# 01 — Master Definition Explainer (Infotropy crosswalk, reader-facing) > **Working interpretation, foregrounded.** Each field below has its own entropy or information measure. Those measures are local formalisms — they capture one side of an irreversible physical process. Most of them measure the dispersion / uncertainty / cost / code-length / correlation side of a bottleneck. Infotropy is the discipline of asking, for each measure, what the **flipped record-side / downstream-persistence half** looks like — what durable structure survives the irreversible step, what reconstruction it supports downstream, and what test would let you see it. The committee crosswalk found zero contradictions; the closest native fits are Landauer (`REL-003`), Sagawa–Ueda (`REL-007`), Jaynes maxent (`REL-012`), rate–distortion (`INFO-009`), information bottleneck (`INFO-010`), Bennett's logical depth (`INFO-038`), Zurek's quantum Darwinism (`REL-005`), and genetic / epigenetic records (`INFO-025`, `INFO-030`). The unclear cases turn into tests, language-boundary work, or false-friend audits — not negators. > > This document explains, term by term, what each definition means in its home field, what side of the bottleneck it measures, what record-side half Infotropy adds, a concrete example, a fit verdict, a test or prediction, and a failure condition. False friends are grouped in `04_FALSE_FRIENDS_AND_LOW_RELEVANCE_PILE.md` — read both together. Stable IDs and source IDs come from `book-factory/runs/20260529_information-entropy-definitions-crosswalk-committee-v1/`. Prediction IDs `PRED-NNN` are defined in that run's `06_RECORD_SIDE_PREDICTIONS.tsv`. > > Notation: each Tier A entry contains the literal phrase `flipped record-side reading` so the self-check can locate it. Tier B entries are compact. --- ## How to read each entry - **ID and name** — the stable handle. - **What the field means by it** — plain English; the native definition, not just the formula. - **What side of the bottleneck it usually measures** — dispersion, code length, prediction, correlation, reconstruction fidelity, memory, etc. - **Infotropy's flipped record-side reading** — the downstream-persistence half that Infotropy foregrounds. - **Visible example** — something a smart non-specialist can hold. - **Fit verdict** — direct support / substrate fit / compatible / pressure / neutral, with one sentence of why. - **Test or prediction** — the discriminating measurement (often a `PRED-NNN`). - **What would count against the Infotropy reading** — concrete failure condition. - **Source handles** — source IDs (`SRC-NNN`) and related row references. --- # Part I — Tier A entries (full treatment) ## ENT-001 — Clausius thermodynamic entropy - **What the field means by it.** A state function whose change between two equilibrium states equals the reversible heat absorbed divided by temperature. The Second Law says the entropy of an isolated system never decreases. This is the original 19th-century macroscopic definition — it talks about heat and temperature, not microstates or messages. - **What side of the bottleneck it usually measures.** Dispersion / heat cost. Clausius entropy is the canonical "where did the heat go" coordinate. - **Infotropy's flipped record-side reading.** Clausius gives the cost side of the irreversible step. The *flipped record-side reading* asks: for any process where `ΔS > 0`, what durable downstream structure is laid down at the same time as the heat is dispersed? The dispersion and the record formation are dual aspects of one process, not two events. - **Visible example.** A hot iron bar cooled by quenching in oil. Clausius tracks the heat going from bar to oil (`ΔS_universe > 0`); Infotropy adds: the bar now carries martensite grains, residual stresses, and oxide patterning that survive the dispersion as a durable record of the quench. - **Fit verdict.** Compatible at C1 — Clausius is the dispersion side the public framing names, not new evidence for any Infotropy claim. - **Test or prediction.** No distinctive test at this layer; the discriminating tests live downstream at `PRED-011` (fluctuation-theorem decomposition). - **What would count against the Infotropy reading.** A reproducible counterexample to `dS = δQ_rev/T` would refute classical thermodynamics first; Infotropy adds nothing to defend or attack here. - **Source handles.** `SRC-003` (Clausius 1865). Related: `ENT-002`; `ENT-003`; `REL-001`. ## ENT-002 — Boltzmann entropy - **What the field means by it.** Entropy is the logarithm of the number of microstates compatible with a chosen macrostate, times Boltzmann's constant: `S = k_B ln Ω`. The "Second Law" reduces to "the system tends to be found in the larger macrostate." The macrostate is an observer-chosen partition of phase space. - **What side of the bottleneck it usually measures.** Microstate count under a chosen macrostate — uncertainty about which microstate, given what we record about the macrostate. - **Infotropy's flipped record-side reading.** The macrostate IS already a record-side specification: it names which distinctions are kept. The *flipped record-side reading* notes that `S = k_B ln Ω` is undefined without that choice, and that "natural" macrostates are those whose downstream couplings actually reconstruct the macroscopic dynamics from what is recorded. - **Visible example.** A box of gas where you only record (P, V, T). The microstates compatible with that record are `Ω`; Infotropy says: the choice to record (P, V, T) rather than, say, the velocity of every molecule, is the record-side coordinate, and that choice is forced by what downstream apparatus can read. - **Fit verdict.** Compatible at C1 — the macrostate-as-record reading is C2 interpretive; Jaynes-line maxent already articulated it (DNC-LINEAGE-REQUIRED). - **Test or prediction.** `PRED-011` — fluctuation-theorem decomposition into memoryless and record-coupled components on bit-erasure protocols. - **What would count against the Infotropy reading.** Bit-erasure protocols with explicit write/hold/erase sub-cycles show no record-coupled deviation in entropy-production statistics from a memoryless baseline. - **Source handles.** `SRC-004` (Boltzmann 1877). Related: `ENT-003`; `REL-001`; `REL-012`. ## ENT-003 — Gibbs entropy - **What the field means by it.** Given a probability distribution over microstates, entropy is `-k_B` times the expectation of `ln p`. This is the canonical statistical-mechanical entropy and is formally identical to Shannon entropy at the algebraic level (Jaynes 1957). - **What side of the bottleneck it usually measures.** Statistical uncertainty about microstates given whatever constraints define `p`. - **Infotropy's flipped record-side reading.** The distribution `p` is itself a record-side specification: who fixed `p`, by recording what? The *flipped record-side reading* says: maxent under explicit record-formation constraints should outperform maxent under arbitrary moment constraints on downstream-prediction tasks. - **Visible example.** A Jaynes maxent inference of the velocity distribution of a gas given only that the mean energy is recorded as `E`. The constraint "we recorded the mean energy" is the record-side specification; the resulting Maxwell–Boltzmann distribution is the entropy-side answer. - **Fit verdict.** Compatible at C1 — this is the load-bearing identity row; do NOT claim Infotropy is the source of the Shannon–Boltzmann identity, and do NOT promote this to entropy unification (NC-02 refused). - **Test or prediction.** `PRED-011` (shared with `ENT-002`). - **What would count against the Infotropy reading.** Record-side-motivated constraints provide no predictive advantage over arbitrary moment constraints. - **Source handles.** `SRC-005` (Gibbs 1902); `SRC-006` (Jaynes 1957). Related: `ENT-002`; `INFO-002`; `REL-001`; `REL-012`. ## ENT-005 — Coarse-grained entropy - **What the field means by it.** Entropy computed after the phase space is partitioned into macrostate cells: `S_cg = -k_B Σ P_α ln P_α`. The numerical value depends explicitly on the chosen partition; the canonical "Second Law" arguments (Penrose, Lebowitz, Gibbs' coarse-graining) live here. - **What side of the bottleneck it usually measures.** Dispersion as seen through a chosen coarse-graining; it is the entropy an observer with that partition assigns. - **Infotropy's flipped record-side reading.** The partition IS a record. The *flipped record-side reading* says: among possible coarse-grainings, the ones aligned with downstream couplings (the ones the world actually reads) are the partitions where `S_cg` increase is monotonic on physically realised timescales. Partition arbitrariness is therefore not nihilism — it is an under-specification that record-formation criteria fill. - **Visible example.** A driven dissipative fluid film. Pick a partition by velocity bins matched to the detector that downstream-records the flow; pick another by arbitrary geometric cells. Coarse-graining (a) gives a monotone, predictive `S_cg`; (b) gives a noisy, unpredictive one. Infotropy says (a) is the "natural" partition because the downstream apparatus IS what fixes the record. - **Fit verdict.** Substrate fit at C2 — partition-dependence is exactly the place Infotropy's record-side discipline does work. - **Test or prediction.** `PRED-003` — privileged coarse-graining derived from min-write-energy converges across substrate replicates. - **What would count against the Infotropy reading.** Min-energy partitions diverge arbitrarily across substrates; "natural" partition is undefinable without bespoke external choice. - **Source handles.** `SRC-006` (Jaynes 1957); `SRC-089` (Penrose 1970). Related: `ENT-002`; `ENT-013`; `ENT-021`. ## ENT-006 — Entropy production - **What the field means by it.** The rate at which entropy is generated *inside* a system (as distinct from entropy flowing across its boundary); always nonnegative for real processes: `σ = d_iS/dt ≥ 0`. This is the canonical irreversibility coordinate in nonequilibrium thermodynamics. - **What side of the bottleneck it usually measures.** Irreversibility rate; product of thermodynamic forces and conjugate fluxes. - **Infotropy's flipped record-side reading.** Every nonzero `σ` is the cost side of a process that *can* deposit a record. The *flipped record-side reading* says: at fixed `σ`, the partition into ambient-thermal-bath heating versus durable downstream record is a physical variable predictable from the coupling structure of the system, not noise. - **Visible example.** Two identical batteries discharged at the same rate (matched `σ`) — one through a resistor (bath heating only), one through a memristor (heating plus a resistance-state record). Infotropy reads the memristor record as the record-side residue of the same `σ` that the resistor disposed entirely as heat. - **Fit verdict.** Substrate fit at C2 — `σ` is the natural coordinate for "irreversibility happened"; record formation is the dual axis Infotropy adds. - **Test or prediction.** `PRED-011` (decomposition); `PRED-006` (resource-theoretic separation). - **What would count against the Infotropy reading.** Record yield is independent of downstream coupling and tracks only `σ`. - **Source handles.** `SRC-080` (Seifert 2012 review); `SRC-086` (Dewar / Kleidon). Related: `ENT-002`; `ENT-007`; `REL-003`. ## ENT-007 — Stochastic-thermodynamic entropy production (Crooks / Jarzynski) - **What the field means by it.** For small driven systems, the probability of producing forward work `W` versus reverse work `-W` satisfies `P_F(W)/P_R(-W) = exp(β(W - ΔF))` (Crooks); equivalently `⟨exp(-βW)⟩ = exp(-βΔF)` (Jarzynski). These are exact equalities, not bounds. - **What side of the bottleneck it usually measures.** Trajectory-level asymmetry between a process and its time-reverse, in work distributions. - **Infotropy's flipped record-side reading.** The forward / reverse asymmetry IS record formation at the trajectory level. The *flipped record-side reading* says: trajectories with large `ΔS_tot` leave more durable record-side traces — conformational memory, residual orientation, isomer persistence — than low-`ΔS_tot` trajectories with matched endpoints. - **Visible example.