A high-order language whose syntax is bound to nature itself.
Ever since Leibniz dreamt of a characteristica universalis, a symbolic language in which every dispute could be settled by calculation, we have been searching for a way to speak about reality without ambiguity. Frege, Russell and Hilbert took up the dream in earnest. Gödel and Turing then showed where it ran aground. Physics, meanwhile, split into two great theories that refuse to be unified. Today, large language models promise a new kind of universality by absorbing the whole written record of humanity.
Yet that is exactly where the problem lies. LLMs are built on human languages, and human languages were shaped for human perception, not for meaningful contact with reality at its fundamental level. This holds even for our most precise language, mathematics. Its symbols are largely the inherited choices of its creators, a historical sediment of conventions laid down by mathematicians, logicians and philosophers. Such languages are countable: they are made of discrete units of tangled, mixed meaning that are of little use in themselves beyond the arc of human, terrestrial development.
Since at least the Enlightenment, mathematics has been the foundational tool of human progress. In the nineteenth and twentieth centuries its finest minds tried not only to ground human certainty in mathematics but to bind it directly to physics. In hindsight the attempt was bound to fail: mathematics was never built for that task. Its arithmetic axioms carry their own incompleteness, and its symbols carry no native link to the processes they describe.
Centrics was born of this realization. If the instrument is unfit for the assignment, the answer is not a better proof. It is a better language.
Enter centrics, an uncountable language. By uncountable we mean a language not bounded by arithmetic axioms, and therefore beyond the limits Gödel and Turing placed on formal systems and machines. Every operation, every operator and every fundamental object in centrics (the so-called primod) has a counterpart in, and a link to, a natural process.
This has a striking consequence. Doing centrics, unlike doing mathematics, is in some measure manipulating nature, or at the very least manipulating our own brains. Its syntax and its syntactic and semantic structure demand complete precision. An imprecise expression does not quietly mean something vague; it produces an error, much like a bug in a computer program. In this sense centrics is a high-order language, set against the low-order languages on which our civilization currently runs.
At its core sits the Heptad: seven intrinsic theories, each carrying a canonical operator and each unfolding into a triality of matter, energy and information. Together they form a closed operator algebra, a universal syntax in which mathematics, physics and computation can be written at both the object level and the meta level.
Our thesis is simple. As a civilization evolves, so must its language systems, whether written, spoken or yet to be imagined. By binding its syntax to nature itself, centrics is the next step in our species' linguistic evolution.
Anchors every operation to the universal limit — undifferentiated potential. Causal product, sum, difference and decomposition act on field elements: fusion, influx, extraction, de-linkage.
To carry centrics from paper into practice, we founded Cendroid AG. We are designing and building two things that belong together: CENTRON, a programming language in which every operation is tagged with its place in the Heptad, and CENDROID, a neuro-symbolic analog computer architecture whose modules mirror the seven theories.
Combined, they form an AI oracle blackbox: a machine that analyzes and computes sentient problems by referring to the fundamental laws of nature, codified in the very language it runs on. Where today's models extrapolate probabilities from the human record and are now meeting its limits, an HLM (a high-order language model) reasons from causes.
We envision a future in which HLMs become standard across science and technology, and in which human evolution itself moves forward with them.
| ASPECT | LLM · CONVENTIONAL | HLM · CENTRICS / CENDROID |
|---|---|---|
| Syntax & semantics | Tokenised; syntax external to the model | Syntax equals semantics; internal machine code |
| Foundation | Countable human languages and arithmetic | Uncountable; every primod linked to a natural process |
| Knowledge | Continuous vectors, no algebraic structure | High-order objects, algebraically rich |
| Reasoning | Probabilistic extrapolation, no causality | Causal, operator-based, globally coherent |
| Errors | Hallucination, prompt injection, drift | Ill-formed output is rejected by the calculus |
| Architecture | Static, digital, von Neumann bottleneck | Analog neuro-symbolic, self-modifying |