Why cybernetics invented the observer and yet remained alone — and why, today, a ‘superlearner’ is buying that same loneliness for 1.1 billion dollars.
"Only a party to a case can really judge, but, being a party, it cannot judge. Hence there is no possibility of judgement in the world but only the glimmer of a possibiliity."
— Franz Kafka, The Eight Octave Notebooks
20. July 2026
The Setup
In 1943, cybernetics found purpose inside the machine. In 1966, Hans Jonas denied it could live there. In 1974, Heinz von Foerster placed the observer inside the system — and declared the loop unsolvable. Thought has been standing there ever since. In 2026, David Silver raises $1.1 billion for a superlearner meant to discover all knowledge from its own experience, without human data. The critics cite Wiener, Jonas, Foerster — and repeat their move: they look for the property inside a single system. One system. One purpose. One observer. Always the singular. But the structure emerges in the plural.
On April 27, 2026, Ineffable Intelligence steps out of stealth. David Silver, the man behind AlphaZero, has raised $1.1 billion — at a valuation of $5.1 billion, before anything exists except a goal. The goal: a superlearner that discovers all knowledge from its own experience. No human data. The company website carries a sentence you read twice: if it succeeds, it would be a breakthrough of Darwin's magnitude — his law explained all Life, ours will explain and build all Intelligence.
The criticism came fast, and it came well-read. It cited eighty years of cybernetics. And it stopped exactly where cybernetics had stopped.
The Chain
In 1943, Arturo Rosenblueth, Norbert Wiener, and Julian Bigelow published “Behavior, Purpose, and Teleology.” Their thesis: Machines can act purposefully. Feedback is sufficient. Wiener’s anti-aircraft predictor calculates where an airplane will be—it pursues a goal that does not yet exist. The purpose has been incorporated into the machine.
In 1966, Hans Jonas counters this in “The Phenomenon of Life.” Automata pursue a purpose that humans have inscribed in them. True purpose requires a metabolic risk: the possibility of dying if it is not achieved. The purpose withdraws once more.
In 1974, Heinz von Foerster draws the conclusion from both: whoever observes is part of the system. Second-order cybernetics—the observation of observation. Objective description: impossible. The loop: epistemically unsolvable.
Three stages. Each brings us a step closer. Then: nothing more. Sixty years have passed since Jonas.
One system. One purpose. One observer.
Listen to the grammar of the questions. Can the machine have a purpose? Can the system observe itself? Can thesuperlearner validate his discoveries? Eighty years of debate, a single grammatical form: the singular.
This is monocausal thinking in its purest form. One property, one place, one cause. Even second-order cybernetics, which brings the observer into the picture, brings in exactly one. The loop closes around a single point—and is promptly declared unsolvable. Foerster was right. Within this grammar.
The failure does not lie in the answers. It lies in the singular form of the question. Cybernetics found the observer. It forgot to give him a counterpart.
What Emerges in the Plural
Gödel: no system can prove its own consistency — one system. Heisenberg: no observer measures without disturbing — one observer. Foerster: no observation without an observer — one. All singular sentences, all true, all thought to their end in the same place.
Paradoxical Interactions begin one step to the side: with what emerges between systems and resides in none of them. The superlearner is not produced by a system but by a constellation. Silver builds — rationally: it is his life's work, and the method conquered Go and chess. Sequoia, Lightspeed, Nvidia fund — rationally: whoever sits out the race for superintelligence has already lost it. The critics criticize — rationally: with the best available theory, which is cybernetics. Every actor acts coherently. Together they produce a structure nobody ordered: a race in which the question of whether the goal is reachable no longer has any influence on the pace.
All are guilty. None are at fault.
The Involuntary Proof
Yet Silver’s architecture itself provides the strongest argument—just not the one that cybernetics, in the strict sense, is based on. It removes human data in order to gain autonomy. Structurally, it thereby removes something else, something more fundamental: the second incomplete system.
AlphaZero had an opponent that was non-negotiable—the rules of the game. They declared, without exception: “You lose.” The superlearner is supposed to learn in the open world, using reward signals from the environment. But a measured value is not an opponent. A thermometer does not contradict. It indicates what the system has learned to ask—and remains silent on everything else.
Jonas sought the deficiency in metabolism. Foerster, in the loop. The deficiency is simpler: the loneliness of the soloist without an audience. Two incomplete systems, however, can make each other’s gaps more visible by illuminating these blind spots for one another. A system on its own, which conceives of itself as complete, renders its gaps invisible—and it is precisely in this that it demonstrates its completeness.
The Hypothesis That Cannot Validate Itself
This is a “playground” text, so it’s allowed to do what it describes. It claims that no single system can validate its own discoveries. Consequently, it cannot validate itself. It needs a second system: you. Yes, you out there, who may have just happened to stop by to read this text.
Perhaps you’re misreading the text. Interpreting it differently than originally intended. That, too, would be a confirmation—just one that this text can no longer register.
Insight is not a way out. But it likes company so it can share itself.
Related Posts
How systems theorists reproduce the enclosure milieus Luhmann warned against
Why systems that analyze closure risk reproducing it — and why structural awareness does not automatically prevent structural hardening
The Phenomenon of Coupled Oscillators
Why smart people reject smarter insights—and act intelligently doing so
On piinteract.org:
- ["Séance of Systems Theorists"] — Wiener, Jonas, Foerster at one table: each describes the loop. None describes what emerges between the describers.
- ["Shared Blindspot"] — A single system makes its own gaps invisible. The shared blind spot only shows itself in the plural.
- ["See Problem, Solve Problem"] — The basic monocausal move: one cause, one location, one solution. The superlearner is its most expensive edition to date.
- ["AI Alignment"] — The race in which the reachability of the goal no longer influences the pace.
See also (external links):
Ineffable Intelligence — Mission & Beliefs — Primary source for the Darwin comparison and the goal of discovering all knowledge without human data. The singular is written into the program: one superlearner, one law.
Silver & Sutton: Welcome to the Era of Experience (PDF) — The superlearner's theoretical foundation, first hand: experience replaces human data, reward replaces judgment.
Rosenblueth, Wiener, Bigelow: Behavior, Purpose and Teleology (1943) — The start of the chain: the peer-reviewed first claim that machines can act purposefully.
Hans Jonas: The Phenomenon of Life (Northwestern University Press) — The 1966 objection: purpose requires the possibility of dying. The publisher's page for the authoritative edition.
TechCrunch: DeepMind's David Silver just raised $1.1B — Evidence for funding, valuation and investor list, April 27, 2026.
Paradoxical Interactions (PI): When rational actors consistently produce collectively irrational outcomes — not through failure, but through structure.
All are guilty. None are at fault.
Peter Senner Thinking beyond the Tellerrand
contact@piinteract.org
https://piinteract.org
Co-created with Claude (Anthropic) — two incomplete systems making each other's gaps visible.