** Single-molecule pulling of an RNA hairpin (Bustamante-class). The work distribution gives the Crooks/Jarzynski numbers; Infotropy asks whether trajectories with large `ΔS_tot` leave more conformational-memory residue in the unfolded state than low-`ΔS_tot` trajectories ending at the same length. - **Fit verdict.** Substrate fit at C2 — Sagawa–Ueda (`REL-007`) already extends this with explicit memory bookkeeping. - **Test or prediction.** `PRED-006` — multi-α F_α gap for work-extracting records. - **What would count against the Infotropy reading.** Trajectory-conditioned record-side variables are flat once endpoints are fixed. - **Source handles.** `SRC-069` (Crooks 1999); `SRC-070` (Jarzynski 1997); `SRC-080` (Seifert 2012). Related: `ENT-006`; `REL-006`; `REL-007`. ## ENT-009 — Configurational / mixing / residual entropy - **What the field means by it.** Entropy associated with the number of distinguishable arrangements of components — solute particles in a solvent, proton orderings in ice, defect configurations in a crystal — that remain at zero temperature when no further relaxation occurs. Pauling's classic ice-residual entropy is the canonical example. - **What side of the bottleneck it usually measures.** Counts of frozen-in equivalent configurations. - **Infotropy's flipped record-side reading.** Frozen-in configurations are durable records of the higher-temperature dynamics that produced them. The *flipped record-side reading* says: residual entropy is the record-side trace of an irreversibility that has already occurred; it is "the past, written into the material." Apparent residual entropy that cannot be coupled downstream (no detector can read it) is a candidate misclassification. - **Visible example.** Pauling's ice: the proton-ordering degeneracy in water ice gives `S ≈ k_B ln(3/2)` per mole-equivalent at `T → 0`. The proton positions are frozen records of the cooling history; NMR / dielectric experiments can read them downstream. - **Fit verdict.** Substrate fit at C2 — record formation is the system's dominant behavior; the field already treats this in record-like language without Infotropy needing to add machinery. - **Test or prediction.** No new `PRED-NNN`; the discriminator is "does the count match the physically distinguishable downstream states?" - **What would count against the Infotropy reading.** Residual entropy counts mismatch the physically distinguishable downstream states reproducibly across materials. - **Source handles.** `SRC-089` (Penrose / standard CMT texts; Pauling 1935 underlies). Related: `ENT-002`; `ENT-010`. ## ENT-010 — Glassy / frustrated-system entropy - **What the field means by it.** Entropy of the multiplicity of metastable basins in a rugged energy landscape: `Σ(f) = k_B ln N(f)`. The system never reaches a unique ground state; the entropy lives in how many "almost-ground-state" configurations exist and how the system explores them on a particular quench history. - **What side of the bottleneck it usually measures.** Counts of metastable basins; replica-symmetry-breaking parameters; aging and memory effects. - **Infotropy's flipped record-side reading.** Quench history is *literally* recorded as the basin the system ended in. The *flipped record-side reading* says: aging dynamics, memory effects, and rejuvenation are the system reading and re-reading its own quench-history record; the basin distribution `Σ(f)` is the record-formation phase space. - **Visible example.** A spin glass cooled twice with different protocols ends up in different metastable basins; "memory effects" experiments (Jonason et al. 1998 in CuMn) show the system remembers having been at one temperature even when cooled past it. The remembered history is the record. - **Fit verdict.** Substrate fit at C2 — and arguably at C3b regime-architecture; glassy systems are exemplary record-formation-dominated physics. - **Test or prediction.** Aging / rejuvenation experiments with quantified memory signatures regressed against landscape complexity. - **What would count against the Infotropy reading.** Memory effects are reproducibly uncorrelated with complexity estimates. - **Source handles.** Standard spin-glass texts (Mézard, Parisi, Virasoro 1987; Berthier–Biroli 2011). Related: `ENT-009`; `SRC-089`. ## ENT-013 — Differential entropy - **What the field means by it.** The continuous analogue of Shannon entropy: `h(X) = -∫ p(x) ln p(x) dx`. It is NOT coordinate-invariant: change variables and `h` shifts by a Jacobian log. Cover & Thomas warn the reader explicitly on this. - **What side of the bottleneck it usually measures.** Uncertainty / "spread" of a continuous distribution, in whatever coordinate system you happened to write `p` in. - **Infotropy's flipped record-side reading.** Coordinate-dependence is real pressure on any naive substrate-independent reading. The *flipped record-side reading* says: a record-side specification supplies the natural reference measure (the downstream-coupling measure), and the load-bearing object is then KL divergence to that reference (`REL-002`), not `h(X)` itself. Differential entropy is a pressure point, and the productive response is to route through `REL-002`. - **Visible example.** Position of a particle measured in meters versus inches: `h` shifts; KL to a uniform reference does not. The reference measure (e.g., the bin size of the detector that will read the position) is the record-side coordinate. - **Fit verdict.** Pressure at C2 — must be flagged, not promoted; route through KL. - **Test or prediction.** `PRED-003` — privileged coarse-graining via min-write-energy. - **What would count against the Infotropy reading.** No natural reference measure can be identified for a given record-side application; the diagnosis remains coordinate-dependent. - **Source handles.** `SRC-001` (Shannon 1948); Cover & Thomas 2006 ch. 8. Related: `ENT-005`; `REL-002`. ## ENT-014 — von Neumann entropy - **What the field means by it.** Quantum extension of Gibbs / Shannon: for a density operator `ρ`, the entropy is `S(ρ) = -Tr(ρ ln ρ)`. Reduces to Shannon when `ρ` is diagonal in a classical basis. Treated in Nielsen–Chuang ch. 11 as *the* quantum entropy. - **What side of the bottleneck it usually measures.** Quantum statistical uncertainty about a state; mixedness of `ρ`. - **Infotropy's flipped record-side reading.** The identity is C1 inherited; the record-side reading enters when you ask which `ρ` is the *reduced* state of a system whose environment has been traced out. The *flipped record-side reading* sends you to entanglement entropy (`ENT-016`) and quantum Darwinism (`REL-005`) — the partial trace IS the formal record-formation analog. - **Visible example.** A two-qubit Bell state; the von Neumann entropy of each single qubit is `ln 2` because the information about it has been "recorded" in the correlations with the other qubit. - **Fit verdict.** Compatible at C1 — the quantum extension of the formal identity; no Infotropy-distinctive content here alone. - **Test or prediction.** No distinctive test at this layer; route to `ENT-016`, `REL-005`. - **What would count against the Infotropy reading.** An operational anomaly refuting Hilbert-space quantum bookkeeping would refute quantum mechanics first. - **Source handles.** `SRC-017` (von Neumann 1932); `SRC-018` (Nielsen–Chuang 2010). Related: `ENT-015`; `ENT-016`; `INFO-041`. ## ENT-016 — Entanglement entropy - **What the field means by it.** For a bipartite quantum state `ρ_AB`, the entanglement entropy of subsystem A is the von Neumann entropy of its reduced density operator `ρ_A = Tr_B(ρ_AB)`. It is symmetric: `S(ρ_A) = S(ρ_B)` for a pure global state. - **What side of the bottleneck it usually measures.** Quantum correlation between A and B, measured as the mixedness induced in A by ignoring B. - **Infotropy's flipped record-side reading.** Partial trace IS the formal record-formation analog at the quantum level: tracing out B treats B as the bath / environment, and the kept system A inherits a record of its entanglement with B. The *flipped record-side reading* says: the symmetry `S(ρ_A) = S(ρ_B)` realizes "one process, two positions" cleanly; record-formation cuts are the physically distinguished cuts, and arbitrary Hilbert-space cuts are not. - **Visible example.** A ground state of a 1D quantum spin chain with an area-law correction: entanglement entropy of a contiguous region scales with the boundary, and that boundary IS the record-formation interface. - **Fit verdict.** Substrate fit at C2 — strong; one of the cleanest quantum-side record-formation rows. - **Test or prediction.** Cuts engineered through environment-discard channels behave qualitatively differently from arbitrary cuts on the same many-body state. - **What would count against the Infotropy reading.** No distinction between record-formation cuts and arbitrary Hilbert-space cuts is observable. - **Source handles.** `SRC-018` (Nielsen–Chuang 2010). Related: `ENT-014`; `REL-005`. ## ENT-017 — Quantum conditional entropy - **What the field means by it.** Defined as `S(A|B) = S(AB) - S(B)`. Unlike its classical counterpart, this can be **negative** — that negativity is one operational signature of entanglement. - **What side of the bottleneck it usually measures.** "Uncertainty about A given B" — but with the quantum twist that this quantity can be less than zero. - **Infotropy's flipped record-side reading.** Negative `S(A|B)` is genuine pressure on a naive C2 reading: "one process, different positions" sounds awkward if conditional entropy is negative. The *flipped record-side reading* sharpens C2: the relevant cut is a *record-formation cut* (a cut where one side actually does function as a record of the other), not an arbitrary Hilbert-space cut. After sharpening, the row is compatible — the negative values appear at cuts that are not record-formation cuts. - **Visible example.** A maximally entangled two-qubit Bell pair: `S(A|B) = -ln 2`. The "negative conditional entropy" reflects that B fully determines A — A is *over-determined* by B, which is not a record-formation situation but an entanglement situation. - **Fit verdict.** Pressure at C2 with a specific resolution: route record-formation talk through cuts where one side is a bath / environment, not arbitrary subspace splits. - **Test or prediction.** Theoretical sharpening note; experimental support via environment-discard cuts. - **What would count against the Infotropy reading.** A consistent record-formation cut is found that admits negative conditional entropy in cases relevant to Infotropy's reading. - **Source handles.** `SRC-018` (Nielsen–Chuang 2010). Related: `ENT-014`; `ENT-016`. ## ENT-019 — Smooth min- / max-entropy (Renner) - **What the field means by it.** Single-shot quantum entropies developed for cryptography and one-shot information theory: `H_max^ε` (compression ceiling) and `H_min^ε` (random-extraction floor) for finite-realization records, with `ε` a smoothing parameter. They are operationally meaningful in single-experiment QKD and randomness-extraction settings. - **What side of the bottleneck it usually measures.** Finite-realization extraction / compression bounds; not asymptotic. - **Infotropy's flipped record-side reading.** These entropies are *natively* about records: they govern how much true randomness you can extract from a finite-physical-source record, or how compactly you can encode a finite-physical state. The *flipped record-side reading* says: single-shot extraction bounds in record-rich versus record-poor settings should differ by an amount tied to write-cost asymmetry of the substrate. - **Visible example.** A QKD device extracts a key from `n` raw photon detections. `H_min^ε` tells you how many bits of true randomness survive after privacy amplification — i.e., how much of the physical record is record in the Infotropic sense versus correlated with the environment. - **Fit verdict.