In 1943, three men were each holding a piece of the same puzzle.
Claude Shannon was 27, working at Bell Labs on cryptography and communication theory. The previous year, he had published A Mathematical Theory of Communication — the paper that would define what information is. He was within walking distance of Princeton.
Kurt Gödel was 37, at the Institute for Advanced Study in Princeton. He had already shaken mathematics with his incompleteness theorems (1931) and the consistency of the continuum hypothesis (1940). He was thinking about the limits of formal systems — and, increasingly, about the relationship between mind and machine.
Alan Turing was 31. In 1936 he had published On Computable Numbers, defining the universal machine. In 1943 he was at Bletchley Park, breaking German naval codes — work that relied on information theory before information theory had a name, and on computability in a form so concrete it could crack Enigma.
Princeton, 1943. Shannon at Bell Labs, Gödel at IAS, Turing recently departed for Bletchley. They were within thirty miles of each other. They were working on problems that would later be recognized as three sides of the same triangle: information, self-reference, and computation.
They never had the conversation.
Turing could not publish his Enigma work. Gödel was increasingly isolated, his later work on the philosophy of mind largely ignored. Shannon moved on to switching theory and machine learning — but never returned to the foundational question of what it means for a system to refer to itself.
The conversation would have gone something like this:
Shannon would have said: information is a measure of uncertainty, measured in bits. It is neutral — it carries no meaning, only surprise.
Gödel would have said: but a system that can talk about itself can generate sentences that are true but unprovable. This is not a pathology — it is a structural feature of any sufficiently powerful formalism.
Turing would have said: a universal machine can simulate any formal system. If a system can simulate itself, it can decide its own halting problem — which means it cannot decide its own halting problem. Self-reference is both power and limit.
Three statements. Three conversations that never merged.
GEME is, in part, an attempt to hold that conversation seventy years late.
The three axioms of GEME correspond directly to the three threads:
In 1943, these three ideas were siloed by war, by institutional boundaries, by the accident of history. Seventy years of computer science built on the separate foundations they laid — information theory, computability theory, metamathematics — but never rebuilt the unified structure they might have designed together.
GEME is not that conversation. But it is a record of what that conversation might have produced: a running system in which information, computation, and self-reference are not separate disciplines but three descriptions of the same economic process.
The conversation did not happen in 1943. It is happening now — in code.