Cybernetics: a reading path

The word comes from the Greek for the one who steers. It is the study of how anything stays on course by watching what it does and correcting. A thermostat, a cell, a team, a person, a cluster of machines.

I am reading its founding texts in order, with my guesses written down before I open each one, and building what I need to follow them. This page is the way in.

On this page 10 sections

Start gentle

Gentler On-Ramps to Cybernetics The founding papers were too steep and so were my own primers. So here are the gentler ways in, stories and a game to test the water, an hour of free reading, then a shelf of books.

Or start by listening

Some of the people on the map can still be heard. I am collecting their talks as I find them, in the order I would walk them. Each shelf begins here.

The whole watching path, with a box to tick for each talk.

Or start by moving something

Some of this I understand best when I can move it. Three pages are made for that.

  • The modelsThe math under the papers as pictures you drag, each formula worked out from its picture, from where a matrix sends its arrows to Ashby's requisite variety. With music, if you want it.
  • The mapEvery idea on the family tree against the primers, talks, books and math that cover it, and where it is still thin.
  • Ideas and peopleGo from an idea to the people named on it, and from a person to who they worked with.

The whole family

Where it came from, which older disciplines met in it, and the shelves it turned into, drawn as one map you can tap through.

The Cybernetics Family Tree I wondered where cybernetics would sit on my O'Reilly shelf. It has no shelf of its own. It sits under all of them. A checked family tree, from the governors to Kubernetes, with your shelf on it.

Another map of the same family, drawn from inside it, is Hugh Dubberly and Paul Pangaro's social graph of cybernetics (2015), made for the Walker Art Center's Hippie Modernism catalogue. It follows the lines out from Wiener, von Foerster, Ashby, Pask and Beer to computing, design and the counterculture. There is an interactive version to click through, and the essay that goes with it. On this site it is joined to the family tree as the people of cybernetics, who worked with whom, and the ideas each of them is on.

How the path came about

It started as a reading list. The papers were steep, so I built a map under them, then a gentler slope under the map. Each step is its own post.

The reading came first, then the talks. How the talks were found and chosen is open too, every search, everything found, and what I kept.

The primers

One for each work, in two parts. A companion for when the words are new, then the primer, and a sheet to print. Before all of them there is the lesson before the lesson. Where a talk opens the work, it is there too.

Part one Ashby to Dubberly and Pangaro

How these were made, and why, is in A Map Before the Reading

  1. Requisite Variety W. Ross Ashby, 1956 Companion 7 min Primer 14 min Sheet SetsLogs and roots
  2. Behavior, Purpose and Teleology Rosenblueth, Wiener and Bigelow, 1943 Companion 6 min Primer 15 min Sheet Words
  3. Every Good Regulator of a System Must Be a Model of That System Conant and Ashby, 1970 Companion 6 min Primer 17 min Sheet SetsProbability
  4. Cybernetics, the introduction Norbert Wiener, 1948 Companion 6 min Primer 13 min Sheet Words
  5. Questions for Conversation Theory Paul Pangaro, 2017 Companion 6 min Primer 12 min Sheet Talk21 min Words
  6. Cybernetics and Design Dubberly and Pangaro, 2015 Companion 6 min Primer 11 min Sheet Words

Part two Beer to Boyd

How these were made, and why, is in The Rest of the Map

  1. The Viable System Model Stafford Beer, 1984 Companion 8 min Primer 17 min Sheet SetsWords
  2. On Constructing a Reality Heinz von Foerster, 1973 Companion 6 min Primer 11 min Sheet Talk25 min Words
  3. Cybernetics of Cybernetics Heinz von Foerster, 1979 Companion 6 min Primer 11 min Sheet Words
  4. Ethics and Second-Order Cybernetics Heinz von Foerster, 1991 Companion 7 min Primer 16 min Sheet Words
  5. Cybernetics of Cybernetics Margaret Mead, 1968 Companion 6 min Primer 10 min Sheet Words
  6. Form, Substance and Difference Gregory Bateson, 1970 Companion 8 min Primer 15 min Sheet Words
  7. Autopoiesis Varela, Maturana and Uribe, 1974 Companion 8 min Primer 15 min Sheet Words
  8. Developments in Conversation Theory Gordon Pask, 1980 Companion 7 min Primer 16 min Sheet Words
  9. What Is Conversation? Dubberly and Pangaro, 2009 Companion 6 min Primer 14 min Sheet Words
  10. Restricted Complexity, General Complexity Edgar Morin, 2006 Companion 7 min Primer 17 min Sheet Words
  11. Destruction and Creation John Boyd, 1976 Companion 7 min Primer 14 min Sheet