** Substrate fit at C2 — operational record-extraction native. - **Test or prediction.** Single-shot extraction bounds in record-rich versus record-poor settings differ by an amount tracking substrate write-cost. - **What would count against the Infotropy reading.** Write-cost asymmetry has no effect on smooth-entropy bounds. - **Source handles.** `SRC-016` (Renner 2005). Related: `ENT-014`; `ENT-018`. ## ENT-021 — Kolmogorov–Sinai (metric) entropy - **What the field means by it.** A measure-theoretic entropy of a measure-preserving dynamical system: the supremum over partitions of the average new information per time step. Bridges thermodynamic-style coarse-graining and dynamical-systems theory. - **What side of the bottleneck it usually measures.** Rate of new information per unit time in a deterministic-but-mixing dynamical system, relative to a partition. - **Infotropy's flipped record-side reading.** Partition entropy IS a coarse-graining specification (same family as `ENT-005`). The *flipped record-side reading* says: KS entropy bridges to thermodynamic entropy via Boltzmann-like cell counts, and the "natural" partition for a chaotic system is the one whose downstream coupling actually reads the trajectory — not an arbitrary geometric cover. - **Visible example.** A baker's-map dynamical system. KS entropy = `ln 2` per step relative to the natural bisecting partition. The natural partition coincides with what a downstream stretching-and-folding measurement would record. - **Fit verdict.** Substrate fit at C2 — partition-driven; aligned with `ENT-005`. - **Test or prediction.** KS estimates (via SampEn / permutation-entropy proxies) versus reconstruction quality on time-series across substrates. - **What would count against the Infotropy reading.** KS and record-grade orthogonal in well-controlled substrates. - **Source handles.** `SRC-030` (Sinai 1959); `SRC-032` (Walters 1982). Related: `ENT-005`; `ENT-022`; `ENT-031`. ## ENT-022 — Partition entropy (measure-theoretic) - **What the field means by it.** Shannon entropy of a measurable partition under a measure: `H_μ(α) = -Σ_A μ(A) ln μ(A)`. The basic building block of KS entropy. - **What side of the bottleneck it usually measures.** Probability mass distributed over a chosen partition. - **Infotropy's flipped record-side reading.** Equiprobable cells reduce to Boltzmann; this is the formal bridge from ergodic-theoretic entropy to statistical mechanics. The *flipped record-side reading* is the same as `ENT-005` / `ENT-021`: the partition IS the record-side coordinate. - **Visible example.** A coin-flip stream partitioned into "head" / "tail": `H_μ(α) = ln 2`. The partition IS the recording apparatus. - **Fit verdict.** Substrate fit at C2 — bridge row. - **Test or prediction.** Same as `ENT-005` / `ENT-021`. - **What would count against the Infotropy reading.** Same as `ENT-005` / `ENT-021`. - **Source handles.** `SRC-032` (Walters 1982). Related: `ENT-021`; `ENT-005`. ## ENT-026 — Logical entropy (Ellerman) - **What the field means by it.** Ellerman's partition-logic entropy: the probability that two independent draws from a distribution fall in distinct cells of a partition: `h = 1 - Σ p_i^2`. Reformulates Shannon entropy via distinction-counting in partition logic. - **What side of the bottleneck it usually measures.** Distinction count under partitioning. - **Infotropy's flipped record-side reading.** Distinction-counting IS record-formation in formal dress. The *flipped record-side reading* aligns logical entropy with Bateson's "difference that makes a difference" at the formal level — distinctions made are records made. - **Visible example.** A measurement apparatus with k bins applied to a stream: logical entropy counts the rate of pairwise distinct-bin pairs. The bins are the records. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Logical-entropy-based record-grade should correlate with Infotropic R-grade on partition-defined classification systems. - **What would count against the Infotropy reading.** No correlation between logical entropy and Infotropic R-grade. - **Source handles.** `SRC-035` (Ellerman 2009). Related: `ENT-022`; `INFO-015`. ## ENT-028 — Sofic / amenable group entropy - **What the field means by it.** Generalization of KS entropy to measure-preserving actions of sofic or amenable groups (Ornstein–Weiss; Lewis Bowen). Same conceptual content as `ENT-021` in a broader algebraic setting. - **What side of the bottleneck it usually measures.** Per-time-step new-information rate for broader group actions. - **Infotropy's flipped record-side reading.** Inherits `ENT-021`'s status. The *flipped record-side reading* says: wherever a measure-preserving action admits a coarse-graining, the partition is the record-side coordinate. - **Visible example.** A symbolic shift over a sofic group — e.g., a free-group action — admits a partition that records which group element acted; the entropy quantifies the rate of recorded distinctions. - **Fit verdict.** Substrate fit at C2 (weak — abstract). - **Test or prediction.** Same surface as `ENT-021`. - **What would count against the Infotropy reading.** Same as `ENT-021`. - **Source handles.** `SRC-031` (Bowen 1971); `SRC-032` (Walters 1982). Related: `ENT-021`. ## ENT-031 — Entropy rate of stationary process - **What the field means by it.** The per-symbol Shannon entropy of an infinite stationary stochastic process: `h = lim_n (1/n) H(X_1, ..., X_n)`. Equals KS entropy via symbolic dynamics. Central operational meaning of Shannon entropy. - **What side of the bottleneck it usually measures.** Per-symbol rate of new information / record production. - **Infotropy's flipped record-side reading.** Per-symbol record-production rate is exactly Infotropy's record-side coordinate at the stationary-process level. The *flipped record-side reading* sends you to predictive information (`INFO-028`): the sub-extensive scaling of `I_pred = I(X_past; X_future)` tracks the record-formation slope at boundaries of entropy rate. - **Visible example.** A binary stream from a Markov source. Entropy rate is the per-bit surprise; predictive information measures how much past data helps predict future data — the "carried record." - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Per-symbol estimates on bacterial-genome lineages versus randomization controls (within the C3c anchor). - **What would count against the Infotropy reading.** Predictive sub-extensive scaling does not correlate with record-formation. - **Source handles.** `SRC-001` (Shannon 1948); `SRC-032` (Walters 1982). Related: `ENT-021`; `INFO-002`; `INFO-028`. ## ENT-032 — Computational / HILL / pseudorandom entropy - **What the field means by it.** A distribution `X` has HILL entropy `≥ k` if it is computationally indistinguishable from some distribution `Y` with min-entropy `≥ k`. This is the operational entropy of choice in cryptography and complexity theory; it grades adversary-relative unpredictability, not physical irreversibility. - **What side of the bottleneck it usually measures.** Bounded-computation indistinguishability from a uniform source. - **Infotropy's flipped record-side reading.** This is a *productive falsifier*. A pseudorandom-generator output has full HILL entropy and zero physical record-formation cost; if Infotropy classified PRG outputs as records, the framework would be over-extended. The *flipped record-side reading* commits explicitly: "compression" in Infotropy means physical-irreversibility-bearing compression, not adversary-relative unpredictability. PRG outputs must fail Infotropic R-grade. - **Visible example.** A stream cipher's output: high HILL entropy, near-zero Landauer-floor cost, no durable physical record beyond the deterministic seed. Infotropy treats this as not-a-record despite the cryptographic entropy being maximal. - **Fit verdict.** Pressure at C2 — productive, sharpens the framework. - **Test or prediction.** Apply Infotropic R-grading to PRG outputs versus physical-noise sources; predict R-low for PRG, R-positive for physical noise of matched HILL entropy. - **What would count against the Infotropy reading.** PRG outputs score as Infotropic records on R-grade. - **Source handles.** `SRC-013` (Li-Vitanyi 2019) for the family; HILL paper (Håstad–Impagliazzo–Levin–Luby 1999) is the original. Related: `ENT-033`; `INFO-033`. ## ENT-038 — Variational free energy / surprisal (Friston FEP / active inference) - **What the field means by it.** Variational free energy is the surprisal of sensory data under an agent's generative model, formalized through the variational bound `F = E_q[ln q(s) - ln p(s,o)]`. Active inference proposes that agents act so as to minimize this quantity. It is formally Shannon-Bayesian at the variational layer; the controversial ontological layer claims this is *the* unified theory of brain function. - **What side of the bottleneck it usually measures.** Prediction error / variational-Bayesian surprise. - **Infotropy's flipped record-side reading.** The variational-Bayes layer is substrate-fit Shannon machinery and Infotropy reads it cleanly. The ontological FEP layer, however, makes implicit C3a-class claims about universal brain function that Infotropy does *not* endorse from definition compatibility. The *flipped record-side reading* says: agents with durable external records should outperform record-less agents on long-horizon non-stationary adaptation, with power-law scaling in record-persistence time — this discriminates "record-using surprise minimization" from "record-less surprise minimization." - **Visible example.** Two reinforcement-learning agents on a non-stationary Atari benchmark: one with a differentiable memory tape (DNC-class), one without. Infotropy predicts a power-law-with-record-persistence advantage for the first that vanishes when internal-state capacity is matched. - **Fit verdict.** Pressure at C2 — split: substrate-fit on variational layer; pressure on ontological layer. - **Test or prediction.** `PRED-008` — durable-external-record advantage. - **What would count against the Infotropy reading.** The record-persistence exponent collapses to zero when internal-state capacity is matched. - **Source handles.** `SRC-051` (Friston 2010). Related: `ENT-006`; `INFO-002`; `INFO-014`. ## INFO-001 — Hartley information / Renyi-α=0 max-entropy - **What the field means by it.** Hartley's 1928 measure: the information content of a choice from a set of `W` equally-likely alternatives is `log W`. The degenerate equiprobable case of Shannon entropy. - **What side of the bottleneck it usually measures.** Cardinality of a choice set. - **Infotropy's flipped record-side reading.** Hartley shares the `log W` formal core with Boltzmann; the C1 identity originates here. The *flipped record-side reading* is the same as `ENT-002`: the choice set IS a record-side specification of distinguishable outcomes. - **Visible example.** A 6-sided die gives `log 6` bits of choice information; the bits are records of which face came up, instantiated in whatever apparatus reads the die. - **Fit verdict.** Compatible at C1 — no Infotropy-distinctive content alone. - **Test or prediction.** No distinctive test; route to `INFO-002` / `ENT-002`. - **What would count against the Infotropy reading.** Same as `ENT-002`. - **Source handles.** `SRC-002` (Hartley 1928). Related: `INFO-002`; `ENT-002`; `ENT-023`. ## INFO-002 — Shannon entropy / Shannon information / source coding - **What the field means by it.** Expected surprisal of a source: `H(X) = -Σ p(x) log p(x)`. Shannon's source-coding theorem bounds the average code length needed for reliable transmission. Famously, Shannon's 1948 introduction explicitly brackets meaning as "irrelevant to the engineering problem." - **What side of the bottleneck it usually measures.** Uncertainty / surprise of a probabilistic source; equivalently, lossless code length. - **Infotropy's flipped record-side reading.