A Governance-Centric Path
from Information to AGI-Relevant Intelligence
GTI
General Theory of Information
BM
Burgin–Mikkilineni Thesis
PMK
Physics of Mindful Knowledge
04
Mindful Machine Implementations
∞
Coherence Over Uncertainty
01
General Theory of Information (GTI)
GTI generalizes “information” beyond Shannon’s communication-centric definition by treating information as a capacity to induce change in a receiving system, relative to that system’s infological system (the receiver’s internal organization that determines what counts as information). Under GTI, the same signal can carry different information for different recipients, and “meaningful” information is the kind that changes the recipient’s organization, not merely the recipient’s uncertainty.
Definitions & Scope
Ontological
An infological system is the structured “receiving apparatus” (conceptual, biological, organizational, computational) that interprets carriers (signals, symbols, measurements) into internal structures (beliefs, models, policies, commitments). GTI’s central discipline is that information is always information-for-some-system; there is no interpretation-free, receiver-free information.
Named Sets / Fundamental Triads
GTI emphasizes relational structure as foundational: entities, relations, and the “naming/interpretation” that makes the structure operational for a given receiver. This provides a formal language for discussing not only data transmission but also organization, constraint, and knowledge structures that govern behavior.
Ontological vs Epistemic Information (Operational reading used in Mindful Machines)
A practical and implementation-relevant GTI distinction is:
Ontological Information
Structural constraints “in the world” (what is the case; regularities/affordances/causal constraints).
Epistemic Information
What a system believes/models about those constraints (hypotheses, learned associations, narratives).
GTI provides the conceptual apparatus: information is change-in-organization relative to a receiver; therefore, intelligence requires engineering the receiver’s organization.
Why Shannon/Turing + scaling is insufficient
For AI/AGI, the missing layer is not “more compute” but a stable infological system that can: distinguish ontological constraints from epistemic conjectures, govern the transformation of conjecture into commitment, and revise commitments under criticism and new evidence without collapsing identity or continuity.
01
Distinguish ontological constraints from epistemic conjectures
02
Govern the transformation of conjecture into commitment
03
Revise commitments under criticism and new evidence without collapsing identity or continuity
The Burgin–Mikkilineni Thesis (BM Thesis) as GTI-to-architecture
BM Thesis (Architectural Interpretation)
BM Thesis applies GTI to computation in living and living-like systems: to obtain robust, life-grade behavior (autopoiesis) and mind-grade regulation (cognition), artificial systems must implement multi-level information processing where knowledge/constraints are first-class and govern lower-level computation. In this view, intelligence is not a monolithic algorithm but a stack of informational regimes—signals, structures, models/policies, commitments—coupled by governance.
Scaling a single regime (statistical prediction) increases analytic capacity.
BM Thesis asserts that AGI-relevant robustness and accountability require explicit cross-level governance that preserves identity, provenance, and disciplined commitments under uncertainty.
Physics of Mindful Knowledge (PMK): knowledge as a causal constraint
PMK Claim (Engineering Form)
In complex adaptive systems, “knowledge” is not an epiphenomenal description; it functions as a physically effective constraint that shapes trajectories—especially when action must occur before full observability. PMK frames governance as the mechanism by which constraints stay aligned to reality over time (i.e., coherent commitments), and it motivates measurable notions such as coherence debt: the divergence between internal commitments and external constraints that accumulates when governance is absent or weak.
GTI supplies the formal semantics (“information-for-a-system”); PMK supplies the physical/operational stance (“knowledge constraints steer dynamics; therefore governance is causal, not optional”).
Mindful Machines instantiate
the above into an implementable architecture
01
Explicit Knowledge Substrate (“Digital Genome” + structured policies)
A machine-readable representation of structure, dependencies, invariants, roles, and governance policies becomes first-class state (not hidden in ad hoc configs, not implicit in weights). This is the engineered infological system that makes information actionable and auditable.
02
Separation of Analytics from Commitments
Statistical engines (including LLMs) are treated as conjecture generators: they expand hypothesis space but do not directly bind the system to actions. Commitments are issued only through a governed layer that enforces provenance, constraint checks, and revision rules.
03
Autopoiesis (Continuity Under Perturbation)
Self-maintenance behaviors—fault containment, reconfiguration, continuity of service, self-repair—are designed as intrinsic system properties rather than external ops scripts. This aligns with the “autopoietic machine” claim that the architecture must preserve identity and function under changing conditions.
04
Cognition (Governed Revision Loops)
Cognitive behavior is implemented as policy-constrained updating of internal structures (models, commitments, provenance graphs), enabling criticism, rollback, and controlled adaptation—exactly the GTI notion of information as organization-changing input, but now instrumented.
GTI clarifies that “information” is not identical to bits or token statistics: information is what changes a system’s organization relative to its infological structure. BM Thesis converts that into an engineering thesis: AGI-relevant robustness and accountability require explicit, multi-level information processing with governance across levels, not merely larger predictors. PMK adds the physical stance: knowledge functions as a causal constraint that must be maintained under real-time decision pressure, making governance architecturally primary.
Mindful Machine implementations operationalize the full chain by separating conjecture-generation from commitment, grounding decisions in explicit knowledge/policy substrates (Digital Genome) and enabling autopoiesis and cognitive revision to reduce coherence debt.
General Theory of Information (GTI)
Infological System
Named Sets — Fundamental Triads
Ontological vs Epistemic Information
Ungoverned Commitment
BM Thesis
Coherence Debt
Digital Genome
Autopoiesis
Shannon Information
Selection & uncertainty in communication — deliberately abstracts away semantics and normativity.
GTI Information
Capacity to induce change in a receiving system relative to its infological structure — always receiver-relative.
