A Cybernetics Reading List, Written Before I Read It
An agent can check a quote for me. It cannot read for me. So before I open Ashby, Wiener and the 1943 paper, here is the list, and what I expect to find, written down first.
An agent opened Ashby's book for me this week. It found the law of requisite variety, quoted it word for word, checked the section number, and marked the row in my ledger as verified. Then it drafted a post correcting something I wrote in June. One sentence in that draft said: "I went to the chapter and read it through."
Except I hadn't read that chapter. The quote was right and the draft was pretty good, and it was about to go out with my name on it saying I'd done reading I hadn't done. And I really hate posting about things I haven't read.
Checked is not read
The ledger is a list of every claim in a course on cybernetics I'm writing for myself. Each claim gets a type (history, a formal result, someone's concept, my own reading) and a note on whether the source has been opened. That was the rule, and an agent can satisfy it in minutes.
So now there's a second column, and it asks a simpler question: did I actually read it? Me, not the agent. Nothing about the field goes on this site until both are true.
| The ledger tonight | |
|---|---|
| Claims in the course | 78 |
| Checked against the source, by an agent | 7 |
| Read by me | 0 |
Zero! Everyone starts somewhere.
I already run two readers on my handwriting and only look where they disagree, and this is the same idea turned around: the agent is the first reader, and the second one has to be me.
The list
What I'm reading, in order, and the question I'm carrying into each one.
| Read | Length | The question I bring | |
|---|---|---|---|
| 1 | W. Ross Ashby, An Introduction to Cybernetics (1956), chapter 11, sections 11/1 to 11/7 | about 6 pages | What has to be as varied as what? |
| 2 | Arturo Rosenblueth, Norbert Wiener and Julian Bigelow, "Behavior, Purpose and Teleology" (1943) | 7 pages | What do they mean by purposeful, and where does feedback come in? |
| 3 | Roger Conant and W. Ross Ashby, "Every Good Regulator of a System Must Be a Model of That System" (1970) | 9 pages | What counts as a model? |
| 4 | Norbert Wiener, Cybernetics (1948), the introduction | 40 pages | Why that name? |
| 5 | Paul Pangaro, "Questions for Conversation Theory" (2017), and Hugh Dubberly and Paul Pangaro, "Cybernetics and Design: Conversations for Action" (2015) | 10 pages each | What makes a conversation a conversation? |
| 6 | Stafford Beer, Brain of the Firm (1972) | a book | Is "only variety can absorb variety" his? |
After that the shelf keeps going: von Foerster, Mead, Bateson, Maturana and Varela, and Pask himself. Most of it is free and some of it is a library card away.
What I expect, written down first
Before opening any of them I wrote down what I think they say, fast and in my own words, no looking anything up. My guesses aren't clean, though. An agent gave me a short summary of each text first, so a bit of that is surely in here, and that's part of what I'm watching too.
Ashby's law.
Something hard and mathematical, resilient to breaking down under argument. Very interested in this and how it relates to technology.
Purposeful, in the 1943 paper.
Very excited about behavior, purpose and teleology (feedback, I think?) and learning how this was thought about in these times, before we have it in our AI models in real-world loops that can be crafted and molded and shaped (with intelligence to boot!). So purposeful here means with intent toward a goal, is my guess or expectation.
A model, in Conant and Ashby.
I'm thinking they mean an understanding of what is happening and how things are connected, and that if you are regulating some process you have to know all the ins and outs, or the levers to switch things on and off.
The name.
Cybernetics gets deeper at the root of the role of technology and where things are going with what we are building and heading toward. Steersman, from the Greek word kyvernetes, makes me think of the control plane in Kubernetes, and of course the links are not lost on me.
That last one isn't a stretch, by the way. Kubernetes takes its name from the same Greek word, helmsman or pilot, so the whole thing really is named for the one at the tiller.
Why write them down
So the next post can measure me. When I've read them I'll set each guess beside what the text actually says and see how far off I was, and the being off is the interesting part, really. That distance is the measurement. Science has a name for writing your prediction down before the experiment, pre-registration, and it's there so you can't decide afterward that you knew it all along.
And I'm fairly sure this field has its own name for watching yourself learn while you're learning, for putting the observer inside the thing being observed. Second-order cybernetics, I think? I haven't read von Foerster or Mead yet, so I'll hold off on the word until I have.
Read along
Come with me if you like. Most of these are free, from the authors, the publishers, or the old cybernetics archives that have kept them online for years.
| Read | Where to find it |
|---|---|
| Ashby, An Introduction to Cybernetics, chapter 11 | Principia Cybernetica, the whole book, free |
| Rosenblueth, Wiener and Bigelow (1943) | Philosophy of Science, through a library or JSTOR |
| Conant and Ashby (1970) | Principia Cybernetica, free, or the journal |
| Wiener, Cybernetics, the introduction | MIT Press open access, free |
| Pangaro, and Dubberly and Pangaro | Pangaro's own site, free: the conversation theory paper and its talk, and the design paper |
| Beer, Brain of the Firm | a bookshop or a library |
One thing before you open them: write down what you expect, a sentence or two each, fast and without looking anything up. Then we can both see how far we moved.
Why now
I can hardly wait to start! This stopped being a history lesson the moment I saw it as something I can run, with real loops and real measurements in the work I already do (systems design, DevOps, architecture modeling) and in how I learn. These papers were written before the loops could think back. Now they can, and that's exciting as hell.
So I'm going to catch my reading up to my borrowing, one text at a time, guesses first, with the ledger keeping score.
Let's go!