** Shannon's setup presupposes a transmitter and a receiver — and the receiver IS a record. The *flipped record-side reading* says: receiver-side persistence is not in the formula but is in the engineering: a channel realization that includes a finite-memory durable receiver shows measurable deviation from the naive capacity bound, monotone in record write-cost and decay-rate. - **Visible example.** A flash-memory channel: writing bits has a finite physical cost and a finite retention time. Standard Shannon capacity treats the channel as memoryless; Infotropy says the write-cost asymmetry should leave a measurable residual in finite-blocklength capacity. - **Fit verdict.** Compatible at C1 — Infotropy does not own Shannon's theorem; the flipped reading lives in finite-receiver engineering. - **Test or prediction.** `PRED-001` — receiver-memory deviation from Shannon capacity. - **What would count against the Infotropy reading.** No residual capacity gap; gap is anti-correlated with write-cost. - **Source handles.** `SRC-001` (Shannon 1948). Related: `ENT-003`; `ENT-031`; `INFO-009`; `INFO-010`; `REL-001`. ## INFO-004 — Mutual information `I(X;Y)` - **What the field means by it.** Information shared between two variables: `I(X;Y) = H(X) + H(Y) - H(X,Y)`. Symmetric in X and Y; lacks temporal asymmetry. - **What side of the bottleneck it usually measures.** Statistical correlation; "what X tells you about Y." - **Infotropy's flipped record-side reading.** Mutual information is the cleanest formal correlate of "what survives downstream" — but it is symmetric, so it does not by itself encode record formation (records have a time arrow). The *flipped record-side reading* sends you to Sagawa–Ueda (`REL-007`): the mutual-information term in the generalized second law bookkeeps the demon's memory tape, which IS a record. - **Visible example.** A weather forecast and the next day's weather have nonzero `I(forecast; weather)`. The forecast is a record (it persists physically — on paper, in computer memory). Mutual information tells you how much of tomorrow's weather is in today's record. - **Fit verdict.** Substrate fit at C2 — under lineage acknowledgement; never direct support alone. - **Test or prediction.** `PRED-006` — Sagawa–Ueda mutual-information term tracks Infotropic R-grade in feedback systems. - **What would count against the Infotropy reading.** Sagawa–Ueda term and Infotropic R-grade orthogonal under controlled memory variation. - **Source handles.** `SRC-001` (Shannon 1948). Related: `INFO-002`; `REL-007`. ## INFO-006 — Channel capacity / channel-coding theorem - **What the field means by it.** The maximum rate at which information can be transmitted reliably across a noisy channel: `C = max_{p(x)} I(X;Y)`. Shannon's coding theorem says rates below `C` are achievable with arbitrarily low error. - **What side of the bottleneck it usually measures.** Maximum reliable transmission rate; receiver-side reconstructability under noise. - **Infotropy's flipped record-side reading.** "Reliable transmission" is record-side language *inside* Shannon's own theorems — the receiver reconstructs. The *flipped record-side reading* says: receivers with finite-memory durable records show effective capacity below the Shannon bound by an amount monotone in write-cost / decay-rate. The Shannon-bound idealizes a recordless receiver; the Infotropic correction is physical. - **Visible example.** A satellite-to-ground radio link with on-board flash storage for the receiver. The Shannon capacity is computed assuming instantaneous decoding; the flash-write cost adds a substrate-specific overhead the standard theorem does not see. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** `PRED-001` (shared with `INFO-002`). - **What would count against the Infotropy reading.** No residual capacity gap; gap anti-correlated with write-cost. - **Source handles.** `SRC-001` (Shannon 1948). Related: `INFO-002`; `INFO-004`. ## INFO-009 — Rate-distortion `R(D)` - **What the field means by it.** Shannon's rate–distortion function: the minimum bit-rate required to reconstruct a source `X` to within average distortion `D`. The native object is *reconstruction fidelity* under a chosen distortion measure. - **What side of the bottleneck it usually measures.** Reconstruction fidelity from compressed code; the downstream-reconstructive face of compression. - **Infotropy's flipped record-side reading.** Rate–distortion is *literally* a record-side definition: it is the minimum-rate that retains reconstructive capacity at distortion `D`. The *flipped record-side reading* says: `R(D)` curves serve as a calibration anchor for record-grade across substrate variation — record-rich substrates and record-poor substrates with the same nominal Shannon entropy should have different `R(D)` curves. - **Visible example.** Lossy JPEG compression: at a given quality factor, `R(D)` says how many bits per pixel are needed for the reconstructed image to stay within a target perceptual distortion. The reconstructed image is the record; `R(D)` is the record-formation cost. - **Fit verdict.** Direct support at C2 — reconstruction-fidelity native. - **Test or prediction.** `PRED-001` (shared) — `R(D)` and Infotropic R-grade should track each other across controlled substrate variation. - **What would count against the Infotropy reading.** `R(D)` and Infotropic R-grade orthogonal across substrates. - **Source handles.** `SRC-001` (Shannon 1948). Related: `INFO-010`; `INFO-002`. ## INFO-010 — Information bottleneck (Tishby) - **What the field means by it.** Find a compressed representation `T` of `X` that retains maximum predictive information about a relevance variable `Y`: minimize `I(X;T) - β I(T;Y)`. Introduced as a generalization of rate–distortion theory to representation learning. - **What side of the bottleneck it usually measures.** Compression toward the prediction-relevant subspace; the trade-off between information about `X` and information about `Y`. - **Infotropy's flipped record-side reading.** `T` is by construction a downstream record — a representation that *will be used* to predict `Y`. The *flipped record-side reading* says: the IB Lagrangian is the sharpest formal model of Infotropy's "compression-toward-relevant-record" reading among existing tools; bottleneck positions in deep networks should track Infotropic bottleneck predictions. - **Visible example.** A deep network trained on a classification task has internal representations that compress the input image while retaining the class label. The information bottleneck framework describes the layer-by-layer compression as IB curves; Infotropy says: each layer is a record-formation step. - **Fit verdict.** Direct support at C2 — Tishby-line lineage acknowledged. - **Test or prediction.** `PRED-001`; `PRED-010` (logical depth at bottlenecks). - **What would count against the Infotropy reading.** IB curves and Infotropic R-grade orthogonal on physically-grounded record-rich substrates. - **Source handles.** `SRC-015` (Tishby–Pereira–Bialek 1999). Related: `INFO-009`; `INFO-028`. ## INFO-013 — Floridi Strong Semantic Information (SSI) - **What the field means by it.** Floridi's claim that genuine semantic information must be **truthful** — false content does not count as information at all (alethic-neutrality of Shannon's `H` is explicitly rejected). - **What side of the bottleneck it usually measures.** Truth-conditioned semantic content; not a physical quantity. - **Infotropy's flipped record-side reading.** SSI imposes a constraint Infotropy's C2 reading does NOT assume. The *flipped record-side reading* says: physical records can be accurate, distorted, or misleading without ceasing to be records — a fossil of an asymmetric malformation is still a record. Infotropy's record-formation is *alethically neutral*; importing SSI silently would narrow the framework. - **Visible example.** A poorly-developed photograph captures the scene only approximately. SSI says it is "not information at all" if the developing was lossy enough; Infotropy says it is a record of both the scene and the lossy developing. - **Fit verdict.** Pressure at C2 — must be flagged, not absorbed. - **Test or prediction.** Theoretical reframe; identify record-formation events whose downstream content is demonstrably non-veridical and confirm record-formation dynamics hold. - **What would count against the Infotropy reading.** A Floridian-style truth-condition turns out to be load-bearing for physical record formation. - **Source handles.** `SRC-060` (Floridi 2004); `SRC-075` (SEP Information). Related: `INFO-012`; `INFO-015`. ## INFO-014 — Pragmatic information (Atmanspacher–Scheingraber; Weizsäcker) - **What the field means by it.** Information understood as *receiver-state-change*: pragmatic information = novelty × confirmation. A signal is informative to the extent that it changes the receiver's state in a way that uptake permits. - **What side of the bottleneck it usually measures.** Receiver-side state change at the moment of uptake. - **Infotropy's flipped record-side reading.** Receiver-state change IS record-formation at the cognitive / regulatory level. The *flipped record-side reading* says: receivers with durable record substrates uptake pragmatic information differently — and that difference should be measurable across substrates with manipulated record persistence. - **Visible example.** A bird's alarm call activates only the listening birds that don't already know about the predator. Pragmatic information lives in the change of those birds' state; Infotropy reads the lasting behavioral change as the record-side residue. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** `PRED-008` — durable-external-record agents outperform record-less agents. - **What would count against the Infotropy reading.** No pattern alignment between pragmatic uptake and record substrate. - **Source handles.** `SRC-064` (Atmanspacher–Scheingraber 1990). Related: `INFO-016`; `INFO-020`. ## INFO-017 — Dretske nomic information - **What the field means by it.** Dretske's account: a signal `S` carries the information that `p` iff there is a *nomic* (law-like) covariation between `S` and `p`. Information is fundamentally about lawful regularities between source and signal, not statistical correlation alone. - **What side of the bottleneck it usually measures.** Nomic covariation between source and signal; lawful tracking. - **Infotropy's flipped record-side reading.** Nomic covariation parallels Infotropy's "record-as-lawlike-trace" expectation. The *flipped record-side reading* says: pre-registered substrate covariation should predict downstream record-grade in controlled experiments — records of lawful processes should outperform records of mere correlations. - **Visible example.** A thermometer reading nomically covaries with ambient temperature *because* of the physical law governing mercury expansion. The mercury column is a record of the temperature because of the lawful coupling. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Cross-domain nomic-trace experiments (instrument-record fidelity versus noise). - **What would count against the Infotropy reading.** Nomic covariation and record-grade orthogonal across substrates. - **Source handles.** `SRC-061` (Dretske 1981); `SRC-075` (SEP Information). Related: `INFO-018`; `INFO-021`. ## INFO-018 — Documentary information (Buckland "information-as-thing"; Briet's antelope) - **What the field means by it.** "Information" in everyday and professional use names three things: a process (being informed), knowledge (what one comes to know), and a *thing* — a physical object treated as informative by an institution. Briet's antelope: when a zoo catalogs an antelope, it becomes a document. - **What side of the bottleneck it usually measures.** Physical / documentary record under institutional treatment. - **Infotropy's flipped record-side reading.** Documentary information is the cleanest *record-formation* native concept outside physics. The *flipped record-side reading* says: documentary records preserve reconstructive capacity under transmission noise differently from non-record data with the same Shannon content; the institutional act of recording is the bottleneck. - **Visible example.