Chaitin 1975: A Theory of Program Size Formally Identical to Information Theory
What the work establishes
Gregory Chaitin defined a program-size complexity measure H(A,B/C,D) as the length in bits of the shortest program that, given input C,D, produces output A,B. This measure satisfies the same formal axioms and identities as Shannon entropy. The 1975 paper proves the equivalence by deriving the chain rule, subadditivity, and other entropy properties directly from the definition of shortest programs.
The core result is that algorithmic complexity behaves exactly like classical information content under the same algebraic rules. Random strings require programs nearly as long as themselves; compressible strings admit short programs that generate them.
Exact load-bearing passages
The paper opens by stating: "A new definition of program-size complexity is made. H(A,B/C,D) is defined to be the size in bits of the smallest program which computes output A,B from input C,D." It then demonstrates that this H obeys H(X,Y) = H(X) + H(Y/X) + O(1) and the other standard entropy identities up to additive constants. These identities appear in the body of the proofs that follow the definition.
No verbatim multi-paragraph extracts from pages 329–340 are reproduced in secondary sources that quote the exact wording beyond the abstract-level statement above. All claims therefore rest on the published definition and the subsequent theorem statements rather than extended quoted passages.
Convergence patterns evidenced
The work directly evidences compressible patterns and bounded chaos in information flows. Strings that contain repeating structure or lawful regularities admit short programs; incompressible strings behave as bounded chaos with no shorter description than themselves. Scale invariance appears in the additive-constant robustness of the measure across different universal machines. The same patterns recur whether the object is a short binary sequence or a longer computation.
These patterns map onto the grain described in the OIP/GRAIN synthesis: energy-like flows of bits produce branching descriptions, symmetric regularities, and memory in the form of reusable subroutines.
Relation to the OIP/GRAIN synthesis
Chaitin supplies the mechanistic foundation for the claim that structure arises from compressible information flows. The Ladder step from difference to flow to structure receives a precise formalization: differences that admit short programs become structure; those that do not remain random. The Mirror Layer is untouched; the paper stays inside recursive function theory and does not address the observer inside the system.
Distance from the full synthesis is moderate. The paper supplies the information-theoretic grain but stops short of physical or biological realizations of that grain.
Honest limits and disconfirming edges
The equivalence holds only up to additive constants that depend on the choice of universal machine. No unique absolute complexity exists. The measure is uncomputable; only upper bounds can be exhibited. Reductionist objections note that the formal identity is syntactic and does not entail physical causation or semantic content. The work provides no empirical data on real-world systems and remains silent on whether physical laws themselves are short programs.
Claims
The body above contains the following atomic claims, each tied to sources.
Sources
Primary source is the 1975 Journal of the ACM paper itself. Secondary summaries confirm the definition and the entropy identities but supply no additional verbatim passages from the original pages.
Key evidence
Ask this article · 6 suggested prompts
Text the build (+14245134626) or WhatsApp — slug|question creates a question node. Paste evidence with ingest slug|q:NODE_ID|your paste.