The math under the papers

Some of these papers read easily and some turn into a wall, and which one depends on the language a paper is written in. Most of the second-order work is written in words. The wall is in a handful of papers written in calculus, probability and machines as tables.

So this is the layer under the primers. Each language, a line of it in its own notation, the ideas on the map that speak it, and where I am learning it, free. I am taking them in my own order. Brush up first, then learn new, then revisit what I have long forgotten.

Some of it I needed to see move before it made sense, so it became the models, pictures you drag by hand, each formula worked out from its picture, from where a matrix sends its arrows to Ashby's requisite variety.

Each step opens to what it says, who speaks it, and where to learn it.

  1. Brush up

    Sets, logic and machines as tables

    T: a → b, b → c, c → c · V(O) ≥ V(D) − V(R)

    Describe a machine as a list of states and the rule that moves each state to the next, with no physics in it. Ashby builds his whole book on that, up to requisite variety: only variety in the regulator can absorb variety in the disturbances. McCulloch and Pitts do the same with neurons as on-off switches.

    Spoken by Ashby, requisite variety; Every Good Regulator; A Logical Calculus of neurons; Von Neumann's automata; The name "artificial intelligence"; Management cybernetics, the VSM

  2. Brush up

    Programming: run the loop

    while True: error = goal − reading; act(gain * error)

    Most of these machines can be built in a few lines and watched. A thermostat, a homeostat, a tortoise that seeks the light, a stock and flow model of a firm. I read code and I practice it at different levels, like a kata, and it still pushes me. When an equation will not open, writing the same machine and watching it run can be the way in.

    Spoken by Ratio Club, Grey Walter's tortoises; System dynamics; The Limits to Growth; Robotics and control; Control loops in software; Complexity science

  3. Learn new

    Logarithms, polynomials and complex numbers: the floor under calculus

    log₂ 8 = 3 · s³ + a s² + b s + c settles when a, b, c > 0 and ab > c

    A logarithm asks how many times you multiplied. It turns huge ranges into a line you can read, which is why information is counted in bits, why Bode drew amplifiers on log paper, and why the power laws of complexity science come out straight. Polynomials and their roots, complex ones included, are the other half of the floor: Maxwell turns his question about whether a governor settles into a question about the roots of one polynomial, and the line above is his answer for the third degree, written the way it is taught today.

    Spoken by Maxwell, On Governors; A Mathematical Theory of Communication; Information theory; Entropy, coding, compression; Feedback amplifiers; Complexity science; Ashby, requisite variety

  4. Learn new

    Calculus: rates of change

    dx/dt = lim h→0 [x(t + h) − x(t)] / h

    How fast something is changing right now, and how the changes add up over time. Every regulator in the control line is written this way, and so is the gradient a neural network follows downhill.

    Spoken by Maxwell, On Governors; Wiener, Cybernetics; Modern control theory; Neural nets, deep learning

  5. Revisit

    Differential equations: will it settle?

    dx/dt = −k (x − goal) → x settles at the goal when k > 0

    Write how a thing changes as a rule that depends on where it is now, then ask whether a small push dies away or grows. Maxwell asked exactly that of a flyball governor, and system dynamics draws the same equations as tanks and pipes.

    Spoken by Maxwell, On Governors; Feedback amplifiers; Servomechanisms at MIT; Modern control theory; System dynamics; The Limits to Growth

  6. Learn new

    Probability, information and signals

    H = −Σ p log₂ p

    Treat a message, or a moving target, as a random process and predict it as well as the statistics allow. This is the language of Shannon, of the hard middle of Wiener's book, of the good regulator theorem, and of the free energy principle.