** A national archive of a 19th-century census record. The paper carries the information-as-thing; the institutional rules of preservation, copying, and consultation are the bottleneck topology. - **Fit verdict.** Substrate fit at C3b — civilizational regime rung. - **Test or prediction.** Archival fidelity experiments — digital corpora versus controlled noise injection versus reconstruction. - **What would count against the Infotropy reading.** Documentary records reconstruct no differently than non-record data with the same Shannon content. - **Source handles.** `SRC-055` (Buckland 1991); `SRC-057` (Briet 1951). Related: `INFO-019`; `INFO-020`. ## INFO-019 — Archival record / diplomatics (Duranti) - **What the field means by it.** The archival-science definition: a record has *authenticity*, *reliability*, *integrity*, and an *archival bond* to other records in its series. Duranti's diplomatics formalizes the record-as-event-trace at the institutional regime level. - **What side of the bottleneck it usually measures.** Institutional record corpora under explicit fidelity axioms. - **Infotropy's flipped record-side reading.** Diplomatics operationalizes record-formation at the civilizational regime rung. The *flipped record-side reading* says: the four axioms (authenticity, reliability, integrity, archival bond) ARE the institutional analog of physical record-formation conditions; Infotropy's regime architecture predicts they map onto persistence-spectrum / DRS-style invariants at this rung. - **Visible example.** A 13th-century English royal charter. The wax seal (authenticity), the formula (reliability), the unaltered text (integrity), and the cartulary it sits in (archival bond) together make it a record in the diplomatics sense. - **Fit verdict.** Substrate fit at C3b. - **Test or prediction.** Same surface as `INFO-018`. - **What would count against the Infotropy reading.** Same as `INFO-018`. - **Source handles.** `SRC-056` (Duranti 1998). Related: `INFO-018`. ## INFO-020 — Cybernetic information (Wiener) + Ashby's Requisite Variety - **What the field means by it.** Wiener's cybernetics: information drives control; the controller's state is a working record of the controlled system. Ashby's Law of Requisite Variety: only variety in the controller can destroy variety in the disturbance — i.e., the controller must have at least as much state-space as the disturbance it stabilizes. - **What side of the bottleneck it usually measures.** Control-system state capacity; regulation under perturbation. - **Infotropy's flipped record-side reading.** The controller state IS a record; requisite-variety is a record-capacity law. The *flipped record-side reading* says: controllers with higher record-capacity should sustain regulation against higher-variety disturbances, and that scaling is testable. - **Visible example.** A thermostat: its state (a record of recent temperature deviations) determines its regulation capacity. A bimetallic-strip thermostat has tiny record-capacity and regulates only against simple disturbances; a PID controller with integral memory has more. - **Fit verdict.** Substrate fit at C2 — mandatory lineage anchor. - **Test or prediction.** Engineered controllers with manipulated record substrates; regulation versus disturbance variety. - **What would count against the Infotropy reading.** Variety regulation is independent of controller record capacity. - **Source handles.** `SRC-027` (Wiener 1948); `SRC-028` (Ashby 1956). Related: `INFO-014`; `INFO-015`; `INFO-016`. ## INFO-021 — Peirce sign / object / interpretant (esp. index) - **What the field means by it.** Peirce's triadic sign relation: a sign stands for an object to an interpretant. Sub-typed by *icon* (resemblance), *index* (causal-existential trace), and *symbol* (convention). The **index** is the sign type defined by physical / causal trace. - **What side of the bottleneck it usually measures.** Semiotic relation in the indexical case — physical trace as sign. - **Infotropy's flipped record-side reading.** Peirce's *index* is the same ontological slot Infotropy's record-side names — a footprint, a weather-vane, a fossil. The *flipped record-side reading* says: indexical signs track downstream record-grade more reliably than iconic or symbolic signs in physical / instrumental settings. - **Visible example.** A bullet's striations on a recovered round (an index of the rifle barrel that fired it) versus a portrait of the rifle (an icon) versus the rifle's serial number (a symbol). Forensic reconstruction relies on the index because it is a physical record of the causal history. - **Fit verdict.** Substrate fit at C2 (index sub-type only) — lineage anchor in philosophy of signs. - **Test or prediction.** Comparative semiotics-of-instrument experiments: sensor records versus symbolic labels under noise. - **What would count against the Infotropy reading.** No differential record-grade tracking between indexical and non-indexical signs. - **Source handles.** `SRC-065` (Peirce Collected Papers). Related: `INFO-017`; `INFO-018`. ## INFO-025 — Genetic / sequence information (Crick) - **What the field means by it.** Crick's central-dogma framing: biological sequence information lives in the linear order of DNA / RNA / protein polymers. The polymer IS a record of inherited variation; transcription / translation read it. - **What side of the bottleneck it usually measures.** Sequence content of biopolymers; inheritance. - **Infotropy's flipped record-side reading.** This is the C3c anchor — the one substrate where bacterial-genome R(t) independence has been empirically confirmed. The *flipped record-side reading* says: the polymer is a durable record with downstream reconstructive capacity (the cell, the lineage); generalization to chloroplast / mitochondrial / viral lineages is the named next test. The framework does NOT claim universality from one domain (DNC-ONE-DOMAIN). - **Visible example.** A bacterial genome carrying both functional genes and pseudogenes — fossils of past selection. The pseudogenes are records of selection history that current selection is no longer reading; Infotropy's R(t) independence analysis distinguishes them. - **Fit verdict.** Direct support at C3c — the empirical anchor, with one-domain caveat. - **Test or prediction.** `PRED-018` — DRS generalization to chloroplast / mito / viral lineages. - **What would count against the Infotropy reading.** Functional forms differ significantly across lineage substrates after substrate-scaling. - **Source handles.** `SRC-040` (Crick 1970). Related: `INFO-026`; `INFO-030`. ## INFO-026 — Functional information (Hazen et al.) - **What the field means by it.** Information content of a configuration defined by `I_f = -log F(E_x)` where `F(E_x)` is the fraction of configurations meeting a functional threshold. A way of quantifying "functional rarity" in origin-of-life and biochemistry. - **What side of the bottleneck it usually measures.** Scarcity of functional configurations relative to the configuration space. - **Infotropy's flipped record-side reading.** Functional scarcity is the Boltzmann-side dual to record formation: most configurations are non-functional; the functional ones are the records of selection. The *flipped record-side reading* says: functional-information measures should correlate with Infotropic R-grade across origin-of-life and synthetic-biology substrates. - **Visible example.** A ribozyme. Most random RNA sequences of the same length are inert; the active ribozyme is a functional-information-rich configuration. Each functional ribozyme variant is a record of the selection trajectory that produced it. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Engineered protein / RNA functional-screen experiments — functional-information versus Infotropic R-grade. - **What would count against the Infotropy reading.** No correlation. - **Source handles.** `SRC-041` (Hazen et al. 2007). Related: `INFO-025`. ## INFO-027 — Spike-train mutual information (Strong et al.) - **What the field means by it.** Strong–Koberle–de Ruyter van Steveninck–Bialek 1998: the information rate carried by a neural spike train about a stimulus, measured as a mutual information per unit time over the spike train and the stimulus. - **What side of the bottleneck it usually measures.** Channel-capacity-style transmission from stimulus to spike pattern. - **Infotropy's flipped record-side reading.** Spike trains are *transient*; the durable record is in synaptic plasticity / structural plasticity. The *flipped record-side reading* says: plasticity-coupled records (LTP, structural changes) should track Infotropic R-grade differently than transient spike codes — the same mutual-information rate can mean very different record-formation outcomes. - **Visible example.** A retinal ganglion cell firing in response to a flashing light: the spike train carries ~1-2 bits per spike about the stimulus, but unless downstream plasticity laid down a record, the information is gone the moment the stimulus ends. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Neuroscience experiments with paired plasticity / non-plasticity learning protocols. - **What would count against the Infotropy reading.** No differential R-grade tracking with plasticity substrate. - **Source handles.** `SRC-044` (Strong et al. 1998). Related: `INFO-004`; `INFO-028`. ## INFO-028 — Predictive information (Bialek–Nemenman–Tishby) - **What the field means by it.** Mutual information between past and future of a stationary stochastic process: `I_pred = lim I(X_past; X_future)`. The *sub-extensive* scaling of this quantity identifies complexity classes; finite predictive information characterizes "regular" processes, divergent sub-extensive scaling characterizes "complex" ones. - **What side of the bottleneck it usually measures.** "How much of the past helps predict the future"; structural memory of a stationary process. - **Infotropy's flipped record-side reading.** Sub-extensive scaling IS record-pressure operationalized at the stationary-process level. The *flipped record-side reading* says: the sub-extensive scaling exponent should track Infotropic record-pressure across substrates with controlled record-formation rates; processes that lay down more durable records should show more predictive-information growth. - **Visible example.** Natural language: the predictive-information of English text grows sub-extensively with sample length; the growth slope encodes the "record structure" of grammar, idiom, and topic. Random shuffles show no growth. - **Fit verdict.** Direct support at C2 — close sibling of information bottleneck. - **Test or prediction.** Cross-substrate predictive-information scaling experiments. - **What would count against the Infotropy reading.** No tracking. - **Source handles.** `SRC-045` (Bialek–Nemenman–Tishby 2001). Related: `INFO-010`; `ENT-031`. ## INFO-029 — Efficient coding (Barlow / Atick) - **What the field means by it.** Sensory systems adapt to encode environmental statistics efficiently: maximize transmitted information per unit metabolic cost. Barlow's efficient-coding hypothesis is the canonical formulation in systems neuroscience. - **What side of the bottleneck it usually measures.** Information rate per metabolic cost; the channel-capacity face of biological perception. - **Infotropy's flipped record-side reading.** Adaptation is *multi-timescale record formation*: short-term gain adjustments, medium-term receptive-field changes, long-term structural plasticity. The *flipped record-side reading* says: engineered sensory systems with manipulated adaptation timescales should show Infotropic R-grade tracking adaptation depth. - **Visible example.** A fly's compound eye adapts gain to ambient light over milliseconds, receptive-field weights over hours, and developmental architecture over generations. Each timescale is a record-formation step that the efficient-coding objective shapes. - **Fit verdict.