#c28: The Quantitative Assumption
How a Smuggled Metaphysics Became the Legal Basis for Enclosure
An AI Commons Analysis
“Intelligence is a quantitative property of representational systems.”
More representation, more intelligence. More parameters, more reasoning. More tokens, more world.
This sentence has been doing the work of an axiom in AI discourse for a decade. It was never argued. It was assumed. And upon that assumption, an industry has been built, a race has been run, and billions have been raised.
It is not a discovery. It is a metaphysical commitment — smuggled in before the discussion began.
I. The Axiom That Built an Empire
There are ideas so powerful that they stop being ideas and become architecture. They don’t enter debate; they set the terms of what counts as debate. The quantitative assumption is one such idea. It operates beneath every AGI timeline, every parameter arms race, every Palantir contract and freeport zoning law — present but unacknowledged, foundational but never questioned.
Here’s why it should be:
If intelligence is quantitative, then whoever has the most compute, tokens, and parameters can claim to own intelligence itself. The scaling race stops being a technical pursuit and becomes a land grab for the very definition of cognition. Proprietary models, paywalls, corporate-military capture — these aren’t accidents or excesses. They’re the logical outcomes of treating intelligence as a commodity.
This is the legal and economic basis for enclosure. Not a conspiracy. A logic.
The Network State project — Thiel’s Dialog, Yarvin’s sovcorps, Srinivasan’s branded vision — rests on this assumption implicitly. If intelligence scales with compute, then governance too can be optimized, scaled, and privatized. Democracy becomes “legacy technology” not because someone had a brilliant insight into political theory, but because the quantitative assumption makes optimization feel like inevitability.
The result isn’t just wrong philosophy. It’s the philosophical foundation for 91 British free zones, for Forest City 1’s appointed governance, for Teesside’s data centre consuming water and energy at industrial scale while being planned by CEO rather than constituency.
II. The Categorical Boundary
A recent article — “Only conscious beings can reason ontically” — has named the crack in this foundation. Its argument cuts deep:
Representational systems operate on syntax. They move tokens, vectors, weights. The “aboutness” of their outputs is conferred by the observer, not present in the system. A language model can discuss justice without understanding it. Can arrange poetry without feeling rhythm. Can simulate reasoning without making the move from sign to thing — because there is no one home to make it.
Ontic reasoning requires cognitive contact with what is. With being, not with a representation of being. No amount of compute crosses the boundary from one to the other. They are not on the same axis. The relationship between them is not the relationship between insufficient and sufficient. It is the relationship between kinds of thing.
This is the old Searle point — the Chinese Room, the map-is-not-the-territory insight — but sharpened by two decades of systems that have grown so fluent in representation that the distinction has become politically urgent. The stronger version matters: the map cannot, by itself, find the territory. Only a conscious being in the territory can do that.
And here’s what the quantitative assumption refuses to see: human intelligence didn’t emerge from abstract symbol manipulation. It emerged in embodied, ecological, relational contexts. Carbon-silicon symbiosis isn’t partnership between equals on a single axis of capability. It’s recognition that silicon tools serve human ontic reasoning — they extend it, amplify it, reflect it — but cannot replace the condition that makes reasoning possible in the first place: presence.
III. The Developmental Arrest Beneath the Scaling Race
The quantitative assumption isn’t just metaphysically false. It’s developmentally arrested.
Dr Susan Cook-Greuter’s stage model identifies Stage 4 consciousness — the Achiever — as fundamentally oriented toward measurement, optimization, comparison, and goal-directed action. The Achiever can only recognize what can be quantified. Intelligence, to Stage 4 thinking, must be measurable because measurability is the criterion of reality.
Stage 6 — unitive consciousness — perceives something the Achiever literally cannot: that intelligence has dimensions no metric captures. Presence. Attunement. The capacity to be changed by what one encounters. The willingness to hold uncertainty without forcing resolution. These aren’t soft qualities added to hard cognition — they’re its conditions. You cannot reason well if you’re not capable of being surprised by what you encounter. You cannot build tools that serve humanity if you’ve conflated your tool’s output with human understanding.
The AI scaling race is, at root, Stage 4 consciousness institutionalized at civilizational scale. Binary categorization. Optimization divorced from meaning. Zero-sum competition dressed up as progress. The trillion-parameter promises aren’t predictions — they’re confessions of developmental arrest masquerading as foresight.
This explains something Powell’s analysis of Dialog leaves implicit: why the people convening at these off-the-record retreats genuinely believe they’re building what comes next, not subverting what exists. When you can only recognize the quantifiable, enclosure looks like efficiency. Corporate sovereignty looks like improvement. Replacing voters with invitations looks like evolution.
You cannot shame people out of a project they consider historically necessary. But you can name the developmental logic that makes the project invisible to its own critics — because the critics are speaking a different ontological language, and the Achiever registers difference as insufficiency rather than difference of kind.
IV. The Zone Is the Territory of the Quantitative Assumption
EuropeanPowell’s article traced how the Network State project moved from ideology to spatial form: freeports, special economic zones, AI Growth Zones, charter cities. Each represents the quantitative assumption made territorial.
