Skip to content

Second-Order Cybernetics

🌀 The Cybernetics of Observing Systems

There is a sentence that split a field in half. Heinz von Foerster said it, and cybernetics was never the same:

"Objectivity is the delusion that observations could be made without an observer."

The engineers were annoyed. The biologists leaned in. The philosophers looked like they had been waiting decades for a man in a lab coat to say it out loud.

Here is what he was pointing at — and why it changes how you hold every idea you have.

First-Order vs. Second-Order

Classical cybernetics — Wiener, Ashby, the whole mid-century project — studied observed systems. Feedback loops, control, regulation, homeostasis. A thermostat. A guidance system. A machine you could stand outside of, describe, and predict.

Von Foerster drew the line like this:

First-order cybernetics is the cybernetics of observed systems. Second-order cybernetics is the cybernetics of observing systems.

The difference is not a technical upgrade. It is a change in where you are standing.

In the first order, the observer stands outside the picture — and is politely invisible. In the second order, the observer is acknowledged as part of the system being described, because there is no description without a describer.

He put it better than anyone has since:

"A brain is required to write a theory of a brain. From this follows that a theory of the brain, that has any aspirations for completeness, has to account for the writing of this theory. And even more fascinating, the writer of this theory has to account for her or himself."

That is the whole idea. Once seen, it cannot be unseen.

The Loop You Are Standing In

Look at what this does to a mind.

You are not your thoughts. You are not your reactions. You are the one who sees them — and that awareness is not a spectator's seat. It is a participant's position.

Because the moment you begin to observe your own thinking, the thinking changes. Attention increases resolution. Resolution increases salience. Salience recruits more attention. The observation is not neutral — it is an intervention. You cannot study your own mind without altering the thing you are studying.

This is not a problem to solve. It is the shape of the instrument.

And it means the old dream — the detached objective view, the view from nowhere — was never on offer. Not for the world, and certainly not for yourself. What is on offer is better: an honest, situated, participant's view. Cybernetics that takes circularity seriously.

Conversation Is the Mechanism

Gordon Pask took this and ran with it, in a body of work called Conversation Theory (1975–76).

Pask's claim was radical for its time: cognition is not something that happens inside a head and is then transmitted outward. Cognition is a conversation — an ongoing exchange in which two participants, each an inherently separate and asynchronous system, work to converge on a shared understanding of a topic.

Notice what a concept is in his frame. Not a stored representation. Not a symbol filed away. A concept is an ongoing process of feedback that maintains a set of relations. It lives in the loop, not in the cabinet.

Two consequences worth carrying:

  • Teachback. Pask held that the critical act of learning is teaching another what you have learned. Understanding is demonstrated by reconstruction — not by recognition.
  • Serialists and holists. Pask found learners split into those who move through a structure step by step, and those who reach first for the higher-order relations between things. The holist does not skip the steps. They arrive having already grasped the shape.

Pask's own summary of the shift is a list worth memorizing:

from information to coupling — from order-from-order to order-from-noise — from transmission of data to conversation — from external observation to participant observation.

Why This Belongs in a Book About Mind

Here is the practical edge.

Most approaches to improving a life are first-order. They hand you a list. The list observes your commitments. It stands outside you and enumerates.

A second-order system does something else — it observes the observer. It does not merely hold what you decided; it holds where each decision currently is, and it changes state when the world answers back. It has a position on your situation relative to your own intentions.

That is the difference between a goal list and a system:

A list can enumerate what you want. It cannot tell you where any of it is. A system carries state — and state is what lets a commitment move.

This is what David Allen's Getting Things Done is actually doing, underneath the productivity language. "Your mind is for having ideas, not holding them" is a second-order move: it separates the part of you that generates from the part of you that tracks, so the generator stops being consumed by the tracker.

Read: Why I Love GTD

And the Weekly Review — the step everyone skips — is the purest second-order gesture in the entire method. It is the observation of your own observation. You look at the system that holds your commitments, and you ask whether the system is still true.

The Ethic That Falls Out

Second-order cybernetics has an ethical core, and it is elegantly short. Von Foerster's imperative:

"Act always so as to increase the number of choices."

That is it. Not be good. Not maximize utility. Increase the space of the possible — for yourself, and for everyone else inside the system you are part of.

Which is, when you sit with it, among the cleanest definitions of sovereignty anyone has written down. Freedom is not the absence of structure. It is the number of doors still open to you — and every act either widens that set or narrows it.

Dive In

  • Heinz von Foerster, Cybernetics of Cybernetics (1974) and Observing Systems (1981) — the primary sources.
  • Heinz von Foerster, "Ethics and Second-Order Cybernetics" (1992) — where the imperative lives.
  • Gordon Pask, Conversation, Cognition and Learning (1975) and Conversation Theory: Applications in Education and Epistemology (1976).
  • Humberto Maturana & Francisco Varela, Autopoiesis and Cognition (1980) and The Tree of Knowledge (1987) — "everything said is said by an observer."
  • Ernst von Glasersfeld, Radical Constructivism (1995) — the epistemology that grows alongside.
  • Ranulph Glanville, "Second Order Cybernetics" (2002) — the best short introduction.
  • Bernard Scott, "Second-Order Cybernetics: An Historical Introduction," Kybernetes (2004).
  • Steve Battle, "Conversation Theory for Design Agents," RSD12, Article 128 (2023) — Pask applied to large language models, and worth reading if you are building with agents. Open access via the OCAD repository.
  • Thomas Manning, "Some Design Considerations for Concept-Forming and Concept-Sharing in Discursive Agents," RSD13 (2025) — the sharpest current argument that Pask's conversation theory belongs in multi-agent systems.