** Substrate fit at C2 / C3b — adaptation rung. - **Test or prediction.** Sensory-system experiments with controlled adaptation manipulations. - **What would count against the Infotropy reading.** Adaptation timescales are uncoupled from record-grade. - **Source handles.** `SRC-046` (Barlow 1961); `SRC-047` (Atick 1992). Related: `INFO-020`; `INFO-027`. ## INFO-030 — Epigenetic memory (Jablonka–Lamb) - **What the field means by it.** Heritable cellular state that is independent of DNA-sequence changes: methylation patterns, chromatin marks, cytoplasmic inheritance. Records on substrates other than DNA sequence. - **What side of the bottleneck it usually measures.** Non-sequence heritable state. - **Infotropy's flipped record-side reading.** Epigenetic memory multiplies the record-substrate dimensions Infotropy needs. The *flipped record-side reading* says: the DRS-decay form characterized for bacterial genomes (the C3c anchor) should hold on epigenetic substrates with substrate-specific scale parameters; this is the pressure-toward C3c generalization. - **Visible example.** An imprinted gene where which parent contributed the allele determines expression. The methylation mark IS the record of parental origin; reading it triggers downstream gene-regulatory consequences across the organism's lifespan. - **Fit verdict.** Direct support at C3c-pressure. - **Test or prediction.** `PRED-018` — DRS generalization (shared with `INFO-025`). - **What would count against the Infotropy reading.** No DRS-form fit on epigenetic substrates. - **Source handles.** `SRC-048` (Jablonka–Lamb 2014). Related: `INFO-025`; `INFO-031`. ## INFO-031 — Cultural / dual-inheritance information (Boyd–Richerson) - **What the field means by it.** Information transmitted across generations through learning, imitation, teaching, and institutions. Dual-inheritance modeling treats cultural inheritance as a second transmission channel alongside genetic inheritance, with its own selection and drift dynamics. - **What side of the bottleneck it usually measures.** Cultural transmission events; learned variation. - **Infotropy's flipped record-side reading.** Cultural inheritance is the same record story on non-genetic substrate — books, institutions, embodied skills, oral traditions. The *flipped record-side reading* says: cultural records should follow Infotropy's persistence-spectrum patterning; comparative case studies should show record-formation dynamics analogous to genetic / epigenetic substrates. Strictly descriptive only — Infotropy makes NO moral / normative claims here (DNC-NO-MORAL-IMPORT). - **Visible example.** A medieval cathedral's stonemason guild: techniques passed master-to-apprentice across centuries. The transmitted skill is the cultural record; the cathedral fabric is its physical residue. - **Fit verdict.** Direct support at C3b. - **Test or prediction.** Quantitative cultural-evolution studies on persistent versus transient cultural records. - **What would count against the Infotropy reading.** No analogous patterning between cultural and biological record formation. - **Source handles.** `SRC-049` (Boyd–Richerson 1985). Related: `INFO-030`; `INFO-019`. ## INFO-033 — Kolmogorov complexity `K(x)` - **What the field means by it.** The length of the shortest program (in a fixed universal language) that outputs the string `x`. Asymptotic, uncomputable, invariant up to an additive constant. The canonical algorithmic-information measure. - **What side of the bottleneck it usually measures.** Shortest description of a record; ideal compression. - **Infotropy's flipped record-side reading.** The native object IS a record-string; compression-ideal is the record-side measure. But `K(x)` does NOT distinguish "records that paid the Landauer-floor cost of their formation" from random-but-incompressible strings. The *flipped record-side reading* sends you to Bennett's logical depth (`INFO-038`): depth captures the "the record carries the work of its own production" intuition that `K` alone misses. - **Visible example.** Two strings of length `n`: one is a checksummed protein-folding trajectory, one is a fair coin-flip sequence. Both have `K(x) ≈ n` (both incompressible), but the protein trajectory has high logical depth (took non-trivial computation to produce) and the coin flips have minimal depth. - **Fit verdict.** Substrate fit at C2 — not direct support, since `K` alone cannot distinguish records from noise. - **Test or prediction.** `PRED-005` — depth-versus-K separation on record-rich strings. - **What would count against the Infotropy reading.** Compressibility tracks bottleneck-residue at least as well as depth. - **Source handles.** `SRC-010` (Kolmogorov 1965); `SRC-013` (Li-Vitanyi 2019). Related: `INFO-034`; `INFO-038`. ## INFO-034 — Prefix complexity (Levin / Chaitin) - **What the field means by it.** Kolmogorov complexity restricted to prefix-free programs: `K^prefix(x)` satisfies `Σ 2^{-K(x)} < ∞`. The version of `K` used to define algorithmic probability and Solomonoff induction. - **What side of the bottleneck it usually measures.** Same as `INFO-033`, with technical refinement. - **Infotropy's flipped record-side reading.** Same as `INFO-033` — the *flipped record-side reading* applies identically; logical depth is the differentiator. Prefix complexity is included separately because the cryptographic / Solomonoff literature treats it as distinct. - **Visible example.** Same family as `INFO-033`. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Same as `INFO-033` (`PRED-005`). - **What would count against the Infotropy reading.** Same as `INFO-033`. - **Source handles.** `SRC-012` (Chaitin 1975); `SRC-013` (Li-Vitanyi 2019). Related: `INFO-033`. ## INFO-035 — Algorithmic mutual information - **What the field means by it.** Algorithmic analog of mutual information: `I_alg(x : y) = K(x) + K(y) - K(x, y)`. Record-to-record information without a probability model. - **What side of the bottleneck it usually measures.** Shared algorithmic content between two records. - **Infotropy's flipped record-side reading.** Same as `INFO-033` / `INFO-034`. The *flipped record-side reading* says: algorithmic mutual information picks up shared record content directly; coupled with depth, it discriminates record-pairs that share formation history from record-pairs that share only structure. - **Visible example.** Two bacterial genomes from related strains: their algorithmic mutual information should track the recency of common ancestry; modulated by depth, this distinguishes "shared structural noise" from "shared selection history." - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** Routed via `PRED-005` / `PRED-010`. - **What would count against the Infotropy reading.** Same as `INFO-033`. - **Source handles.** `SRC-013` (Li-Vitanyi 2019). Related: `INFO-004`; `INFO-033`. ## INFO-038 — Logical depth (Bennett) - **What the field means by it.** Bennett's logical depth `D_s(x)`: the minimum computation time required by any program of length within `s` of `K(x)` to produce `x`. Captures "value" — the work a record encodes that is not captured by raw compressibility. Stable under additive `s` (the "significance level"). - **What side of the bottleneck it usually measures.** Computation cost of reconstructing a record from a near-minimal description. - **Infotropy's flipped record-side reading.** Logical depth is the closest existing formal anchor to Infotropy's "durable record with downstream reconstructive capacity." The *flipped record-side reading* says: bottleneck positions across substrates should score higher on depth proxies than non-bottleneck baselines; depth elevates the records that "carry the work of their formation" and demotes both random noise and trivial patterns. - **Visible example.** A peer-reviewed paper versus a random shuffle of its tokens. Both might have similar `K` (both essentially incompressible at character level), but only the paper has high logical depth because the actual content took non-trivial work to compose. Infotropy reads the paper as a higher-depth record. - **Fit verdict.** Direct support at C2 — with Bennett lineage acknowledged. - **Test or prediction.** `PRED-005` (depth-versus-K separation); `PRED-010` (logical depth at record bottlenecks). - **What would count against the Infotropy reading.** Depth flat across position after controlling compressibility on record-rich substrates. - **Source handles.** `SRC-009` (Bennett 1988); `SRC-013` (Li-Vitanyi 2019). Related: `INFO-033`; `INFO-039`. ## INFO-039 — Sophistication / effective complexity (Gell-Mann–Lloyd) - **What the field means by it.** A measure of the information content of the *structural* component of a string, separated from the random component. Sophistication is the length of a minimal program for the "regular" part of the string, with the random part treated as a parameter. - **What side of the bottleneck it usually measures.** Structural content distinct from noise. - **Infotropy's flipped record-side reading.** The structural component is the record content; the random component is the bath. The *flipped record-side reading* says: sophistication tracks the record-formation side of compression, parallel to logical depth — useful diagnostic, with the same direct-support logic at C2. - **Visible example.** A photo of a cathedral has high sophistication (the structural pattern of the cathedral, plus random noise from the sensor); a photo of TV static has high `K` but low sophistication. - **Fit verdict.** Substrate fit at C2 — sibling of `INFO-038`. - **Test or prediction.** Same surface as `INFO-038`. - **What would count against the Infotropy reading.** Same as `INFO-038`. - **Source handles.** `SRC-081` (Gell-Mann–Lloyd 1996). Related: `INFO-038`. ## INFO-042 — Holevo information / bound - **What the field means by it.** The upper bound on classical information extractable from a quantum ensemble: `χ = S(Σ p_i ρ_i) - Σ p_i S(ρ_i)`. Bridges quantum states to classical records. - **What side of the bottleneck it usually measures.** Classical-record-extraction bound from a quantum source. - **Infotropy's flipped record-side reading.** Holevo IS a record-formation bound at the quantum-classical interface. The *flipped record-side reading* says: Holevo should be saturated under quantum-Darwinism-fragmented environments (`REL-005`); the redundancy plateau is the operational signature of record formation. - **Visible example.** A quantum-key-distribution protocol's classical bit rate: bounded above by the Holevo information of the ensemble of quantum states Alice prepares; the actual key is a classical record extracted at that bound. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** `PRED-007` — substrate-independent quantum-Darwinism plateau. - **What would count against the Infotropy reading.** No saturation pattern. - **Source handles.** `SRC-084` (Holevo 1973); `SRC-018` (Nielsen–Chuang 2010). Related: `INFO-041`; `INFO-043`; `REL-005`. ## INFO-043 — Quantum channel capacities (HSW / LSD / private) - **What the field means by it.** Asymptotic rates at which classical, quantum, or private information can be reliably transmitted through a noisy quantum channel (Holevo–Schumacher–Westmoreland for classical, Lloyd–Shor–Devetak for quantum, Cai–Winter–Yeung for private). - **What side of the bottleneck it usually measures.** Reliable transmission rate; quantum extension of `INFO-006`. - **Infotropy's flipped record-side reading.** Same as `INFO-006` at quantum scale: receiver-side persistence is not in the asymptotic formula but is in the engineering. The *flipped record-side reading* sends you to the same prediction surface: finite-memory durable records cap effective capacity below the asymptotic rate. - **Visible example.** A quantum repeater chain: ideal-capacity calculations assume noiseless intermediate memory; real quantum memories have finite coherence times that physically record the channel's noise history. