Ashby's Law of Requisite Variety

In our previous lesson, we established a precise, quantifiable meaning for complexity through W. Ross Ashby's concept of variety. We saw that it represents the number of distinguishable states a system can possess, a measure that is fundamentally tied to information theory and the perspective of an observer. Now, we will take the next crucial step. Knowing how to measure complexity is one thing; knowing how to manage it is another.

This lesson addresses that challenge directly by introducing cybernetics' most fundamental principle of regulation. Our learning outcome is to state and explain Ashby's Law of Requisite Variety. This law provides the essential logic for how control is possible in any system, from a simple mechanical device to the complex political and economic systems that inspire your interest in this field.

1. Stating the Law: A Core Principle of Cybernetics

The Law of Requisite Variety is elegant in its simplicity and profound in its implications. Let's begin by hearing it stated by two key figures. First, a concise summary, and then a more passionate declaration from Stafford Beer, a central character in our story.

Management Cybernetics: The Law of Requisite Variety

This short clip provides a direct and clear statement of Ashby's Law.

Watch the opening of this video, from the beginning until dependent on experiment. Listen for the memorable phrasing of the law.

As the video states, the law's central tenet is: "Only variety absorbs variety." It's presented not as a scientific finding subject to experimental proof, but as a self-evident truth, almost like a logical or mathematical axiom.

Now, let's watch Stafford Beer—the architect of Project CyberSyn—introduce this same law. His presentation captures the weight and significance he and other cyberneticians attribute to it.

Ashby's Law of Requisite Variety and Autonomy / Stafford Beer // Javier Livas

In this excerpt from a lecture, Stafford Beer introduces Ashby's Law as the "major law in cybernetics."

Please watch from it will exert itself. Notice how Beer frames the law not as a mere suggestion but as an inexorable principle, akin to a law of nature.

Beer's emphasis is critical. This is not a "guideline" or a "best practice." Ashby's Law is presented as a fundamental constraint on the universe of control. If a system fails to abide by it, the law still "exerts itself"—not by being broken, but by producing failure, chaos, or gridlock.

2. Unpacking the Law: Regulation as a Battle of Variety

To understand what "Only variety can absorb variety" means, let's model a general regulatory situation. Imagine three components:

The System: The thing we want to keep stable. Its essential variables (e.g., body temperature, a company's profit, a nation's food supply) need to be kept within certain limits.
The Disturbances (D): Unpredictable events or forces from the environment that threaten to push the system's essential variables outside their acceptable limits. The variety of these disturbances, $V_{D}$ , is the amount of complexity the system has to face.
The Regulator (R): The mechanism responsible for countering the disturbances. The variety of the regulator, $V_{R}$ , is the number of distinct responses or actions it can take.

The goal of regulation is to ensure that for every possible disturbance that the environment throws at the system, the regulator has an appropriate response to keep the system's outcome stable.

Ashby's Law states that for regulation to succeed, the variety of the regulator must be at least as great as the variety of the disturbances that need to be managed.

$V_{R} \geq V_{D}$

If the regulator has less variety than the disturbances, there will inevitably be a disturbance for which the regulator has no answer. At that point, control is lost. The following image provides a simple but powerful visualization of this mismatch.

This diagram illustrates the core of Ashby's Law. On the top, the variety of responses matches the variety of problems, allowing each problem to be dealt with. On the bottom, there is insufficient (or incorrect) variety in the responses, leaving some problems unaddressed. This leads to a failure of regulation.

Stafford Beer's examples in the video you watched illustrate this perfectly:

A traffic policeman (regulator) who can't see all the traffic (disturbances) lacks the requisite variety to control the intersection, resulting in gridlock.
A finance minister (regulator) using only one or two tools like interest rates has insufficient variety to manage the immense variety of a national economy (disturbances), leading to unintended consequences and economic instability.

3. A Formal View: The Game of Regulation

We can express the law more formally by considering the variety of the Outcomes (O). The variety of outcomes, $V_{O}$ , represents the spread of possible states the system might end up in after the regulator has acted. Our goal is to minimize this outcome variety, ideally keeping it at 1 (the single, desired state).

Let's turn to a paper by Raul Espejo, a collaborator of Stafford Beer's who was deeply involved in the Cybersyn project. He explains the law using a "black box" model.

