or why should I bother?
(With a hat-tip to Neil Postman, whose book Amusing Ourselves to Death described how technology - in his case broadcast media - eroded our capacity for critical thought, I write this about the new technologies of social media and Generative AI).
The Spark: Cuts and Constructivism in Cybernetics
I was inspired to write this by Harish’s terrific post, An Introduction to Cybernetic Constructivism. It struck a chord with my ongoing interest understanding Biological Constraints or the Cybernetic Cut.
The “TL;DR” of my investigations is whether biology is “just” physics or something fundamentally more. While I’ve been using my “QSet” approach - an AI-augmented human memory - to synthesise leading concepts from Biology to Complexity Science, Harish approaches a similar structural argument without any AI from thinkers in Philosophy.
Crossing The Divides:
| My QSet (Biological/Relativism) | Harish’s Sources (Philosophical) |
|---|---|
| Waddington, Rosen, Pattee, Noble | Heidegger, von Foerster |
| Feynman (Physics) | Maturana, Varela |
| Juarrero, Kaufmann (Complexity) | Merleau-Ponty, Bachelard |
| Lissack, Luhmann (Management) | Ludwig Wittgenstein |
While my focus has been the “Cybernetic Cut,” Harish explores “Cybernetic Constructivism.” The structure of the arguments are similar and the challenge he poses is profound: Are these constraints real, or are they merely artifacts of language and consciousness?
Ontology vs. Epistemology: Shifting the Fulcrum
Harish’s writing forced me to move my own stance. I used to view constraints as objective “realities” (Ontology). Now, I’m forced to also consider if they are primarily about how we perceive and describe them (Epistemology).
-
The Old View: Constraints are real; our knowledge of them is secondary.
-
The New View: Constraints, like systems, are dependent on the observer.
Taking the old world stance the onus is to explain away the epistemic ‘side’ of the argument, however if I adopt the new world stance the primary point of view is epistemic with ontological considerations then subsidiary.
Adhering to just one “pole” - asserting complexity is purely real or alternatively that complexity is purely in your head - is a fool’s errand. As Herb Simon brilliantly illustrated with his ant trail: the path is simple to the ant, but complex to the human observer. Any “truth” lives in the juxtaposition of both.
The Bayesian Warning and “Newspeak”
To add more I’ve also been intrigued by the post The Mind-Bending World of Bayesian Probability, which led me to the eminent scientist E.T. Jaynes. In his paper, What’s Wrong with Bayesian Methods?, Jaynes offers a warning that underpins the importance of Harish’s writing even if you disagree with his premise:
“Our communication problems arise in large part from the difficulty that orthodox terminology is not adapted to expressing Bayesian ideas (in this respect it reminds one of the Orwellian NEWSPEAK, a language within whose vocabulary and grammar it was not possible to express dissenting views).”
This is the crux of the matter. Whether Harish is "right" or "wrong" is less important than ensuring our methods and approaches don't exclude certain possibilities of explanation before we even begin to think.
To Post or Not to Post?
After synthesising all of this into a powerful prompt for my QSet:
Can you include ET Jaynes' observation about Orwell's Newspeak that we need to ensure the medium of description does not exclude certain possibilities of causality
the response was illuminating and thought provoking response. With my finger on the publish button I had a moment of hesitation…
Should I even post this?
In the economy of “likes” and “retweets,” a breakthrough observation can often be met with the deafening tumbleweed of indifference. It can be soul-destroying to offer up a piece of your ego only for it to vanish into the digital void. There’s an old saying that a real scientist would rather share their toothbrush with a stranger than their idea.
So, in a way, I’m writing this not as a bid for your attention, but as an anchor for my own memory. If you've read this far, consider yourself an accidental passenger in my internal dialogue.
But as a trailer to the next step of my investigations …
Would you like me to help you refine that QSet prompt further or perhaps explore how Jaynes’ critique of “Newspeak” applies specifically to modern Large Language Models?
Acknowledgments
The author thanks Google’s Gemini, whose suggested revisions have greatly improved the paper.