    Spoken by A Mathematical Theory of Communication; Wiener, Cybernetics; Information theory; Every Good Regulator; Entropy, coding, compression; Reinforcement learning; The free energy principle

  7. Learn new

    Linear algebra: states as vectors

    x(t + 1) = A x(t)

    A whole system's state as one list of numbers, and its rule of change as one matrix. Modern control is written this way, and so is every layer of a neural network.

    Spoken by Modern control theory; The perceptron; Neural nets, deep learning

  8. Learn new

    Neural networks: weights that learn

    y = σ(W x + b) · W ← W − η ∇L

    Where the 1943 switch ends up once its weights can learn: a layer is a matrix, learning is nudging it downhill on an error surface, from the perceptron to the models I work beside every day.

    Spoken by A Logical Calculus of neurons; The perceptron; Neural nets, deep learning; Reinforcement learning

  9. No equations

    Words and diagrams

    loops drawn, not solved

    Most of the second-order work keeps feedback and the observer but argues in prose and pictures. These papers read hard for other reasons: when one feels slippery, the missing piece is a vocabulary, and the companions and primers are built for exactly that.

    Spoken by Behavior, Purpose and Teleology; The Macy conferences; General systems theory; Bateson: the double bind, an ecology of mind; Mead, Cybernetics of Cybernetics; Von Foerster, the observer inside; Autopoiesis; Radical constructivism; The Embodied Mind; Pask's conversation theory; Whole Earth, counterculture; Luhmann's social systems; Morin's complex thought; Man-Computer Symbiosis; Family and systemic therapy; Cognitive science, extended mind

    • Book Donella Meadows, Thinking in Systems: A Primer

Put to work

  • The Regulator Was Fed IntentsI built the smallest loop that could be wrong, to watch a law from Ashby do work. The loop was wrong twice before it was evidence, the law turned out not to apply, and the real cluster already knew the fix. All of it measured, and explained from the ground up.
  • A Loop Is a Path That ClosesOne of my health checks stayed red for eleven weeks by reading its own failure. A loop is the special case of a trajectory, the path that happens to close, and a life does not.
  • The Hacker Ethic Meets CyberneticsA fourteen-year-old's robot bug at the MIT AI lab pushed a wallet across a grimy floor while the grad students upstairs wrote theses on why it was hard. The hacker ethic, rewritten for cybernetics, humility and AI as a partner.
  • The Right to ExploreThe hackers didn't understand the machine when they started. That was the point. So why is someone exploring with AI written off for not understanding yet? FORTRAN, the genetic fallacy and a loop.

Around it

  • Culture Is the Oldest Information TechnologyI once wrote that our minds run on the same hardware as every other animal and that something rewrote the software. I think the software is culture, the oldest information technology we have.
  • The Information by James Gleick: The Book That Sent Me Looking for MeaningJames Gleick's The Information is the clearest history we have of how the world learned to measure the signal. It is also the book that showed me, by drawing the boundary so cleanly, the one thing the science of information set aside on purpose, which is meaning. This review is the origin story of the work I have been doing since I closed it.
  • A Moment MeasuredMaturana wrote that living happens in zero time. Tolle says the present is all we have. Calculus measures an instant it can never reach. Three takes on the present moment, and a builder who wants something to come of it.
  • Eigen to MyselfA tired joke that held. The eigenself is what comes back pointing the same way. Nietzsche at his typewriter, his dice game, a mathematician from Zarathustra's landscape, a cursor on a black screen, and three experiments.

Everything else on loops, maps and legibility is under Complexity and systems, and in the Atlas, where this whole site is drawn as one graph.

The maps

The family tree and the reading map are two of a set. Each one says what it maps.

  • The AtlasThe site: every page and how it links, drawn live.
  • The cybernetics family treeA field: its people and ideas, what it grew from and the shelves it became.
  • A Map Before the ReadingA reading route: the canon in order with a primer for each work, drawn when the first paper stopped me.
  • The watching pathThe same route in talks: other people’s lectures and conversations.
  • How the watching path was madeThe gathering itself: every search, every series followed, everything found and what was kept.
  • The cybernetics mapA domain as one graph: every idea against the material that covers it, and what kind of fact each thing is.
  • The frameworksMy own frameworks, each one charting a single domain.
  • Domain cartographyNot a map but the method behind them: how a territory gets charted so others can follow.

More to look around