Cybersyn, big data, variety engineering and governance - PMC

This article discusses Cybersyn and variety engineering. We will focus on the section that formally explains the Law of Requisite Variety using a diagram.

First, find the section titled "Variety engineering..." and read the first few paragraphs to understand the context of the "black box" description. Then, focus your attention on Fig. 4 and the paragraph that explains it. Start with the paragraph that begins right after the figure. Also, take a look at Footnote 2 for a concise definition of variety.

The article presents Ashby's law as an inequality concerning the variety of outcomes:

$V_{O} \geq \frac{V _{D}}{V _{R}}$

Here, $V_{D}$ is the variety of disturbances, $V_{R}$ is the variety of the regulator's responses, and $V_{O}$ is the variety of the final outcomes.

This equation is a powerful summary of the regulatory challenge. To make the system's behavior more predictable (i.e., to reduce $V_{O}$ ), you have two choices:

Decrease the variety of the disturbances ( $V_{D}$ ).
Increase the variety of the regulator's responses ( $V_{R}$ ).

Since $V_{R}$ is in the denominator, increasing the regulator's variety has a powerful effect, reducing the variety of the outcome. This is the mathematical expression of "Only variety can absorb variety."

A general model of regulation. The "system regulated" is buffeted by disturbances, creating variety. The "regulator" must absorb this variety. To do so effectively, it uses strategies of attenuation (to reduce the incoming variety from the system) and amplification (to increase the variety of its own regulatory actions).

4. Requisite Variety and Information Theory

Your background in computer science and our previous discussions on Shannon's information theory provide a unique lens through which to view Ashby's Law. The law is not just an analogy to information theory; it is a direct generalization of it.

Let's read a short passage from a paper by Klaus Krippendorff, who was a student of Ashby's.

[PDF] Ross Ashby's information theory - University of Pennsylvania

This paper directly connects Ashby's work to Shannon's, highlighting how one is a special case of the other.

In the PDF, find the paragraph that begins "Part of Ashby's". Read from there to the end of the next paragraph, ending at the comparison to Shannon's 10th Theorem.

As the text explains, there is a direct parallel between Ashby's Law and Shannon's 10th Theorem, the noisy-channel coding theorem.

Shannon's Theorem: Communication through a noisy channel can be made error-free if you add a correction channel whose capacity (measured in bits/sec) is equal to or greater than the entropy rate of the noise source (also in bits/sec).
Ashby's Law: A system can be regulated against disturbances if the regulator has a variety (measured in bits) that is equal to or greater than the variety of the disturbances (also in bits).

The "disturbances" in Ashby's Law are equivalent to the "noise" in Shannon's model. The "regulator" is the "correction channel." The goal of maintaining a stable system is equivalent to the goal of receiving a clean signal. Ashby's Law generalizes this principle from communication channels to any process of regulation and control. This connection elevates the law from a clever heuristic to a principle with deep roots in the mathematics of information.

Conclusion

In this lesson, we have moved from measuring complexity to understanding the fundamental law that governs its management. You have now encountered one of the most powerful and broadly applicable ideas in all of cybernetics.

Let's summarize the key takeaways:

Ashby's Law of Requisite Variety states that "only variety can absorb variety."
For a regulator to successfully control a system, its own variety of actions ( $V_{R}$ ) must be at least as great as the variety of the environmental disturbances ( $V_{D}$ ) it has to face.
This relationship can be formally expressed as $V_{O} \geq V_{D} / V_{R}$ , where $V_{O}$ is the variety of the outcomes. To minimize $V_{O}$ , one must have a sufficiently large $V_{R}$ .
The law is a direct generalization of Shannon's noisy-channel coding theorem, treating disturbances as noise and the regulator as a correction channel.

You have now stated and explained the law. In our next lesson, we will build on this foundation and move from theory to practice, where you will apply the Law of Requisite Variety to analyze a regulatory problem. We will take a concrete scenario and use the law as a diagnostic tool to understand why it might be failing and what would be required to make it work.

Recommended resources

1. Stating the Law: A Core Principle of Cybernetics

2. Unpacking the Law: Regulation as a Battle of Variety

3. A Formal View: The Game of Regulation

4. Requisite Variety and Information Theory

Conclusion