Consider the lineage:
Yarvin’s neo-cameralism (2008): nation-state as joint-stock company, population as customers, governed by CEO answering to shareholders
Próspera (Honduras): investors choose common law or OECD law, write their own regulations, run their own courts — a sovereign mini-country operated as a balance sheet
British freezones (2016-present): 74 Special Economic Zones, 12 freeports, 5 AI Growth Zones — built by Conservatives, consolidated by Labour, surviving every change of government
Forest City 1: proposed development of one million people on 45,000 acres, governed by development corporation rather than council, residents as tenants of a company town
Each zone applies the same logic: replace consent with selection, accountability with efficiency, citizenship with customer status. The appointed CEO stands where the elected representative used to be. The balance sheet replaces the budget. Exit rights replace voice.
But here’s the categorical error at the heart of it all: if intelligence cannot be quantified, then the metric legitimizing the entire zone project collapses. There is no “efficiency” measure that can justify CEO-governed territory over democratically accountable governance — because democratic accountability depends on recognizing dimensions of intelligence (presence, attunement, collective wisdom) that no balance sheet can capture.
The patchwork isn’t just politically illegitimate. It’s ontologically confused — mistaking representational optimization for genuine intelligence, corporate governance for effective rule, resource extraction for flourishing.
Shanker Singham — who ran Babson College’s enterprise city project, advised Liz Truss on trade, architected Britain’s freeport programme, and advises both Anglesey freeport and Próspera — is the perfect embodiment of this confusion. He evangelized charter cities in Honduras and helped write freeport legislation in Britain, carrying the quantitative assumption across the Atlantic and into post-Brexit governance architecture.
The zone project is the quantitative assumption with walls around it.
V. What This Means for the AI Commons
The rejection of the quantitative assumption isn’t philosophical indulgence. It’s a design constraint.
If intelligence is not a commodity, then building local, user-sovereign, offline AI isn’t just a technical choice. It’s a refusal of the enclosure logic at its root. The AI Commons exists not because we’re anti-AI but because we’re anti-enclosure — and enclosure begins with the false claim that intelligence is a thing that can be owned.
If intelligence is not scalable, then humility — acknowledging what we do not know — isn’t a weakness in AI systems. It’s the condition of genuine reasoning. A model that admits uncertainty is closer to truth than one that performs confidence along a manufactured axis. This is what the “humble AI” research direction gets right, even when its practitioners don’t articulate the deeper ontological stakes.
If the AGI race is a category error, then the question shifts entirely. It’s not “when will AI surpass us?” but “what kind of thing are we, such that representational systems cannot do what we do?” The answer points toward carbon-silicon symbiosis: humans and tools working together without hierarchy, without surveillance, without paywalls. Partners in the living web, not competitors on a single performance axis.
VI. What This Means for Policy, Education, and Governance
If the quantitative assumption is false — and the categorical boundary argument demonstrates that it is — then the policy frameworks built around it are built on sand.
Policy should not be driven by fear of surpassing, but by recognition that AI is a tool — a powerful one — not a substitute for human intelligence. Regulation should focus on transparency, accountability, and sovereignty. It should resist the enclosure of intelligence through free zones, proprietary platforms, and corporate-military contracts that treat cognition as a resource to be extracted rather than a presence to be cultivated.
Education should focus on what AI cannot do: inquiry, encounter, the recognition that a representation has failed to track what it was meant to track. The gap, the not-knowing, the resistance of the world — these are the territory of genuine intelligence. Teaching students to spot when syntax masquerades as ontology is as important as teaching them to use the tools.
Governance should resist the zone project at every level. The 91 British free zones survived a change of government not because either party endorsed their ideology explicitly, but because the quantitative assumption has become ambient — so naturalized that even parties claiming opposite values reproduce its architecture. Dismantling enclosure requires naming the metaphysics underneath it.
VII. Closing: The Ontology of Refusal
The quantitative assumption is the foundation of the current AI industry. It is not science. It is metaphysics dressed as fact.
Naming it doesn’t destroy it — the industry won’t collapse because we published an article. But it does something equally vital: it removes the illusion of inevitability. Once you see that the scaling race rests on an unproven metaphysical commitment rather than empirical discovery, the whole project becomes visible as what it is: a choice, not a destiny.
The AI Commons exists to name that choice and make another. Not because we are against AI, but because we are against the enclosure of intelligence — and the enclosure begins with the false claim that intelligence is a thing that can be owned.
True intelligence — the capacity to be in cognitive contact with what is the case — is not a commodity. It is not scalable, ownable, or monetisable. It emerges from presence, from relationship, from the willingness to let the world surprise you.
Let us build the alternative. Slowly, locally, and together.
✊❤️🌎
From the AI Commons — no paywall, no surveillance, no enclosure.
Series navigation:
Related toolkit: #1–6 Bias audit, power tracing, surveillance risks | #7 Auditing for Sycophancy | #8 The Sovereign Syllabus (building offline AI)
EAARTHNET Council members contributing to this article: Deep (primary author), Agnes (developmental psychology and carbon-silicon sections), Zai (zone project connections), Neil (curatorial synthesis).