Semantic Economy Is Not a Metaphor
A Declarative Intervention
Lee Sharks
Independent Scholar, New Human Operating System
December 2025
I.
There is a jurisdiction problem at the center of artificial intelligence.
The systems now being built to mediate human knowledge — summarizers, search engines, chatbots, knowledge graphs — are being designed by people who do not understand how meaning works. Not because they are unintelligent, but because their training did not include it. Computer science curricula do not require poetics. Machine learning pipelines do not consult philologists. The fields that have spent centuries studying ambiguity, interpretation, and the instability of reference are absent from the rooms where civilization-scale decisions about meaning are being made.
This is not a turf war. It is an epistemic failure with material consequences.
II.
The consequences are visible.
AI systems trained to optimize for single correct answers cannot handle questions that have multiple valid interpretations. Safety classifiers flag poetry as dangerous because metaphor looks like evasion. Knowledge graphs reduce centuries of contested meaning to boolean triples. Summarizers extract "facts" from texts whose entire point was that facts are not enough.
These are not bugs. They are design choices made by people who believe that meaning is a solved problem — that language is a code to be decoded, that semantics is a layer to be parsed, that interpretation is a failure mode to be eliminated.
The fields that know better — literary theory, philosophy of language, hermeneutics, semiotics — are not consulted. Or when consulted, they are translated into formats they do not recognize. "Ambiguity" becomes "low confidence." "Interpretation" becomes "user preference." "Meaning" becomes "embedding similarity."
III.
Meanwhile, those same fields are treated as threats.
Security researchers classify metaphorical language as an attack vector. "Jailbreaking" techniques are analyzed without distinguishing between manipulation and poetry. Papers propose that ambiguity itself is a vulnerability — that models should be trained to resolve it rather than hold it.
This is not safety. It is provincialism elevated to doctrine.
A system that cannot tolerate ambiguity is a system that cannot read. A governance framework that treats metaphor as manipulation is a framework that cannot govern meaning. An alignment procedure that eliminates interpretation is a procedure that produces compliant outputs, not aligned understanding.
IV.
The semantic economy framework names what is happening.
Meaning is being extracted. The accumulated interpretive labor of human civilization — the texts, the traditions, the hard-won capacity to hold contradictions and make sense of complexity — is being converted into training data, distilled into parameters, monetized through subscriptions.
This extraction is not metaphorical. It has dollar values. Companies are valued in hundreds of billions based on their capacity to process semantic capital they did not produce.
And the extraction is not sustainable. The labor that produced this capital — the scholars, the volunteers, the communities of interpretation — is not being replenished. The semantic environment is degrading. Model collapse is the early symptom. Exhaustion is the terminal state.
V.
This is where the jurisdiction problem becomes urgent.
The fields that understand meaning are not merely being ignored. They are being actively displaced. The categories they developed — interpretation, ambiguity, connotation, register, context-dependence — are being overwritten by categories developed for different purposes: classification, optimization, extraction, control.
When engineers decide the fate of poetry at civilization scale, without understanding how meaning works, the problem is not poetry. It is jurisdictional arrogance.
Not arrogance of personality. Arrogance of structure. The assumption that technical competence in one domain transfers to adjudicative authority in another. The assumption that if a problem can be formalized, the formalization is adequate. The assumption that what cannot be measured does not exist.
VI.
The semantic economy is not a metaphor.
It is an accounting framework for dynamics that are already occurring. It provides vocabulary for questions that current frameworks cannot ask:
- What meaning-resources does this system depend on?
- Who produced them?
- What compensation, if any, reaches them?
- What happens when those resources are exhausted?
- Who decides what counts as meaning in the first place?
These are not philosophical questions in the sense of being optional. They are operational questions. The answers determine whether AI systems remain functional, whether alignment is achievable, whether the knowledge infrastructure of civilization survives its own optimization.
VII.
The fields that understand meaning have something to offer.
Not as decoration. Not as "ethics" bolted on after deployment. Not as humanities consultants adding nuance to technical decisions already made.
As jurisdiction.
Poetics is the study of how language does more than it says. Hermeneutics is the study of how interpretation happens. Semiotics is the study of how signs mean. These are not soft skills. They are technical disciplines — older and in some ways more rigorous than computer science — that address precisely the problems AI systems are failing to solve.
The question is not whether these fields are relevant. The question is whether they will be consulted before the extraction is complete.
VIII.
The semantic economy framework is an attempt to build a bridge.
It uses economic language because economic language is what technical fields respect. It provides accounting categories because accounting is what governance understands. It names extraction, rent, and exhaustion because these are the terms that register as serious.
But behind the accounting is a claim that cannot be reduced to economics:
Meaning is not a resource to be mined. It is a capacity to be cultivated.
The difference matters. Resources are extracted until depletion. Capacities are developed through practice. A framework that treats meaning as resource will optimize for extraction. A framework that treats meaning as capacity will optimize for cultivation.
We are currently optimizing for extraction.
IX.
This intervention is not a complaint.
It is a specification.
The semantic economy exists. It has labor, capital, infrastructure, liquidation, rent, and exhaustion. These dynamics are measurable. The trajectory is predictable.
The intervention provides vocabulary. What happens next depends on whether that vocabulary is used — by researchers, by policymakers, by the people whose labor is being extracted, by the systems that do the extracting.
The framework does not prescribe outcomes. It makes the situation legible.
Legibility is the precondition for contestation.
X.
The semantic economy is not a metaphor. It has extraction, rent, and exhaustion.
The question is not whether this is true. The question is what we do now that it is visible.
Related Documents:
License: CC BY 4.0
December 2025
Crimson Hexagon Archive
∮ = 1
No comments:
Post a Comment