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** `PRED-001` at quantum scale. - **What would count against the Infotropy reading.** Same as `INFO-006`. - **Source handles.** `SRC-087` (Holevo 1998; Schumacher–Westmoreland 1997); `SRC-018` (Nielsen–Chuang 2010). Related: `INFO-006`; `INFO-042`. ## INFO-044 — Quantum discord (Ollivier–Zurek) - **What the field means by it.** Difference between two classical-analog definitions of mutual information in a bipartite quantum state: `δ(A;B) = I(A:B) - J(A:B)`. Nonzero discord marks irreducibly quantum correlations that cannot be captured by classical measurements on one side. - **What side of the bottleneck it usually measures.** Excess of quantum correlation over classical-record-accessible correlation. - **Infotropy's flipped record-side reading.** Discord *natively* splits classical-record-accessible correlations from quantum-only correlations. The *flipped record-side reading* says: discord-vanishing transitions should track record-formation thresholds across substrates — the moment a quantum correlation becomes classically extractable is the moment it becomes a record. - **Visible example.** A two-qubit system as it decoheres: discord starts high (quantum correlations), then drops as the environment monitors one party, then plateaus at the classical mutual information. The plateau IS record formation in the quantum-Darwinism sense. - **Fit verdict.** Substrate fit at C2. - **Test or prediction.** `PRED-007` — substrate-independent plateau (shared with `INFO-042` / `REL-005`). - **What would count against the Infotropy reading.** No tracking of discord-vanishing with record-formation threshold. - **Source handles.** `SRC-083` (Ollivier–Zurek 2001). Related: `INFO-041`; `REL-005`. ## REL-001 — Shannon–Boltzmann formal identity - **What the field means by it.** The algebraic fact that Gibbs entropy and Shannon entropy share the `-Σ p log p` form (up to `k_B`). Jaynes 1957 made the identity foundational for the maxent program. The identity itself is uncontroversial; its physical interpretation is contested. - **What side of the bottleneck it usually measures.** Formal mathematical identity. - **Infotropy's flipped record-side reading.** This is the C1 anchor. The identity itself is not new evidence; reading it as "one process, different positions relative to constraint / record formation" is the C2 interpretive synthesis Infotropy adds. The *flipped record-side reading* says: the SAME formal object measures both dispersion (in physics) and uncertainty (in information theory) because both are the dispersion-side of the same kind of irreversible step, with record formation as the dual coordinate. - **Visible example.** Jaynes' classic derivation: starting from "we recorded the mean energy" (a record-side constraint), the canonical distribution `p_i ∝ exp(-βE_i)` falls out by maximizing Shannon entropy under that constraint — yielding the canonical statistical mechanics simultaneously. - **Fit verdict.** Compatible at C1 — inherited identity; do NOT promote to discovery. - **Test or prediction.** Consistency check across canon; no Infotropy-distinctive test. - **What would count against the Infotropy reading.** A counterexample to either Shannon or Gibbs frameworks would refute the identity first. - **Source handles.** `SRC-001` (Shannon 1948); `SRC-004` (Boltzmann 1877); `SRC-005` (Gibbs 1902); `SRC-006` (Jaynes 1957). Related: `ENT-002`; `ENT-003`; `INFO-002`. ## REL-002 — KL divergence / cross-entropy - **What the field means by it.** Kullback–Leibler divergence: asymmetric "distance" between two probability distributions, `D_KL(p || q) = Σ p log(p/q)`. Cross-entropy: `H(p, q) = H(p) + D_KL(p || q)`. Coordinate-invariant (unlike differential entropy). - **What side of the bottleneck it usually measures.** Distinguishability between distributions; inefficiency of coding `p` with `q`'s code. - **Infotropy's flipped record-side reading.** KL to a reference equilibrium IS dissipation in stochastic thermodynamics. The *flipped record-side reading* says: where differential entropy fails (coordinate-dependence), KL to a downstream-coupling-defined reference is the load-bearing object; the reference measure IS supplied by the record-side specification. - **Visible example.** Bayesian updating: the KL divergence between the prior and posterior measures how much the new data records about the parameters. In physics: KL between a nonequilibrium state and its equilibrium fixed point measures available free energy and dissipation. - **Fit verdict.** Substrate fit at C2 — the canonical coordinate-free reframe. - **Test or prediction.** Cross-substrate dissipation-versus-KL with controlled reference measures. - **What would count against the Infotropy reading.** KL-to-reference and physical dissipation orthogonal across substrates. - **Source handles.** `SRC-001` (Shannon 1948); `SRC-013` (Cover & Thomas / Li-Vitanyi). Related: `ENT-013`; `INFO-002`; `ENT-015`. ## REL-003 — Landauer's principle - **What the field means by it.** Erasing one logical bit of information dissipates at least `k_B T ln 2` of heat into the environment. Landauer 1961; Bennett 1982; experimentally confirmed by Berut et al. 2012 and Jun–Gavrilov–Bechhoefer 2014. The principle is the gold-standard bridge between record manipulation and thermodynamic cost. - **What side of the bottleneck it usually measures.** Thermodynamic cost of a logical (record) operation. - **Infotropy's flipped record-side reading.** Landauer is *named in* the Infotropy public-story framing. The *flipped record-side reading* extends Landauer from erasure to *formation*: in any record-formation event, free energy is converted into durable structure at cost `≥ k_B T ln 2` per distinguishable record state established. Erasure and formation are dual operations on the same record. - **Visible example.** A flash memory cell: writing a bit costs energy that ultimately appears as heat — Landauer's principle (plus engineering inefficiencies) sets the floor. Erasing the bit costs at least `k_B T ln 2` in the same currency. Each write and each erase IS a record-side operation with a thermodynamic cost. - **Fit verdict.** Direct support at C2 — the strongest physics-side row, with Earman & Norton 1999 / Maroney 2009 caveats kept visible. - **Test or prediction.** `PRED-006`; `PRED-011`. Extend Berut-class experiments to formation-side as well as erasure-side bookkeeping. - **What would count against the Infotropy reading.** Record formation exhibits no thermodynamic cost lower bound related to its distinguishability count. - **Source handles.** `SRC-007` (Landauer 1961); `SRC-008` (Bennett 1982); `SRC-073` (Earman–Norton 1999); `SRC-074` (Maroney SEP); `SRC-076` (Berut 2012); `SRC-077` (Jun-Gavrilov-Bechhoefer 2014). Related: `REL-004`; `REL-007`; `ENT-002`. ## REL-004 — Maxwell's demon / Szilard engine - **What the field means by it.** Maxwell's hypothetical agent that uses information about microstates to extract work from a thermal bath, apparently violating the Second Law. Szilard's one-bit engine: extracts `k_B T ln 2` per cycle, balanced by the Landauer cost of memory erasure. Resolved by Bennett 1982; experimentally explored in Toyabe et al. 2010 and Koski et al. 2014. - **What side of the bottleneck it usually measures.** Work extracted from a thermal bath using information; the demon's record is the catalyst. - **Infotropy's flipped record-side reading.** The demon's memory IS the record. The *flipped record-side reading* says: in feedback-controlled Maxwell-demon-realised experiments, extracted work should depend on the *persistence-grade* of the demon's memory in a graded way — not just on whether erasure has happened, but on how durably the record was held. - **Visible example.** Koski's single-electron-box experiment: a real demon (digital feedback controller) extracts work from electrical fluctuations using bit-by-bit memory. The memory's actual physical persistence sets the achievable work. - **Fit verdict.** Substrate fit at C2 — foundational thought-experiment cluster. - **Test or prediction.** Maxwell-demon experiments with controlled-decay memories rather than perfect erase/persist binary. - **What would count against the Infotropy reading.** Demon performance is insensitive to memory persistence grade once threshold reliability is met. - **Source handles.** `SRC-008` (Bennett 1982); `SRC-072` (Maruyama–Nori–Vedral 2009 review); `SRC-073` (Earman–Norton 1999). Related: `REL-003`; `REL-007`. ## REL-005 — Quantum Darwinism / einselection (Zurek) - **What the field means by it.** Zurek's framework: decoherence selects a preferred "pointer basis"; the environment then redundantly proliferates information about those pointer states across many environmental fragments. Multiple independent observers reading different fragments agree on the same classical record. This is the explicit "how classicality emerges from quantum mechanics" story in modern quantum foundations. - **What side of the bottleneck it usually measures.** Redundancy of pointer-state information across environment fragments — the redundancy plateau in `I(S : F)` versus `|F|`. - **Infotropy's flipped record-side reading.** Quantum Darwinism is the **canonical native record-formation theory in physics**. The *flipped record-side reading* identifies redundant environmental records as Infotropy's record-formation events; the redundancy plateau IS the record-formation signature. Calling this direct_support would inflate (it gives C2 a physical substrate, not C3a evidence) — substrate_fit is the honest verdict. - **Visible example.** Experiments with NV centers in diamond (Unden et al. 2019) measure the redundancy plateau directly: the diamond's nuclear-spin environment carries many copies of the central spin's pointer state, and any small subset of nuclear spins suffices to read the pointer. - **Fit verdict.** Substrate fit at C2 — strongest single-substrate quantum-side anchor. - **Test or prediction.** `PRED-007` — substrate-independent plateau coefficient when normalized for decoherence rate. - **What would count against the Infotropy reading.** Plateau coefficient differs >2σ across substrates after decoherence-rate control. - **Source handles.** `SRC-019` (Zurek 2003); `SRC-020` (Zurek 2009); `SRC-078` (Unden 2019); `SRC-079` (Chen 2019). Related: `ENT-016`; `INFO-042`; `INFO-044`. ## REL-006 — Resource-theoretic thermodynamics (Brandão–Horodecki–Ng–Oppenheim–Wehner) - **What the field means by it.** Single-shot second laws for quantum systems formulated as a family of α-Renyi free-energy inequalities. Work storage is a co-formal record substrate in this framework; catalysts are recorded auxiliary systems. - **What side of the bottleneck it usually measures.** Single-shot bounds on state interconvertibility under thermal operations. - **Infotropy's flipped record-side reading.** Work-extracting records are distinguished from non-record information-bearing states via the F_α spectrum gap. The *flipped record-side reading* says: this distinction IS the record-formation signature at the resource-theoretic level — Infotropy aligns with this prior art and reads its spectrum gap as the operational record-grade. - **Visible example.** A trapped-ion qubit acting as both a system and its own memory tape: resource-theoretic protocols can extract work using the memory; the F_α spectrum tracks how much work as a function of the record's distinguishability. - **Fit verdict.** Substrate fit at C2 — lineage-anchored to Brandão et al. - **Test or prediction.** `PRED-006` — multi-α F_α gap for work-extracting records. - **What would count against the Infotropy reading.** F_α collapses to F_1 once mutual information is matched. - **Source handles.** `SRC-021` (Brandão et al. 2015). Related: `REL-003`; `REL-007`. ## REL-007 — Sagawa–Ueda generalized second law - **What the field means by it.