Moore's Laws of Bureaucracy

Preliminary Observations

(1) A bureaucracy is a servo-mechanism. It computes an error signal by processing the current state of the external system against the desired state. It then generates corrective commands to minimize the future error signal, and hence to drive the state of the system in the desired direction.

(2) The state of the system is measured in many variables, by samples from sensors. The state is modified by the actions of effectors.

(3) In a bureaucracy, there are a large number of sensors (low level bureaucrats) and a large number of effectors (also low level bureaucrats). Goals are provided by an external authority.

(4) In order to acquire and filter information from the sensors, process that information against the goals, and propagate and refine the correction (difference) signal to the effectors, a management hierarchy is required.

(5) The larger the bureaucracy, the deeper the hierarchy. The depth is logarithmically proportional to the sum of the number of sensors and the number of effectors. The top of the hierarchy is a single bureaucrat with a title like "Director," "Secretary," "Administrator" or "Commander." The middle layers of the hierarchy are populated by mid-level bureaucrats.

* * * * * *   The Laws Themselves   * * * * * *

Failure to Achive Goals

(1a) The probability of the decision element at the top correctly measuring the system state decreases exponentially with the depth of the hierarchy. Each level adds noise to the information as it passes through. Thus the measurement signal is very noisy in a large bureaucracy.

(1b) The same logic applies to the transmission of the correction signal back downwards through the system.

(1c) Thus the likelihood of the system correctly responding to the inputs and goals goes down at a high rate (approximately K to the 2N power where (1-K) is the error rate of one individual and N is the depth of the bureaucracy).

(1d) Thus LARGE BUREAUCRACIES CANNOT POSSIBLY ACHIEVE THEIR GOALS!

Thrashing

(2a) Furthermore, even if the signal were to be transmitted correctly, it would be delayed by the depth of the hierarchy.

(2b)Servo theory tells us that corrective feedback, applied with a significant time lag, causes oscillations of ever increasing magnitude.

(2c) Thus LARGE BUREAUCRACIES WILL THRASH WILDLY ABOUT, CAUSING MUCH COST, PAIN AND DESTRUCTION!

Evil

(3a) In addition, the probability of a person in society being highly amoral (a sociopath) is about 5%. A sociopath will act totally selfishly in almost all circumstances.

(3b) Thus the probability that a sociopath will not be in the feedback loop is at most (.95^2N)).

(3c) But in reality, the power wielded by a bureaucracy attracts sociopaths, so their density will be much greater in a bureaucracy than in the population at large.

(3d) Thus, the probability that a sociopath will not be in the feedback loop is around (.8^2N))

(3e) So a typical bureaucracy with 12 levels has a 99.5 percent chance that one or more sociopaths are in each control path. The actual percentage is somewhat lower because the sensor and effector information paths often overlap.

(3f) Thus LARGE BUREAUCRACIES ARE EVIL!

Heartless

(4a)In order to try to thwart the noise and sociopaths in the system, those at the top will create rules (policies) that apply directly to the sensors and effectors. Rules are also needed by government bureaucracies in order to provide an appearance of fairness or even-handedness.

(4b) This, of course, is doomed to failure since the information on which these policies are based was created by the noisy system.

(4c) But since bureaucrats must believe in their own wisdom and power, they will persist in this silliness.

(4c) So the effectors, who actually deal with the public, will be totally constrained in their behavior by rigid rules.

(4d) Thus LARGE BUREAUCRACIES HAVE NO HEART!

Perverse

(5a) By the way, bureaucracies are created to produce long term results.

(5b) But, bureaucrats have their own goals of rising rapidly within the organization.

(5c) So bureaucrats will demand to be measured on their short term performance.

(5d) So the bureaucracy will focus on these short term goals.

(5e) Thus LARGE BUREAUCRACIES ARE PERVERSE! [see Note 3]

Immortal

(6a) Of course, Bureaucracies never achieve their goals (see #1 above).

(6b) Thus, they must keep striving.

(6c) So they keep acquiring new bureaucrats as the old ones wear out.

(6d) Thus LARGE BUREAUCRACIES ARE IMMORTAL!

Grow Without Bounds

(7a) Finally, the power of a bureaucrat is determined by how many people report to her.

(7b) Thus she is motivated to increase her staff by increasing the reach of her portion of the bureaucracy.

(7c) Thus LARGE BUREAUCRACIES WILL GROW WITHOUT BOUND! [See Notes 1,2]

Notes and Links

[1] As was pointed out by Clark Jones (jones@azterra.San-Jose.ate.slb.com):
"The amount of matter in the universe is finite (and the sum of matter [times c**2] and energy at least appears to be constant). Every bureaucrat has some finite mass. This will, eventually, serve to limit the number of bureaucrats, and thus, in turn, impose an upper limit on the size of bureaucracies."

[2] The trend of governments to attempt to regulate the Internet constitutes a recent example of this law. The government is itself an enormous bureaucracy, and as such seeks power. In the case of he Internet, however, the outcome will be more consistent with conclusion 1d: LARGE BUREAUCRACIES WILL FAIL TO CONQUER THE INTERNET!

[3] Law of Bureaucracies apply outside of governmental institutions. For example, the administration of private pension funds is certainly affected by the Fifth Law of Bureaucracy: the individuals are administering funds whose owners have long term goals, but whose fiduciaries have short term goals - their annual bonuses and their reputations. Likewise, publicly regulated monopolies tend to incorporate the worst combination of governmental and private bureaucracies. See, for example, QWest, Inc.

John Moore

last update 10/15/2000