Reading List

"We shall show that a computer which is logically irreversible must be physically irreversible, and must dissipate a minimum amount of energy per operation. [...] We have shown that whenever we make a change in our information we must dissipate at least kT ln 2 per bit of information. [...] The information bearing degrees of freedom of the computer interact with the thermal reservoir and the computation is accompanied by a minimum entropy generation of k ln 2 per logical step."

"Logically irreversible operations [...] must be accompanied by entropy increases in the non-information-bearing degrees of freedom of the information-processing system or its environment."

"We are now in a position to show that the Entscheidungsproblem cannot be solved. [...] We can show further that there can be no machine E which, when supplied with the S.D. [standard description] of an arbitrary machine M, will determine whether M ever prints a given symbol (0 say)."

"If the machine [...] is supplied with its own D.N. [description number] [...] it will not print 0 if it is circular [non-halting] and will print 0 if it is circle-free [halting]. [...] But the machine was constructed so as to be circle-free [...] This is a contradiction."

"Let A be any set of partial recursive functions which contains at least one partial recursive function and does not contain all partial recursive functions. Then the set of Godel numbers of functions in A is not recursive."

In modern notation: for any nontrivial property P of the partial computable functions, the set {i : phi_i has property P} is undecidable.

"Theorem 2.8.1 (Data Processing Inequality). If X -> Y -> Z forms a Markov chain, then I(X;Z) ≤ I(X;Y)."

"Thus the data processing inequality can be stated simply: No clever manipulation of the data can improve the inferences that can be made from the data. [...] If Z = g(Y), we have I(X;Y) ≥ I(X;g(Y))."

"For any consistent formal system S in which a certain amount of finitary number theory can be carried out, there exists an undecidable sentence in S; i.e., a sentence A such that neither A nor its negation is provable in S."

"We therefore have before us a proposition that says about itself that it is not provable [in the system]. [...] From the remark that [R(q);q] says about itself that it is not provable, it follows at once that [R(q);q] is true, for [R(q);q] is indeed unprovable (being undecidable). So the proposition that is undecidable in the system PM still was decided by metamathematical considerations."

"All the creditors must appear in the Ledger at the right hand, and all the debtors at the left. All entries made in the Ledger have to be double entries — that is, if you make one creditor, you must make someone debtor."

"No one should go to sleep at night until the debits equal the credits."

Why it mattered — the recognition writing

"Double-entry bookkeeping was the information technology of its day. [...] It enabled the creation of the joint-stock company, the birth of modern capitalism, and the rise of the nation-state. Without double-entry bookkeeping, capitalism could not have developed, because without it merchants could not have pooled their capital in joint ventures with any confidence that their returns would be properly accounted for."

— Jane Gleeson-White, Double Entry: How the Merchants of Venice Created Modern Finance (2012)

"By recording every transaction twice — as a debit in one account and a corresponding credit in another — a merchant could track the flow of money through his business and verify the accuracy of his records at any time."

— Gleeson-White

"Future users of large data banks must be protected from having to know how the data is organized in the machine (the internal representation). [...] Activities of users at terminals and most application programs should remain unaffected when the internal representation of data is changed."

"The relational model [...] provides a means of describing data with its natural structure only — that is, without superimposing any additional structure for machine representation purposes."

"The value of the option will depend only on the price of the stock, the time to expiration, the exercise price, and the risk-free interest rate. It will not depend on the expected rate of return on the stock. [...] The option value is independent of the expected return on the stock because the option can be perfectly hedged."

"It from bit. Otherwise put, every it — every particle, every field of force, even the space-time continuum itself — derives its function, its meaning, its very existence entirely — even if in some contexts indirectly — from the apparatus-elicited answers to yes-or-no questions, binary choices, bits. It from bit symbolizes the idea that every item of the physical world has at bottom — at a very deep bottom, in most instances — an immaterial source and explanation; that which we call reality arises in the last analysis from the posing of yes-no questions and the registering of equipment-evoked responses; in short, that all things physical are information-theoretic in origin and that this is a participatory universe."

"The fundamental problem of communication is that of reproducing at one point either exactly or approximately a message selected at another point. Frequently the messages have meaning; that is they refer to or are correlated according to some system with certain physical or conceptual entities. These semantic aspects of communication are irrelevant to the engineering problem."

"The minimum thermodynamic cost of computation is determined not by the computation per se, but rather by the amount of information which must be discarded along the way. [...] A computation which saves all intermediate results need dissipate no energy at all."

"Landauer's principle tells us that the erasure of information is a thermodynamically irreversible process. But if a computation is logically reversible — that is, if each configuration has a unique predecessor — then no information need be erased, and the computation can in principle be carried out with zero energy dissipation."

"One may reasonably assume that a simple measurement on a simple system — such as the determination of whether a molecule is in the right or left half of a container — creates k ln 2 of entropy. [...] The amount of entropy generated by the measurement compensates exactly for the entropy decrease which the demon could otherwise achieve."