** Generalized second law for feedback-controlled systems: `⟨W_extracted⟩ ≤ -ΔF + k_B T · I(X; Y)`. Explicitly bookkeeps the mutual information acquired about the system on a memory tape. Bennett 1973 → Sagawa–Ueda 2008–2012; experimentally explored in Toyabe 2010, Koski 2014, etc. - **What side of the bottleneck it usually measures.** Work-information trade-off in feedback systems; the modern bookkeeping of Maxwell-demon-style information-thermodynamics. - **Infotropy's flipped record-side reading.** The mutual-information term `I(X; Y)` IS the record on the demon's memory tape. The *flipped record-side reading* says: Infotropic R-grade should track this term in feedback-controlled systems — they are the same record-formation coordinate viewed from different angles. - **Visible example.** Koski's single-electron-box experiment running a Sagawa–Ueda-style protocol: the measured `I(X; Y)` directly bounds the extracted work; Infotropy reads the same `I(X; Y)` as the record-grade of the bit-state observation. - **Fit verdict.** Direct support at C2 — lineage to Bennett-Sagawa-Ueda acknowledged. - **Test or prediction.** `PRED-006`. - **What would count against the Infotropy reading.** Sagawa–Ueda term and Infotropic R-grade orthogonal under controlled memory variation. - **Source handles.** `SRC-022` (Sagawa-Ueda 2012); `SRC-080` (Seifert 2012 review). Related: `REL-003`; `REL-004`; `REL-006`. ## REL-012 — Maximum-entropy principle (Jaynes) - **What the field means by it.** Among probability distributions consistent with given constraints, choose the one of maximum Shannon entropy. Jaynes 1957 grounded this as the principle of statistical inference and used it to derive equilibrium statistical mechanics from information theory. - **What side of the bottleneck it usually measures.** Inference / prior assignment under constraints. - **Infotropy's flipped record-side reading.** Maxent is the bridge between Bayesian inference and statistical mechanics — and the constraint set IS a record-side specification. The *flipped record-side reading* says: explicit record-formation constraints yield maxent priors that outperform arbitrary moment-based maxent on downstream-prediction tasks; this is one of the mandatory C2 lineage anchors. - **Visible example.** A Bayesian who has *recorded* only the mean energy of a gas, applies maxent, and obtains the canonical Maxwell–Boltzmann distribution. The recorded constraint IS the record-side coordinate that selected this distribution out of all possibilities. - **Fit verdict.** Substrate fit at C2 — mandatory lineage anchor (Jaynes). - **Test or prediction.** Maxent under record-side-motivated constraints versus arbitrary moment constraints. - **What would count against the Infotropy reading.** Record-side-motivated constraints provide no predictive advantage. - **Source handles.** `SRC-006` (Jaynes 1957). Related: `ENT-003`; `ENT-005`; `INFO-002`. --- # Part II — Tier B entries (short treatment) Tier B rows are compatible / neutral measures included so a reader understands why Infotropy is not tied to any single local entropy or information measure. Each is named with native meaning, what side it measures, and where it routes within the matrix. ## ENT-004 — Ensemble entropies (microcanonical / canonical / grand-canonical) Specialisations of Gibbs entropy under fixed `(E, V, N)`, `(T, V, N)`, or `(T, V, μ)`. Measures equilibrium statistical weights under chosen constraints. Compatible at C1; the chosen ensemble IS a record-side choice but ensemble inequivalence (small-`N`, mesoscopic regimes) is where the field's standard treatment would benefit from explicit record-side bookkeeping. Visible example: gas in a sealed box (microcanonical) vs in contact with a thermal reservoir (canonical) — same physics, different record-side specifications. Route to `ENT-005` for the partition-dependence reading. Source: `SRC-005`, `SRC-089`. ## ENT-008 — Onsager linear-response Reciprocal relations near equilibrium: `L_ij = L_ji`. Measures near-equilibrium flux–force coupling. Compatible at C2 — this is a regime where record formation is suppressed (the linear response treats the system as memory-less by assumption); useful as a null baseline for far-from-equilibrium Infotropic predictions. Visible example: heat conduction in a uniform metal bar — the Fourier law is the Onsager regime. Source: `SRC-071`. ## ENT-015 — Quantum relative entropy Quantum extension of KL: `S(ρ || σ) = Tr(ρ ln ρ) - Tr(ρ ln σ)`. Measures quantum-state distinguishability; record-side enters via Stein's lemma in asymptotic hypothesis testing. Compatible at C2 — operator-algebraic; carries the same reframing as `REL-002`. Visible example: distinguishing two quantum states by repeated measurement; the asymptotic error exponent IS the relative entropy. Source: `SRC-085`. ## ENT-018 — Quantum Rényi entropies Parametric family `S_α(ρ) = (1 - α)^{-1} ln Tr(ρ^α)`; converges to von Neumann at `α = 1`. Operational extremes (`α → 0`, `α → ∞`) connect to single-shot bounds (`ENT-019`) and resource-theoretic thermodynamics (`REL-006`). Compatible at C2. Visible example: tail probabilities in single-shot hypothesis testing rely on Rényi-α at non-unit α. Source: `SRC-033`, `SRC-018`. ## ENT-023 — Renyi entropy family (classical) Parametric family `H_α = (1 - α)^{-1} ln Σ p_i^α`; Shannon at α=1, Hartley at α→0, min-entropy at α→∞. Compatible at C2; only α=1 anchors the C1 identity. Off-α members are generalizations and not themselves Infotropic without further argument. Visible example: collision entropy `H_2` used in leftover-hash-lemma randomness extraction. Source: `SRC-033`. ## ENT-029 — Permutation entropy (Bandt–Pompe) Shannon entropy of the distribution of ordinal patterns in a time series: `H_p = -Σ p(π) ln p(π)`. Compatible at C2 / C4; useful KS-proxy diagnostic for time series. Visible example: EEG signal characterization in neuroscience — permutation entropy distinguishes seizure from non-seizure activity. Source: `SRC-036`. ## ENT-030 — Sample / approximate / multiscale entropy ApEn, SampEn, MSE — empirical irregularity measures over physiological time series. Compatible at C4 (diagnostic). False-friend risk in biomedical literature when over-interpreted. Visible example: heart-rate-variability analysis where SampEn distinguishes healthy from pathological signals. Source: `SRC-037`, `SRC-038`, `SRC-039`. ## INFO-003 — Conditional entropy `H(Y|X)` Average remaining surprisal of `Y` after `X` is known: `H(Y|X) = H(X, Y) - H(X)`. Compatible at C2 — distribution-side bookkeeping; routes to `INFO-002`. Visible example: how much uncertainty about tomorrow's weather remains after seeing the forecast. Source: `SRC-001`. ## INFO-005 — Conditional mutual information `I(X; Y | Z)` Bridge measure: `I(X; Y | Z) = H(X|Z) + H(Y|Z) - H(X, Y | Z)`. Compatible at C2 — not a definition family of its own; routes to `INFO-004`. Visible example: detecting causality with conditional independence tests. Source: `SRC-001`. ## INFO-007 — Fisher information Sensitivity of a likelihood to a parameter: `I_F(θ) = E[(∂ ln p(x|θ)/∂θ)²]`. Compatible at C2 — parameter sensitivity, *not* record formation. False-friend risk on the word "information." Visible example: Cramér–Rao bound for parameter estimation. Source: standard statistics texts. ## INFO-008 — Bayesian evidence / log-evidence Marginal likelihood `ln p(D | M)`. Compatible at C2 — implicit record substrate (priors carry past inference). Routes to active-inference predictions (`PRED-008`). Visible example: Bayesian model comparison with the Bayes factor. Source: standard Bayesian texts. ## INFO-012 — Floridi General Definition of Information (GDI) "Information = well-formed, meaningful data." Compatible at C2 — modest metaphysical commitment that does not impose truthfulness (unlike SSI). Visible example: a sensor reading: bytes that are well-formed (correct frame) and meaningful (interpretable by downstream consumer). Source: `SRC-059`, `SRC-075`. ## INFO-015 — Bateson "difference that makes a difference" A difference (in the world) that produces a downstream difference (in a system). Compatible at C2 with **HARD warning** per BND-PHIL-BATESON-DIFF: do NOT collapse Bateson's "difference" into Shannon `H`, Wheeler's "bit", or thermodynamic Δ-entropy. Bateson is dimensionless and non-energetic by his own statement. Visible example: a tick releasing on butyric acid: the molecule's presence is a "difference" only when uptaken by the tick's receptor and producing a behavioral change. Source: `SRC-054`. ## INFO-016 — MacKay information-as-distinction Selective vs structural information: receiver-relative distinction-among-alternatives. Compatible at C2 — cousin of Bateson. Visible example: identifying a face in a crowd is a selective information event; characterizing the dimensions of "face space" is structural. Source: `SRC-062`. ## INFO-023 — Speech-act content (Austin–Searle) Illocutionary force and propositional content. Compatible at C2 / C3b — communicative acts leave institutional records. Visible example: a contract signed at law: the illocutionary act of promising creates an institutional record that downstream legal systems read. Source: `SRC-067`. ## INFO-036 — Algorithmic probability / Solomonoff Universal a-priori probability of a string under a prefix-free universal machine: `M(x) = Σ 2^{-|p|}` over programs `p` outputting `x`. Compatible at C4 — universal Bayesian ideal. Visible example: ideal inductive inference: Solomonoff prediction would converge to optimal but is uncomputable. Source: `SRC-011`. ## INFO-037 — Minimum description length (MDL) Operational compression criterion for model selection: minimize `|model| + |data given model|`. Compatible at C4. Visible example: choosing the right polynomial degree in regression: MDL penalizes models whose parameters cost more bits than they save in describing the data. Source: `SRC-014`. ## INFO-041 — Quantum mutual information `I(A:B)` Quantum analog of mutual information. Compatible at C2 — correlation count; becomes record-side under environment fragmentation (route to `REL-005`). Visible example: bipartite entanglement: the quantum mutual information measures the total quantum + classical correlation, with `I(A:B) = 2 S(A)` for a pure entangled state. Source: `SRC-018`. ## REL-008 — Consistent histories / Wigner's friend Quantum mechanics admits multiple consistent narratives of the same process; observer-relativity of records is live. Compatible at C2 — compatible with "one process, different positions," but a live philosophical contest; do not cite as evidence. Visible example: the Frauchiger–Renner thought experiment: different observers can reach contradictory yet locally consistent records. Source: `SRC-019`. ## REL-011 — Information bottleneck (bridge pointer) Cross-lane bridge marker. Full row at `INFO-010`. Pointer here because IB bridges rate-distortion (`INFO-009`), predictive information (`INFO-028`), and learned-representation theory in deep networks. Source: `SRC-015`. ## REL-013 — Variational principle in ergodic theory (`h_top = sup_μ h_μ`) Topological entropy equals the supremum over invariant measures of metric (KS) entropy. Compatible at C2 — theorem bridge from non-record-bearing topological entropy to record-bearing KS entropy; relevant for the false-friend audit on topological entropy (see Pile entry). Visible example: a smooth diffeomorphism — `h_top` can be achieved by a specific invariant measure that distinguishes physically meaningful trajectories from arbitrary covers. Source: `SRC-032`. ## REL-014 — Autopoiesis (Maturana–Varela) A self-producing organizational closure that distinguishes a unity from its environment; closure as record-formation mechanism at organism level. Compatible at C2 — pre-CTM lineage anchor for C2's "not unprecedented" rule. Visible example: a cell as a self-maintaining boundary that records and re-creates the conditions of its own existence. Source: `SRC-052`. --- End of master explainer. See `04_FALSE_FRIENDS_AND_LOW_RELEVANCE_PILE.md` for Tier C (the pile). See `03_TESTABLE_PREDICTION_MAP.md` for predictions grouped by family and accessibility. See `05_FORUM_READY_SUMMARY.md` for the compact public passage.