ruleset you make up that solves some problemsets better, will solve other problemsets worse.

This is easy to visualize. Say you have a hill-climbing algorithm. It finds peaks by going up a gradient until it can't go any higher. You can design a situation where that approach will fail, simply by finding places that the algorithm will avoid -- lowest valleys -- and transporting the top half of each peak with sheer walls to such a valley. The algorithm will then avoid all the best places.

Any genetic algorithm that does better than random chance, does better by preferentially choosing some things over others, and it must parlay its good choices into better choices. It's always possible to create a pathological case where the good choices will not in fact lead to better choices. And then it will fail.

So genetic algorithms must be constructed for local conditions. Any algorithm will work in some times and places, and not others. Try to use it in the places where it will work.

For bureaucracies, we have the added problem of intelligent opponents. When there are people who need to outwit the bureaucracy, they will try to create situations where the bureaucratic rules work for themselves and not for the bureaucracy. They can adapt faster than bureaucracies can. So the bureaucracies must try to design their rules to minimize their losses in the bad cases, more than to maximize their gains in the good cases.

Also, bureaucracies must try to prevent others from changing the context.

I doubt this is particularly a slow gradual process from the time a bureaucracy is designed to fulfil idealistic purposes, to a later time when the bureaucracy has changed to maintain its own survival. It starts that way at day one, unless it's somehow created to be suicidal. It may develop an institutional memory of methods that have helped it survive threats in the past.

10:36 AM  

Post a Comment

<< Home