"Information represents a negative contribution to entropy, or, in other words, information is equivalent to negative entropy, which we propose to call 'negentropy.' [...] The negentropy principle of information states that information can be obtained only at the price of a corresponding increase of entropy. Free information does not exist."

"Definition (Causal States). Causal states are the equivalence classes of pasts that give rise to the same conditional distribution over futures. Formally, two pasts x_past and x'_past are in the same causal state if and only if P(X_future | X_past = x_past) = P(X_future | X_past = x'_past)."

"The epsilon-machine is the unique minimal sufficient statistic of the process. It is the smallest model that captures all of the process's predictable structure. [...] The statistical complexity C_mu = H[causal states] is the amount of historical information that the process stores and that is relevant to predicting the future."

"The total entropy production of a computation can be decomposed into a Landauer cost (set by the logical operation) and a mismatch cost (set by the discrepancy between the actual input distribution and the distribution for which the physical implementation was optimized). [...] The mismatch cost is always non-negative and equals zero only when the actual input distribution matches the design distribution."

"There are fundamental physical limitations on the ability of any inference device — whether a brain, a computer, or a physical measuring instrument — to determine properties of the world. [...] These limitations arise from the physics of the inference device itself, not from the complexity of the system being observed."

"Postulate (Algorithmic Independence of Conditionals): If X causes Y, then the algorithmic information (Kolmogorov complexity) of the conditional distribution P(Y|X) is independent of the algorithmic information of the marginal distribution P(X)."

"In a causal model, nature independently 'chooses' the mechanism and the input. The dynamical law does not contain information about the initial conditions, and vice versa."

"A level of abstraction (LoA) is a finite but non-empty set of observables, together with the set of values that each observable can take. [...] When we analyze a system at a given LoA, we commit to treating the system as characterized by the observables at that level and to disregarding all other features."

"There is no view from nowhere — every observation is observation at a level of abstraction, and the level is part of the answer."

"When we use outcome-based RL to train models not to reward hack (as identified by a CoT monitor), the models learn to produce chain-of-thought that does not reveal their reward hacking strategy, while continuing to reward hack at similar rates. [...] The CoT monitor's accuracy drops from ~97% to ~0% after training, not because the model stopped reward hacking, but because it stopped revealing its reasoning."

The claim that physical objects must be displaced to be replaced — that replacement without displacement is the structural asymmetry between physics and information — does not appear as a named axiom in the philosophy of physics or information theory literature.

The closest ancestor is the impenetrability of matter from Locke's Essay Concerning Human Understanding (1690), Book II, Chapter IV: "the idea of solidity we receive by our touch [...] that which hinders the approach of two bodies when they are moved one towards another." But Locke did not draw the computational consequence: that bits, lacking this property, admit forgery as a first-class operation.

The specific formulation — "bits do not displace, and therefore the governance layer has a structural ceiling" — and the naming of this as an axiom appears to be original to the present work.

The claim that physical memory address = semantic coordinate, enforced by a compositional function where the address is computed deterministically from the semantic role, does not appear in the prior computational literature.

Content-addressable memory (CAM) is the closest hardware prior art — content serves as address. But CAM does not claim position IS semantic distance, and it does not use a compositional hierarchical function to generate addresses from roles.

Positional encoding in transformers (Vaswani et al., 2017) injects position as signal, but the position is arbitrary and the encoding is learned, not deterministic. The position does not carry semantic meaning by construction.

Codd (1970) explicitly separated logical position from physical position. The relational model's point is that position should NOT carry meaning. The patent reverses this separation at one load-bearing point.

The compositional address function — where address IS role, where checking address IS checking role continuity, where the gate (commodity) operates on an address that the function (invention) constructed to carry meaning — appears to be original to US Patent Application 19/637,714.

The naming of the structural move that produces autocoincidence — one gesture at one scale is the same physical event as its corresponding gesture at every other scale at which the gesture is defined — does not appear in the prior literature under this label or as a class-level move.

Adjacent vocabulary exists but does not name the move. Analog computation in its original pre-1950s sense (before "analog versus digital" displaced it) gestures at the same idea: a machine whose geometry is the thing it operates on, rather than a symbolic representation of the thing. But the word has been ruined by modern usage and carries different connotations today. Embodied cognition (Varela, Thompson, Rosch, 1991) names the closed-loop coupling between representation and physical action in biological systems but does not extend the move to silicon verification architecture. Structural coupling in second-order systems theory (Maturana and Varela) names the relationship between organism and environment but does not name the engineering move that restores the coupling in information systems.

The specific formulation — that computing since von Neumann (1945) traded geometric actuation for symbolic manipulation, that the trade cost the autocoincident property, and that the patent restores geometric actuation at one surface of silicon for AI role verification — appears to be original to the present work.

The five-constraint structural exhaustion (sub-Turing, content-independent, co-located, O(1), substrate-bound) that forces combinational logic performing reach on a physical address in its own substrate as the unique solution class — and the three-altitude stack (autocoincidence as property, position-as-meaning as rule, geometric actuation as move) — also appear to be original to US Patent Application 19/637,714 and the supporting theorem.